TeraWulf WULF NASDAQ

TeraWulf's record describes a company that financed a capital-intensive buildout first through structured, insider-heavy private capital and then through progressively more dilutive instruments as conditions tightened. The 2022 sequence shows a stack being assembled quickly: a $500M universal shelf and $200M ATM (claims #14103, #14135), a $124M term loan (claim #14100), a Yorkville standby facility of $50M with a $15M convertible advance (claims #13900, #13901), and an additional $50M term loan facility accompanied by lender warrants (claims #13873, #13874). The Yorkville conversion price was amended within months (claim #13463), and by October 2022 the company was raising single-digit millions through direct unit placements (claims #13554, #13534). That shift in ticket size reads less as strategy than as patching around a balance sheet under pressure. The working capital deficiency of $112M reported in March 2023 (claim #12850) supports that reading.

The February 2023 offering at $0.68 per share, roughly $27M after the over-allotment (claims #13404, #13309), was tied directly to lender terms. The Fifth Amendment conditioned amortization relief on $33.5M of equity proceeds (claim #13210), and the lenders received penny and dollar warrants on roughly 40M shares combined (claims #13211, #13212). From that point equity raising served debt management rather than growth, and the warrant overhang became a standing feature of the capital structure. Insiders participated repeatedly on the buy side, including the CEO's share of a $5M private investment (claim #13976) and the $12M Prager-entity exchange (claim #13429). The record shows no insider sales.

Use of proceeds is harder to trace. The stated purpose of the Series A offering (claim #14277) is not itemized here. The large miner purchase agreements and the $156M Talen joint venture commitment (claims #14137, #14138) remain not yet verifiable. The later $17M Nautilus cash contribution (claim #13712) is of a different scale, and the gap between commitment and deployment requires explanation. Unit conventions also vary across claims, which limits precise reconciliation.

Going forward, a sophisticated reader should watch three things. The first is whether the debt repayments of $22M and $30M (claims #12093, #12014) satisfied the $40M condition for the maturity extension (claim #13199), which the record does not confirm. The second is how the authorized share increase to 600M (claim #12077) is used against the standing ATM. The third is the pace of warrant exercise. The record is truncated in April 2024, so later activity is not assessed here.

TeraWulf's record shows dilution that accelerated through 2022 and early 2023 and then tapered. The pre-2022 base was built on private placements and a $124M term loan (claims #14098, #14099, #14101, #14100). By mid-2022 the company was running several channels at once: a $200M ATM (#13972), a $20M public offering alongside $5M from existing holders including the CEO (#14066, #13976), a $50M Yorkville standby facility with a $15M convertible advance (#13900, #13901), and a $50M term loan add-on carrying three tranches of lender warrants (#13873, #13874–#13876). The October 2022 raises were much smaller: a $9.5M private placement and unit placements of $1.75M and $2.0M (#13554, #13534, #13535). In the same month the Yorkville conversion price was amended (#13463). Together these suggest capital was being assembled in increments rather than on a planned schedule.

The February 2023 offering at $0.68 per share (#13404) marks the low point of financing terms. The term-loan amendment linked amortization relief to a $33.5M equity raise (#13210). It added roughly 26.7M penny warrants and 13.3M dollar warrants for lenders (#13211, #13212), and it disclosed a working-capital deficiency of $112M (#12850). Debt was being restructured through equity, which makes the mix look more like patching than a coherent capital structure. One mitigating feature is that insider-linked entities were consistent buyers and exchangers rather than sellers (#13976, #13401, #13429). No insider sales appear on the record.

It is not possible to test whether announced use of proceeds matched deployment. The stated Series A use (#14277) has no recorded detail. The largest committed outflows, including the Bitmain miner agreements and the Talen joint-venture contribution (#14137, #14138, #14136), remain not yet verifiable, as does the Nautilus cash contribution (#13712). Some figures are recorded at inconsistent scales. For example, the Series A appears as both $10,000 and $10,000,000 (#14243, #14276), which needs to be reconciled before any reconciliation of sources and uses.

The 2024 entries point to deleveraging, with repayments of $22M and $30M (#12093, #12014). The same period saw a charter amendment raising authorized shares to 600M (#12077). A careful reader should watch for several things: how much of the renewed $200M ATM (#12129) has been drawn; whether the $40M repayment condition for the maturity extension (#13199) was met; how and when the lender warrants are exercised; and what the truncated April 2024 authorization item (#12013) ultimately covers. The record ends there, so any later activity is not yet reflected.

The record shown here ends partway through April 2024, so this read covers roughly 2022 through early 2024 and should not be taken as a view of the company's current capital structure. Within that window, dilution came quickly and in many small pieces rather than through a few large raises. In March and April 2022 alone the company did a Series A convertible preferred offering (claim #14276), set up a $500M universal shelf and a $200M at-the-market program (claims #14103, #14102), and priced a $20M public offering alongside a $5M purchase by existing investors that included the CEO (claims #14066, #13976). By mid-2022 it had added a Yorkville standby equity facility and a $15M convertible advance (claims #13900, #13901). Each round of lender negotiations came with warrants. The February 2023 offering was priced at $0.68 per share (claim #13404), and the record shows penny and dollar warrants to lenders of roughly 27.8M and 13.9M respectively (claims #12150, #12151). Taken together, the pattern points to dilution that sped up as the share price fell.

The financing mix looks more like patching driven by the balance sheet than a planned sequence. The early structure leaned on a $124M term loan (claim #14134) to fund miner purchase commitments (claims #14136, #14137) and the Nautilus joint venture (claim #13712). After that, most financings answer to lender conditions. The 2023 equity raise was explicitly tied to amortization relief under the Fifth Amendment, which required roughly $33.5M in equity proceeds (claims #13210, #13198). The Yorkville note's conversion price and maturity were amended within months of issuance (claims #13463, #13464). Related-party participation recurs throughout: Allin WULF, Lake Harriet Holdings and Stammtisch Investments appear in unit, warrant and note transactions (claims #13534, #13395, #13400, #13176). That repeated insider support is a constant of the record and deserves disclosure-level attention.

The announced use of proceeds can only be partly traced. The stated purpose of the Series A offering is recorded (claim #14277), but its contents are not shown here. Deployment toward miner deposits (claim #13878) and toward debt reduction is visible, including repayments of $22M and $30M (claims #12093, #12014). These appear to address the condition that the company repay $40M by April 2024 to extend the loan maturity (claim #13199). Whether that condition was formally met is not established in this record.

A reader should next look for confirmation of the maturity extension, for how the increase in authorized shares is used (claim #12077), for the annual automatic increases in the incentive plan share pool (claim #12667), and for remaining warrant overhang. Unit labels in the source data are also inconsistent, mixing raw dollar amounts and millions, so individual figures require verification against the underlying filings.

TeraWulf's record from 2021 through early 2024 shows financing that was frequent, layered, and mostly driven by lender requirements rather than by a set issuance plan. Dilution came in several overlapping channels. A $200 million at-the-market program was first disclosed in March 2022 (claim #14135) and restated through November 2023 (claim #12129). Alongside it ran a $50 million Yorkville standby equity agreement with a $15 million convertible advance (claims #13900, #13901), small unit placements with Allin WULF and Bayshore (claims #13534, #13535), and a February 2023 underwritten offering priced at $0.68 per share (claim #13404). The March 2024 charter amendment raising authorized shares to 600,000,000 (claim #12077) indicates that management expected further issuance capacity to be needed.

The mix looks more like patching than design. The December 2021 term loan of $124 million (claim #14100) became the main constraint on later decisions. Lender warrants were issued repeatedly (claim #13848), and the Fifth Amendment then added roughly 26.7 million penny warrants and 13.3 million dollar warrants (claims #13211, #13212). That amortization relief depended on raising $33.5 million of equity (claims #13210, #13198). The equity therefore served largely to secure debt concessions. The Yorkville note followed the same pattern: its conversion price and maturity were amended within months of issuance (claims #13463, #13464). Insiders consistently bought rather than sold, with the CEO and related entities participating in placements and exchanges (claims #13976, #13429). No insider sales appear on record.

Tracking proceeds against deployment is difficult on this record. The stated use of the Series A proceeds (claim #14277) has no matching deployment claim. The Bitmain purchase commitments of $169 and $33 (claims #14137, #14136) and the Nautilus contribution (claim #13712) remain not yet verifiable. Several figures are recorded without consistent units, which limits precision. The working-capital deficiency of $-112 (claim #12850) suggests that proceeds went largely to sustaining operations and servicing debt.

The record breaks off in April 2024, so later activity is outside this read. Readers should check whether the 2024 repayments of $22 and $30 (claims #12093, #12014) satisfied the $40 million condition for extending maturity to December 2024 (claim #13199), because the timing of claim #12014 requires explanation. They should also watch how much of the expanded authorized share count gets drawn, and whether future financing shifts from lender-imposed equity toward a self-directed capital plan.

TeraWulf's record describes a company that financed a capital-intensive buildout with whatever instrument was available at each moment. The early capital stack was conventional for a pre-revenue miner: private placements in March and December 2021 (the $30M and $76M placements, per claims #14098 and #14101), a $50M Series A preferred (claim #14099), and a $124M term loan (claim #14100). From mid-2022 onward the instruments multiplied and their terms worsened. Within roughly ninety days the company layered a $200M ATM (claim #14102), an April 2022 public offering of $20M (claim #14066), a Yorkville standby equity facility of $50M with a $15M convertible advance (claims #13900, #13901), and an additional $50M term loan that came with lender warrants (claim #13873). The Yorkville conversion price was then amended in October 2022 (claim #13463), and the February 2023 offering priced at $0.68 per share (claim #13404). That sequence reads as reactive patching rather than a planned mix. The lender relationship drove much of the dilution: the Fifth Amendment traded amortization relief for a $33.5M equity raise (claim #13210) and penny and dollar warrants, whose counts were later restated upward (claims #13211 and #12150), a revision that requires explanation. Insiders consistently bought rather than sold. The CEO took part in the April 2022 raise (claim #13976), and a Prager entity exchanged $12M of shares for warrants (claim #13429). No insider sales appear in this record. Use of proceeds cannot be reconciled from this data. The Series A use statement (claim #14277) is recorded without detail, and the major deployment claims are all not yet verifiable: the Bitmain purchase agreements of $33M and $169M (claims #14136, #14137), the $156M Talen contribution (claim #14138), and the Nautilus cash contribution (claim #13712). The record does show deleveraging in 2024. Repayments of $22M and $30M (claims #12093, #12014) together exceed the $40M threshold tied to the maturity extension (claim #13199), but the record does not confirm that the condition was formally met. Looking forward, the increase in authorized shares to 600 million (claim #12077) and the incentive-plan pool, which grows automatically through fiscal 2026 (claim #12667), keep dilution capacity in place even as the debt burden shrinks. The record ends abruptly in April 2024, with the final entry cut off mid-line, so anything after that date is outside what this analysis can support.

TeraWulf's record through April 2024 shows dilution driven first by financing needs and later by lender terms, not by a planned schedule. The record supplied ends partway through an April 2024 authorization entry (claim 12013), so this read stops there. In the first half of 2022 the company used every available channel at once: a Series A convertible preferred (claim 14276), a $200 million ATM (claim 13972), an underwritten common offering with an insider side-car that included the CEO (claims 14066, 13976), and a $50 million Yorkville standby facility with a $15 million convertible advance (claims 13900, 13901). By the second half of 2022 the deals had become smaller and more bespoke. There were $1.75 to $2 million direct unit placements (claims 13534, 13535), a $7.5 million term-loan draw (claim 13555), shares issued to Somerset for a lease amendment (claim 13700), and a Yorkville note amended for both conversion price and maturity (claims 13463, 13464). This sequence looks less like a coherent capital structure and more like successive patching, and the ATM and the standby facility overlapped. In 2023 the term lenders effectively set the issuance calendar. The Fifth Amendment tied amortization relief to $33.5 million of equity by March 15, 2023 (claims 13197, 13210) and extended maturity on a $40 million repayment (claim 13199). The lenders also received penny and dollar warrants on 26.7 million and 13.3 million shares (claims 13211, 13212), and these counts were later adjusted upward (claims 12150, 12151). The $0.68 per share February 2023 offering (claim 13404) came after a disclosed working-capital deficit of $112 million (claim 12850). Whether proceeds were used as announced cannot be confirmed from this record. The stated intent behind the Series A (claim 14277) carries no amount, and the large deployment commitments, including the Bitmain purchases and the Talen joint-venture contribution (claims 14137, 14138), remain not yet verifiable. The verified debt repayment of $30 million (claim 12014), following $22 million (claim 12093), is the clearest evidence of deployment, and it points to deleveraging rather than growth spending. Going forward, readers should watch four things: the outstanding lender warrant overhang, how much of the ATM capacity re-registered in 2023 (claim 12272) has actually been drawn, the effect of the March 2024 increase in authorized shares to 600 million (claim 12077), and whether the earlier joint-venture and miner commitments are eventually reconciled to actual spending.

TeraWulf's disclosed record through April 2024 describes a company that assembled a broad financing toolkit at once and then spent much of 2022 and early 2023 renegotiating it. The early disclosures show several instruments in parallel: the $124 million term loan (claim 14134), a $200 million ATM (claim 14102) under a $500 million shelf (claim 14103), and the Series A convertible preferred (claim 14276). Within months the cadence turns incremental. It adds the $50 million Yorkville standby facility and $15 million convertible advance (claims 13900, 13901), small October 2022 unit placements of $1.75 to $2 million each (claims 13533–13535), and a Yorkville note whose conversion price was revised (claim 13463). Against the working capital deficiency of $112 million (claim 12850), this sequence reads more as successive patching than a planned capital structure.

Dilution came through channels a headline share count can understate. Warrant grants accompanied the term loan expansion (claims 13874–13876). Penny and dollar warrants to lenders were stated at roughly 26.7 million and 13.3 million (claims 13211, 13212) and later at 27.8 million and 13.9 million (claims 12150, 12151), so debt relief was paid for partly in equity. The February 2023 offering at $0.68 per share (claim 13404) is the lowest stated price in the record. The Fifth Amendment tied amortization relief to a $33.5 million equity raise (claim 13210) and a maturity extension to a $40 million repayment by April 2024 (claim 13199). The recorded repayments of $22 million and $30 million (claims 12093, 12014) appear to meet that threshold, though the record does not explicitly confirm the extension.

Use of proceeds cannot be tested against deployment from this data. The stated intent for the Series A (claim 14277) is on record without detail. Miner deposits of $60 million (claim 13878) and the Nautilus contribution (claim 13712, not yet verifiable) are the only deployment markers. No insider sales appear. Insider activity runs toward buying, including the $5 million investment by existing shareholders including the CEO (claim 13976) and the Allin WULF warrant purchase (claim 13401).

A sophisticated reader should watch how the authorized share increase (claim 12077) is used, whether the ATM (claim 12129) becomes the primary funding channel, and how the lender warrant overhang is exercised. Several figures are recorded in inconsistent units (claims 14243, 14276), which requires reconciliation. The record ends mid-entry in April 2024, so later events are not covered here.

The record runs from March 2022 to April 2024 and is cut off mid-entry, so this read covers that window only. Several amounts appear to be recorded in millions without a unit, and the readings below assume that.

In 2022, dilution came frequently and in small increments. A $10 million Series A convertible preferred (claim 14276) was followed within weeks by a $20 million underwritten offering (claim 14066) and a $5 million insider-led placement (claim 13976). The company then added a $50 million Yorkville standby equity facility with a $15 million convertible advance (claims 13900, 13901), and in October it took $9.5 million in private placements alongside a $7.5 million term-loan draw (claims 13554, 13555). Throughout, a $200 million ATM was kept registered (claims 14135, 13972). The Yorkville note's conversion price and maturity were amended within five months of issuance (claims 13463, 13464), which suggests the original terms did not hold.

The mix looks less like a sequenced plan than like repeated responses to lender conditions. The $124 million term loan (claim 14100) drew successive warrant grants: lender warrants of $5.79 million (claim 13848), then penny and dollar warrants of $26.7 million and $13.3 million (claims 13211, 13212). Amortization relief depended on raising $33.5 million of equity (claim 13210). The February 2023 offering, priced at $0.68 per share (claim 13404), appears to have been sized to meet that requirement. A working capital deficiency of $112 million (claim 12850) sets the context. Insider activity in this record consists of purchases and warrant exchanges, such as the $12 million Prager-entity exchange (claim 13429), rather than sales.

The stated use of proceeds for the Series A (claim 14277) is on record but not itemized here. Miner deposits of $60 million (claim 13878) and the Nautilus contribution (claim 13712) are consistent with capacity buildout. By 2023, however, proceeds were mostly going to the lender. The $22 million and $30 million repayments (claims 12093, 12014) suggest that the $40 million maturity-extension condition (claim 13213) was being worked toward, though the record does not confirm it was met.

Going forward, the main thing to watch is the charter amendment raising authorized shares to 600 million (claim 12077). It increases capacity for future issuance, and how much of it is used, and through which channels, will show whether the company has moved past financing driven by lender conditions.

The record runs from March 2022 into April 2024 and is truncated mid-claim, so this read covers that window only. Over that period the capital pattern moves from planned growth funding to a lender-directed recapitalization, and then partly back.

The 2022 stack began with several layers at once. The term loan, the $124M per claim #14134, sat underneath the $10M Series A convertible preferred (claim #14276), a $200M ATM (claims #14102, #13972) and an April common raise of $20M (claim #14066). After that, the instruments became smaller and costlier. The Yorkville facility paired a $50M standby equity line with a $15M convertible advance (claims #13900, #13901). Its conversion price and maturity were amended within five months (claims #13463, #13464). Lender warrants accompanied the added term facility (claims #13873, #13848), and 8.5M shares went to Somerset for a lease amendment (claim #13700). Taken together, this reads as patching under pressure, not a sequenced plan.

The February 2023 offering at $0.68 per share, $25M per claim #13404, marks the low point. It coincided with a disclosed working capital deficiency of $112M (claim #12850), which suggests the raise was driven by need. The Fifth Amendment then tied amortization relief to $33.5M of new equity (claim #13210) and loan extension to $40M of repayment (claim #13213). It also added penny and dollar warrants, later restated at about 27.8M and 13.9M (claims #12150, #12151). From that point, lenders were setting the dilution schedule as much as management was. The 2024 repayments of $22M and $30M (claims #12093, #12014) suggest that schedule was being honored.

Insider behavior runs consistently toward buying. The CEO joined the April 2022 placement (claim #13976), affiliated entities bought units and warrants (claims #13534, #13401), and the record contains no insider sales. On use of proceeds, the stated intent for the Series A (claim #14277) is not detailed in the record. Miner deposits of $60M (claim #13878) are disclosed, but the Talen and Nautilus commitments (claims #14138, #13712) remain not yet verifiable. Deployment therefore cannot be fully reconciled with announced intent.

The items to watch next follow from the authorized-share increase to 600M (claim #12077) and the automatic expansion of the incentive pool through fiscal 2026 (claim #12667). Both leave substantial headroom for dilution. Readers should track whether the outstanding penny warrants are exercised and whether the ATM is drawn again. They should also check whether the not-yet-verifiable joint-venture contributions are confirmed in later filings.

TeraWulf's record shows a company financing its buildout in rapid, overlapping increments rather than through a planned capital stack. Between March 2021 and February 2023 the filings record at least eight distinct equity events. These include the $30M March 2021 placement (claim 14098), the $50M June 2021 preferred (claim 14099), the $76M December 2021 placements (claim 14101), the $20M April 2022 public offering (claim 14066), the $9.5M October 2022 placement (claim 13554) and the $25M February 2023 offering at $0.68 per share (claim 13404). Dilution was steady through this period, and the per-share price fell sharply over it. The $200M ATM (claim 13972) and the $50M Yorkville standby facility (claim 13900) appear repeatedly in the filings, but the record does not show how much was actually drawn on either.

The mix looks more like reactive patching than a coherent strategy. The $124M term loan (claim 14134) was followed within eighteen months by several things. Lender warrants were issued in successive rounds, rising to 27.76M penny warrants and 13.88M dollar warrants (claims 12150, 12151). Amortization was suspended on the condition of a $33.5M equity raise (claims 13198, 13210). Maturity was extended only if $40M was repaid by April 2024 (claim 13199). The Yorkville note was amended twice on conversion price and maturity (claims 13463, 13464). These are lender-driven terms that the company accepted under pressure. The disclosed $112M working capital deficiency (claim 12850) supports that reading. Insider participation runs in one direction, toward buying. Existing holders including the CEO invested $5M (claim 13976), and Prager-affiliated entities did warrant and share exchanges (claims 13429, 13400). No insider sales appear in the record.

The stated use of proceeds for the Series A (claim 14277) has no recorded content to compare against. What the record does show going out is miner deposits of $60M (claim 13878), the Nautilus contributions (claim 13712, not yet verifiable), and debt repayments of $22M and $30M (claims 12093, 12014). The pivot from capex to deleveraging is visible in that sequence, but the record does not reconcile it to any original guidance.

The record goes no later than April 2024 and is cut off mid-entry. A reader should look next at what the 600M-share authorization increase (claim 12077) is used for. The other items to watch are how the warrant overhang is handled and whether the auto-increasing incentive pool (claim 12667) becomes a material source of dilution.

TeraWulf's record through early 2024 shows a company that raised capital in small, frequent tranches, with the terms of those raises worsening as leverage tightened. The first half of 2022 combined a shelf registration ($500M, claim #14103), an ATM program ($200M, claim #14102), a $20M underwritten offering (claim #14066), and Series A convertible preferred (claim #14276). By mid-2022 the mix shifted toward structured instruments: a $50M Yorkville standby equity facility and a $15M convertible advance (claims #13900, #13901), whose conversion price and maturity were amended within months (claims #13463, #13464). The February 2023 offering priced at $0.68 per share (claim #13404), and the lender warrants that came with the amended term loan, roughly 26.7M penny warrants and 13.3M dollar warrants (claims #13211, #13212), mark the low point. On this record, dilution was driven increasingly by creditor demands rather than by the company's own choices.

The mix reads more as sequential patching than as a planned capital stack. Several equity raises were explicitly tied to loan covenants: amortization relief depended on $33.5M of equity by March 15, 2023 (claim #13197), and the maturity extension depended on $40M of repayment by April 1, 2024 (claim #13199). The recorded repayments of $22M (claim #12093) and $30M (claim #12014, dated April 7) cover that amount in total, but whether the April 1 condition was met on time requires explanation. Insiders, including the CEO, appear on the record as buyers in private placements (claims #13976, #14048). No insider sales are disclosed.

Use-of-proceeds tracking is limited. The stated uses included $60M of miner deposits (claim #13878), a $156M Talen joint venture commitment (claim #14138), and a $17M Nautilus contribution (claim #13712). The latter two remain not yet verifiable on this record, so the link from announced use to actual deployment cannot yet be confirmed. Unit notation is also inconsistent across duplicate claims, for example the Series A recorded as both $10,000 and $10,000,000 (claims #14243, #14276), which calls for caution with the raw figures.

The disclosed record appears to end around April 2024, after a charter amendment that raised authorized shares to 600M (claim #12077). A careful reader should watch how that new authorization is used, whether the ATM continues to be a main funding source, how the outstanding lender warrants are exercised, and whether later filings reconcile the joint venture contributions with their original commitments.

TeraWulf's record describes a company that financed its build-out through many small, overlapping instruments rather than a few large raises, and then shifted its capital toward servicing debt. Dilution was frequent through 2022 and early 2023. The Series A convertible preferred of roughly $10 million (claim #14276) was followed within weeks by a $20 million public offering (claim #14066), a $200 million ATM authorization (claim #13972), a $50 million Yorkville standby facility with a $15 million convertible advance (claims #13900, #13901), and an October 2022 private placement of $9.5 million (claim #13554). The February 2023 offering priced at $0.68 per share (claim #13404) shows how far equity terms had moved by then.

The mix reads more as sequential patching than a planned capital structure. The Yorkville advance had its conversion price amended and its maturity extended (claims #13463, #13464). Amortization relief on the term loan under the Fifth Amendment depended on raising $33.5 million of equity (claim #13210) and came with 26.7 million penny warrants and 13.3 million dollar warrants to lenders (claims #13211, #13212). Those warrant counts later appear as 27,759,265 and 13,879,630 (claims #12150, #12151). The record does not explain the change. Insider activity runs entirely to purchases, including CEO participation in the April 2022 placement (claim #13976) and an exchange of shares for warrants with a Paul Prager entity (claim #13429). No insider sales appear.

The record does not allow announced uses of proceeds to be reconciled with actual deployment. The intended use of the Series A proceeds is noted (claim #14277) but not itemized here. The larger commitments, the Bitmain agreements and the $156 million Talen joint-venture contribution (claims #14137, #14138), remain unverified. The clearest deployment that can be confirmed is debt reduction: $22 million (claim #12093) and $30 million (claim #12014) in early 2024. Together these exceed the $40 million repayment that conditioned the maturity extension (claim #13213).

The data provided ends mid-April 2024, and claim #12013 is cut off, so later events are outside this read. Readers should watch whether the expanded share authorization (claim #12077), the incentive-plan pool that increases automatically through fiscal 2026 (claim #12667), and the recurring $200 million ATM authorizations are drawn on. They should also watch how the outstanding lender warrants are ultimately settled.

TeraWulf's record through early 2024 shows rapid, layered dilution that was driven by the balance sheet rather than by a growth plan. Between March and October 2022 the company drew on nearly every available instrument. It issued Series A convertible preferred ($10M, per claim #14276), priced a $20M public offering (claim #14066) alongside a $5M insider placement that included the CEO (claim #13976), and signed a $50M Yorkville standby equity facility with a $15M convertible advance (claims #13900, #13901). It also added a $50M term loan facility that came with lender warrants (claim #13873) and closed a $9.5M private placement (claim #13554). The Yorkville conversion price was later amended (claim #13463). The financings got smaller and arrived closer together over the year, which reads as opportunistic patching rather than a staged plan.

The 2023 sequence makes the pattern explicit. The $25M offering priced at $0.68 per share (claim #13404) was tied to a lender requirement of roughly $33.5M in equity proceeds in exchange for amortization relief (claims #13210, #13198). The lenders were also paid in penny and dollar warrants (claims #13211, #13212). The company reported a working capital deficiency of $112M at the time (claim #12850). The warrant counts were later shown at 27,759,265 and 13,879,630 (claims #12150, #12151), an upward adjustment from the earlier figures that requires explanation. Debt repayments of $22M and $30M followed in early 2024 (claims #12093, #12014). The record does not confirm whether these satisfied the $40M maturity-extension condition (claim #13199).

Use of proceeds is difficult to test against this data. The Series A use-of-proceeds claim (claim #14277) carries no detail. The capital commitments to the Talen joint venture and Nautilus (claims #14138, #13712) are not yet verifiable, and miner deposits are disclosed only as a balance (claim #13878). In practice, a meaningful share of 2022–2023 equity appears to have gone toward lender conditions rather than the originally described build-out, though the record does not allow a precise reconciliation.

The disclosed history is truncated after April 2024 and contains no open-market insider sales, so conclusions about the later period cannot be drawn here. A careful reader should watch four things. The first is how much of the recurring $200M ATM authorization (claims #13982, #12129) is actually drawn. The second is how the authorized-share increase to 600 million (claim #12077) is used. The third is the timing of exercises on the outstanding lender warrants. The fourth is whether subsequent filings reconcile the joint-venture commitments with actual cash deployed.

The record shows a company that financed itself in short, overlapping increments rather than through a single anchored raise. The first year alone layers a Series A convertible preferred (the $10,000,000 offering, per claim #14276), an underwritten common offering (the $20,000,000 deal, per claim #14066), a $200,000,000 ATM program (per claim #13972), a Yorkville standby equity facility (the $50,000,000 SEPA, per claim #13900), and a convertible advance (the $15,000,000 note, per claim #13901). This sequence was followed within months by lender warrants attached to an added term loan facility (per claims #13873 and #13848). The dilution trajectory then steepens. The February 2023 offering priced at $0.68 per share (per claim #13404), a level well below the period's earlier raises. It coincided with a disclosed working capital deficiency (per claim #12850) and a loan amendment that conditioned amortization relief on a $33.5 million equity raise (per claim #13210). That amendment also issued penny and dollar warrants, later stated at 27,759,265 and 13,879,630 respectively (per claims #12150 and #12151). The mix reads less as a pre-set capital plan than as successive accommodations to lender terms, though each step is individually disclosed and documented. Insider activity on record runs toward participation rather than sales, including the CEO in a private investment (per claim #13976) and Allin WULF in direct units (per claim #13534). The Yorkville note's conversion price was amended (per claim #13463), and that change requires explanation alongside the terms of the original advance. On deployment, the stated use of proceeds for the Series A (per claim #14277) is recorded without detail sufficient to test against outcomes. The large equipment and joint-venture commitments (the $169 Bitmain agreement and $156 Talen contribution, per claims #14137 and #14138) remain not yet verifiable. Debt repayments of $22 and $30 appear in 2024 (per claims #12093 and #12014). Whether they satisfy the $40 million maturity-extension condition (per claim #13199) is not stated in this data. Readers should watch how much of the authorized-share increase to 600,000,000 (per claim #12077) is drawn down, along with the exercise pattern of the lender warrant overhang and the auto-increasing incentive pool (per claim #12667). The record supplied here is truncated in April 2024, so later developments cannot be assessed.

TeraWulf's record describes financing driven by liquidity needs, not a pre-set capital plan, though the later entries show the balance sheet stabilizing. The 2021 base was mixed: a $124 million term loan (claim #14134) beside private placements and preferred issuance (claims #14098, #14099, #14101). From March 2022 the cadence tightened sharply. A $10 million Series A convertible preferred (claim #14276) was followed within weeks by a $20 million public offering (claim #14066), a $200 million ATM (claim #13972), a $50 million Yorkville standby facility with a $15 million convertible advance (claims #13900, #13901), and an additional $50 million term loan carrying lender warrants (claims #13873, #13848). The Yorkville note's conversion price and maturity were later amended (claims #13463, #13464). Revisions within months of issuance usually mean the original terms did not hold. By February 2023 common stock was priced at $0.68 per share for a $25 million raise (claim #13404). That raise was tied to a loan amendment: amortization relief depended on $33.5 million of equity (claim #13210) and came with roughly 26.7 million penny warrants and 13.3 million dollar warrants to lenders (claims #13211, #13212). Alongside it the company disclosed a working capital deficiency of about $112 million (claim #12850). In this period dilution served debt covenants more than growth.

Insider participation stands out, and it runs toward buying. Existing holders including the CEO invested $5 million (claim #13976), and affiliated entities took up direct unit placements and warrant exchanges (claims #13534, #13535, #13429). The record shows no insider sales.

The record does not let us test whether stated use of proceeds matched deployment. The intended use for the Series A is noted (claim #14277) but its content is not shown. Miner deposits of $60 million (claim #13878) and the Talen joint venture commitment of $156 million (claim #14138) are visible, but the latter is marked not yet verifiable.

The most recent entries show debt repayments of $22 million and $30 million (claims #12093, #12014). Together these exceed the $40 million needed for the maturity extension (claim #13199), but whether the extension took effect is not on the record. The authorized share count was raised to 600 million (claim #12077), and the incentive plan pool increases automatically through fiscal 2026 (claim #12667). A careful reader should watch how much of that new capacity, the ATM, and the remaining warrant overhang is used. The record shown here ends in April 2024, so anything later is beyond this read.

TeraWulf's record describes a company that raised capital in many small, overlapping pieces rather than following a planned sequence. The disclosed history ends abruptly at a common stock authorization entry in April 2024, so this read covers only 2021 through early 2024.

Dilution came in three phases. The first was 2021 to early 2022, when the base was built from private placements, preferred equity and the December 2021 term loan of roughly $124 million (claims 14098–14100). The second, from mid-2022 to early 2023, was a period of tight financing. The company issued units directly to related parties in October 2022 (claims 13533–13535). It sold shares at $0.68 in February 2023 (claim 13404). It gave lenders penny and dollar warrants covering more than 40 million shares (claims 13211–13212) in exchange for loan amendments. The third phase is marked by the authorized share count rising to 600 million (claim 12077) and the incentive plan's automatic share increases through 2026 (claim 12667). Both point to ongoing issuance capacity rather than a pause.

The mix looks reactive more than strategic. The $200 million ATM is re-disclosed almost every quarter (claims 14102, 12272, 12129). The Yorkville standby facility and convertible advance (claims 13900–13901) were amended within months (claims 13463, 13487). The lender relief in the Fifth Amendment depended on a $33.5 million equity raise by March 15, 2023 (claim 13210). That makes the February 2023 offering look driven by a covenant deadline. Insiders bought rather than sold. The CEO took part in the April 2022 private investment (claim 13976), and a Prager entity exchanged warrants for $12 million in shares (claim 13429). No insider sales appear in the record.

It is hard to check how proceeds were used against how they were described. The stated use of the Series A proceeds (claim 14277) has no detail in the record. Most capital commitments are marked not yet verifiable, including the Bitmain purchases (claims 14136–14137), the Talen JV commitment of $156 million (claim 14138) and the Nautilus cash contribution (claim 13712). Debt repayments of $22 million (claim 12093) and $30 million (claim 12014) add up to more than the $40 million needed to extend maturity (claim 13199). However, the April 7, 2024 entry falls after the April 1 deadline, and the timing needs explanation.

A careful reader should look for the missing post-April 2024 disclosures. They should check whether the extended maturity was actually triggered, how much of the ATM was used, and whether the $112 million working capital deficiency (claim 12850) was closed with equity or with new debt.

TeraWulf's record through early 2024 shows dilution that was persistent and, in its sharpest phase, priced from weakness. The 2022 sequence moves within weeks from Series A convertible preferred (claim 14276) to underwritten common stock with insider participation (claims 14066, 13976), a $200 million ATM (claim 13972), and a $50 million Yorkville standby facility with a $15 million convertible advance (claims 13900, 13901). By October 2022 the Yorkville advance's conversion price had been amended and its maturity made extendable (claims 13463, 13464), and small unit placements and private placements were filling gaps (claims 13554, 13534). The February 2023 offering at $0.68 per share (claim 13404) is the low point. It was explicitly tied to the lenders' $33.5 million equity condition for amortization relief (claim 13210), and it came alongside a reported working capital deficiency of $112 (claim 12850).

The mix reads less as a planned capital structure than as successive accommodations negotiated around the term loan (claim 14134). Each loan amendment carried an equity cost. Lenders received warrants in 2022 (claim 13827), then penny and dollar warrants of 26,666,669 and 13,333,333 (claims 13211, 13212). Those counts later appear as 27,759,265 and 13,879,630 (claims 12150, 12151), and the revision requires explanation in the filings. Insiders and affiliated entities consistently bought or exchanged rather than sold (claims 13429, 13401). Shares also went to related parties for a lease amendment and for services (claims 13700, 12155). The record contains no insider sales.

Use of proceeds cannot be meaningfully tested. The stated intent for the Series A (claim 14277) has no paired deployment figure. The joint venture and miner commitments are mostly not yet verifiable (claims 14138, 14137, 13712). What can be observed is that debt reduction followed: repayments of $22 and $30 (claims 12093, 12014) appear against the condition that $40 million be repaid by April 2024 to extend maturity (claim 13199). The record does not show whether that condition was formally satisfied.

The disclosed history is truncated in April 2024, so later events are outside this read. A careful reader should next look for confirmation that the maturity extension took effect, the exercise status of the lender warrants, and how much of the authorized-share increase to 600 million (claim 12077) is used. They should also watch whether issuance shifts from conditional, lender-driven raises toward funding growth on the company's own terms.

TeraWulf's record through late 2023 shows a company financing its buildout in small, frequent increments rather than through a planned capital stack. In the first half of 2022 the mix looked deliberate. It included a $200 million ATM (claim 13972), a $20 million underwritten offering paired with a $5 million private investment by existing holders including the CEO (claims 14066, 13976), and a $50 million Yorkville standby facility with a $15 million convertible advance (claims 13900, 13901). From mid-2022 the pattern shifts toward patching. Lender warrants accompany successive term loan accommodations (claims 13874, 13848). The Yorkville conversion price was amended within five months (claim 13463). October 2022 brought a $9.5 million private placement and unit placements of $1.75 million to $2 million with related entities (claims 13554, 13534, 13535).

Dilution accelerated into early 2023, when common stock priced at $0.68 per share raised $25 million (claim 13404). Penny and dollar warrants were issued to lenders alongside it, and a stock-for-warrant exchange was done with a Prager entity (claims 13211, 13212, 13429). That raise appears tied to a lender condition. The Fifth Amendment linked amortization relief to $33.5 million of equity by March 15, 2023 (claim 13210), and it linked maturity extension to repaying $40 million by April 2024 (claim 13199). Equity issued at that point served covenant management more than expansion, against a stated working capital deficiency of $112 million (claim 12850).

Use of proceeds is difficult to reconcile from this record. The Series A purpose is stated (claim 14277), but its specifics are not captured. The Bitmain purchase commitments and the Talen joint venture contribution remain not yet verifiable (claims 14137, 14138), as do most of the 2023 debt amendments. The $200 million ATM is reaffirmed in successive filings through November 2023 (claims 12931, 12254, 12129), but no utilization figures appear in the claims.

A sophisticated reader should watch three things: whether the $40 million repayment condition was met, how much of the ATM has actually been drawn, and how the lender warrant overhang resolves. The penny and dollar warrant issuances are recorded at roughly 27.8 million and 13.9 million (claims 12150, 12151). The record supplied ends in January 2024 and contains no insider transaction data, so this read does not extend to later financings or insider selling.

TeraWulf's record through early 2024 describes a company that financed a capital-intensive buildout in small, frequent increments rather than through a few planned raises. After the term loan (the $124 figure, per claim #14100) and the $200 ATM shelf (per claim #14143), the company returned to the market roughly every few months. It raised a $20M public offering in April 2022 (per claim #14066), signed a $50M Yorkville standby facility with a $15M convertible advance in June (per claims #13900 and #13901), closed a $9.5M private placement in October (per claim #13554), and sold a $25M offering at $0.68 per share in February 2023 (per claim #13404). Over that span, dilution moved from priced equity toward warrants. Lender warrant grants appear in June 2022 (per claims #13873 through #13876). Penny and dollar warrants follow in 2023, recorded at 26.7M and 13.3M units (per claims #13211 and #13212) and later restated at roughly 27.8M and 13.9M (per claims #12150 and #12151). These carry economic cost while barely registering as proceeds. The mix reads less as a single coherent strategy than as a sequence of accommodations with the term lenders. The Fifth Amendment exchanged amortization relief for a $33.5M equity raise (per claim #13210) and a $40M paydown condition for maturity extension (per claim #13213). Amortization suspension is described again in August and November 2023 (per claims #12252 and #12122), which suggests the debt terms were renegotiated more than once. Insider participation is consistent and directional: the CEO and existing holders bought $5M in April 2022 (per claim #13976), and affiliated entities bought warrants and units through 2022 and 2023 (per claims #13401 and #13535). The record shows no insider sales. Tracking use of proceeds is difficult on this record. The Series A use-of-proceeds statement (per claim #14277) carries no quantified allocation. The miner purchase agreements and the $156 Talen joint venture commitment are tagged as delayed (per claims #14137 and #14138), and the March 2023 working capital deficiency of $112 (per claim #12850) indicates that deployment ran ahead of funding. Some figures in the record are stated in inconsistent units, which limits precise reconciliation. A sophisticated reader should watch three things: whether the $40M paydown condition was met, how much of the $200M ATM was actually drawn after its re-filing (per claim #12129), and how quickly the accumulated warrant overhang converts into outstanding shares.

TeraWulf's record from March 2022 through January 2024 shows dilution that steepens as the cost of capital rises. The early layers were conventional: the December 2021 term loan of $124M (claim #14100), a $200M ATM (claim #14102) and a $500M universal shelf (claim #14103). By mid-2022 the mix had shifted toward instruments that carry heavier per-dollar dilution. These include the Yorkville standby equity facility of $50M (claim #13900), a $15M convertible advance (claim #13901) whose conversion price was later amended (claim #13463), and successive warrant grants to lenders (claims #13874, #13848). The February 2023 common offering priced at $0.68 per share for $25M (claim #13404) marks the low point of that path. The lender penny and dollar warrant figures, recorded at 26.7M and 13.3M (claims #13211, #13212), were later restated upward to 27.8M and 13.9M (claims #12159, #12151). The mix reads less as a planned capital structure than as sequential accommodation to the term lenders. The $33.5M equity raise (claim #13210) was a condition of amortization relief rather than a growth decision, and principal amortization was suspended again later in 2023 (claims #12252, #12122). Insider activity in this record runs toward buying, not selling. The CEO participated in the April 2022 private investment (claim #13976), and an entity tied to Paul Prager exchanged $12M of shares for warrants (claim #13429). That supports alignment, but it also means insiders absorbed part of the dilution burden. The record supports only a partial test of use of proceeds. The Bitmain purchase agreements (claims #14136, #14137), the Talen joint venture commitment of $156M (claim #14138) and the Nautilus cash contribution (claim #13712) are all marked delayed. The working capital deficiency of $112M (claim #12850) suggests that proceeds were going to liquidity and lender terms rather than the expansion first described. The $200M ATM is restated repeatedly through November 2023 (claim #12129) with no disclosed utilization in this dataset, and that requires explanation. Several items are worth watching. The first is whether the $40M repayment condition for the December 2024 maturity extension (claim #13199) was met. The second is how much of the ATM was drawn and at what prices. The third is the cumulative overhang from penny warrants, which the data does not fully reconcile. The fourth is the omnibus plan's automatic share pool increases through fiscal 2026 (claim #12667). The record ends in January 2024, so later refinancing is outside what the data supports.

TeraWulf's record from early 2022 through January 2024 describes a company financing a capital-intensive buildout in small, frequent increments rather than through a planned sequence. Raises arrive roughly monthly across 2022: a Series A convertible preferred (claim #14276), an April public offering with a concurrent insider placement (claims #14066, #14048), a Yorkville standby equity facility and convertible advance (claims #13900, #13901), and October unit placements in the low single-digit millions (claims #13534, #13535). The trajectory steepens in early 2023, when the underwritten offering priced at $0.68 per share (claim #13404). That price, set against a lender warrant package that grew from 26,666,669 penny warrants (claim #13211) to 27,759,265 (claim #12159), indicates that each successive raise bought less capital per unit of dilution.

The mix reads more as patching than as a coherent strategy. The $200M at-the-market program is re-disclosed repeatedly (claims #14102, #12870, #12272, #12129) without a corresponding record of how much was drawn. Meanwhile, the Yorkville conversion price was amended (claim #13463) and the term loan (claim #14100) was restructured several times through warrant issuance and amortization relief. The Fifth Amendment tied that relief to a $33.5M equity raise (claim #13210), so equity issuance was in part a condition set by lenders rather than a funding choice. Insiders appear on the buy side, in the CEO-inclusive private investment (claim #13976) and the Allin WULF and Lake Harriet warrant purchases (claims #13401, #13395). The record contains no insider sales.

Use of proceeds is hard to reconcile against deployment. The stated intended use of the Series A (claim #14277) carries no detail in the record. The Talen joint venture commitment (claim #14138), the Bitmain purchase agreements (claims #14137, #14136), and the Nautilus contribution (claim #13712) are all tagged delayed. The $112M working capital deficiency (claim #12850) suggests that a meaningful share of proceeds went to servicing obligations rather than expansion.

A careful reader should watch three items. The first is whether the April 2024 conditions on the term loan, the $40M repayment tied to maturity extension (claim #13213), were met. The second is how much of the $200M ATM has actually been sold. The third is the cumulative overhang from lender warrants and the auto-increasing incentive plan pool (claim #12667). The record ends in January 2024, so later outcomes on these points are not yet visible in the data.

TeraWulf's record through January 2024 shows a capital structure built on secured debt and then supported by repeated, increasingly small equity raises. The base was set in 2021 with the $124 million term loan (claim #14134), $76 million of private placements (claim #14101) and $50 million of Series A preferred (claim #14099). After that, the raises came every few weeks: a $10 million convertible preferred (claim #14276), $15 million of direct offerings (claim #14104), a $20 million public offering (claim #14066), and a $200 million ATM (claim #13972), all inside roughly six weeks of 2022. Pricing fell sharply over the period. The February 2023 offering cleared at $0.68 per share for $25 million (claim #13404).

The mix reads more as a response to lender terms than as a planned sequence. The Yorkville standby facility and its $15 million convertible advance (claims #13900, #13901) had their conversion price amended within months (claim #13463). Later financings were shaped by the term loan. The Fifth Amendment tied amortization relief to a $33.5 million equity raise (claim #13210) and came with penny and dollar warrants to lenders (claims #13211, #13212). Those warrant counts were later revised upward to about 27.8 million and 13.9 million (claims #12159, #12151). Insider activity runs the other way from selling. Existing holders including the CEO joined a $5 million placement (claim #13976), and affiliated entities bought units and convertible notes (claims #13535, #13176). No insider sales appear in the record.

The data is too thin to test use of proceeds properly. The intended use for the Series A is noted (claim #14277), but its specifics are not captured. Several deployment commitments are marked delayed, including the Bitmain purchase agreements (claims #14136, #14137), the $156 million Talen joint-venture commitment (claim #14138) and the Nautilus cash contribution (claim #13712). The disclosed working capital deficiency of $112 million (claim #12850) helps explain why the raises were tied to obligations rather than to growth.

A careful reader should watch a few items. The first is whether the $40 million repayment that conditions the maturity extension (claim #12908) was met. The second is whether the suspended amortization (claim #12122) holds. The third is how much of the twice-renewed $200 million ATM (claim #12272) is drawn. The fourth is the combined overhang from lender warrants, the April 2024 warrants (claim #12133), and the auto-increasing incentive plan pool (claim #12667).

TeraWulf's disclosed record runs from March 2022 to January 2024 and shows financing that accelerated under pressure rather than following a staged plan. The early 2022 stack was layered on top of the $124M term loan from December 2021 (claim #14134). It included a Series A convertible preferred (claim #14276), a $200M ATM with a $500M universal shelf (claims #14102, #14103), and an April public offering of roughly $20M alongside a $5M insider-linked placement (claims #14066, #13976). By mid-2022 the instruments became more expensive and more structurally dilutive. A $50M Yorkville standby equity facility and $15M convertible advance (claims #13900, #13901) followed, along with tiered lender warrants (claims #13874–13876). The October 2022 term-loan tranches arrived with a $9.5M private placement attached (claims #13555, #13554).

The mix reads less as a coherent capital plan than as successive patches to the term loan. The clearest example is the Fifth Amendment. It conditioned amortization relief on $33.5M of new equity by mid-March 2023 (claim #13210) and came paired with penny and dollar warrants to lenders, recorded at 26,666,669 and 13,333,333 (claims #13211, #13212). The February 2023 offering at $0.68 per share (claim #13404) appears to be the equity that satisfied that condition. The record shows no insider selling. Insiders and affiliated entities appear repeatedly as buyers or warrant counterparties, including Allin WULF, Lake Harriet and a Prager entity (claims #13401, #13395, #13429). That signals alignment, but it also concentrates economic exposure among related parties.

On use of proceeds, the Series A intent is stated (claim #14277), but the data does not specify it in enough detail to test against deployment. Several commitments that proceeds were presumably meant to fund carry a delayed status, including the Talen joint-venture contribution and both Bitmain purchase agreements (claims #14138, #14137, #14136). That gap requires explanation. The reported working-capital deficiency of $112M (claim #12850) suggests proceeds went largely to keeping the debt structure current rather than to growth.

A careful reader should watch several items. The first is whether the $40M repayment that triggers the maturity extension (claim #13213) is met through ATM issuance or new debt. The second is how much of the ATM authorization, re-recorded through November 2023 (claim #12129), has been drawn. The third is whether the lender warrant overhang, reissued in September 2023 and January 2024 (claims #12159, #12105), is exercised into the float.

TeraWulf's record from March 2022 through January 2024 shows dilution that accelerated sharply before it stabilized. Most of it was driven by balance-sheet pressure, not a pre-planned capital stack. The early instruments looked like a conventional growth-stage toolkit: a $500M universal shelf (claim #14103), a $200M ATM (claim #14102), and the $124M December 2021 term loan (claim #14134). Within months the mix broadened into a $50M Yorkville standby equity facility paired with a $15M convertible advance (claims #13900, #13901). That convertible was later amended on conversion price and maturity (claims #13463, #13464). The company also ran a series of small direct unit placements in the low single-digit millions (claims #13534, #13535). That sequence reads as incremental patching more than a coherent strategy.

The term loan is the center of gravity. Lender compensation escalated from 5,787,732 warrants (claim #13848) to 26,666,669 penny warrants and 13,333,333 dollar warrants under the Fifth Amendment (claims #13211, #13212). Those counts were restated at 27,759,265 and 13,879,630 (claims #12159, #12160). Amortization relief was conditional on raising $33.5M of equity (claim #13210). Extending maturity to December 2024 required a $40M paydown by April 1, 2024 (claim #13213). The February 2023 offering priced at $0.68 per share for $25M (claim #13404) appears to have been the vehicle for meeting that condition. A stated working capital deficiency of $112 (claim #12850) frames the urgency. Insider participation runs entirely in the buying direction: the CEO-linked private investment (claim #13976) and the Prager-entity exchange (claim #13429). The record contains no insider sales.

Use of proceeds is harder to test. The Series A preferred carries a stated intended use (claim #14277), but the record does not show its content. Several capital commitments are tagged delayed: the $169 Bitmain purchase (claim #14137), the $156 Talen joint venture contribution (claim #14138), and the Nautilus cash contribution (claim #13712). That suggests deployment was resequenced against funding availability. Without later reconciliation, how closely spending tracked the original plans requires explanation.

The data thins after late 2023 and ends with a January 2024 warrant restatement (claim #12105). Readers should look for whether the April 2024 $40M paydown condition was met. They should also track how heavily the renewed $200M ATM (claim #12129) was drawn, and whether the Omnibus Plan's automatic share-pool increases through fiscal 2026 (claim #12667) add a steady base layer of dilution beneath the financing activity.

TeraWulf's record describes a company that set up a broad financing toolkit early and then relied on increasingly expensive, lender-driven equity as its capital needs outran it. In March 2022 the structure was a full suite: a Series A convertible preferred of $10 million (claim #14276), a $200 million ATM (claim #14102), a $500 million universal shelf (claim #14103), and a December 2021 term loan of $124 million (claim #14134). Within roughly three months the company added a $50 million Yorkville standby equity facility (claim #13900) with a $15 million convertible advance (claim #13901), whose conversion price was later amended (claim #13463). Small placements followed: $9.5 million in October 2022 (claim #13554), plus unit sales of $1.75 million and $2 million (claims #13534, #13535). Taken together, this looks less like a planned sequence than a series of patches, each sized to the shortfall in front of it. Pricing moved in the same direction. The February 2023 offering was priced at $0.68 per share for $25 million (claim #13404). Lender compensation shifted from percentage-of-equity warrant grants (claims #13874–#13876) to penny warrants of about 26.7 million shares and dollar warrants of about 13.3 million (claims #13211, #13212), later restated at about 27.8 million and 13.9 million (claims #12159, #12160). The term loan amendments tie amortization relief to equity raises, including $33.5 million by March 2023 (claim #13210), so lenders were effectively directing the equity issuance. A working capital deficiency of $112 million (claim #12850) frames that pressure. On use of proceeds, the record states an intended use for the Series A (claim #14277) but contains little that confirms deployment. The commitments that would show it are marked delayed: the Talen joint venture contribution of $156 million (claim #14138), the Bitmain miner purchases (claims #14136, #14137), and the Nautilus cash contribution (claim #13712). A reader should watch whether the $40 million repayment needed to extend maturity to December 2024 (claim #13199) is met, how the April 2024 warrants are exercised (claim #12133), how much of the renewed $200 million ATM is drawn (claim #12255), and the effect of the auto-increasing incentive pool (claim #12667). The record shows insider and affiliate buying, including CEO participation (claim #13976), but no insider sales. Some claims report amounts in inconsistent units, so a few figures need reconciliation against the underlying filings.

TeraWulf's disclosed record from March 2022 through January 2024 reads as a sequence of financings driven more by near-term liquidity than by a pre-set capital plan. The window opens on an already layered base: a $30M private placement, $50M of Series A preferred, a $124M term loan and $76M of private placements (claims #14098, #14099, #14100, #14101). Alongside these sat miner and joint-venture commitments of $169M, $33M and $156M, each marked delayed (claims #14137, #14136, #14138). What follows is a dense run of small, overlapping raises rather than a few sized ones.

Dilution came faster, and the price per share fell. The April 2022 offering was about $20M plus a $5M placement from existing holders including the CEO (claims #14066, #13976). By February 2023 the company sold $25M of stock at $0.68 per share (claim #13404). In between came the Yorkville standby facility and its $15M convertible advance (claims #13900, #13901), whose conversion price was later amended (claim #13463), and sub-$10M unit placements with related parties (claims #13554, #13535, #13534). Taken together, the mix reads as opportunistic patching rather than a coherent structure, with each instrument closing a specific gap.

The term loan is the organizing pressure. Lender warrants first appear in mid-2022 (claim #13848). They then expand to penny and dollar warrants recorded as 26.7M and 13.3M (claims #13211, #13212), granted in exchange for amortization relief. That relief was conditional on a $33.5M equity raise (claim #13210), and the maturity extension on a $40M paydown (claim #13213). The warrant figures are later restated as 27.8M and 13.9M (claims #12159, #12160), a revision that requires explanation. A $112M working-capital deficiency is stated (claim #12850).

The record states an intended use for the Series A proceeds (claim #14277) but carries no figure-level deployment data, so use of proceeds cannot be tracked from this record. The $17M Nautilus contribution was itself delayed (claim #13712).

A careful reader should watch four things. The first is whether the $200M ATM, reauthorized repeatedly (claims #13972, #12254, #12129), is actually drawn. The second is whether the April 2024 $40M repayment condition (claim #13199) is met. The third is how penny-warrant exercises (claim #12105) feed into the share count. The fourth is that no insider sales appear in this record, so insider behavior cannot be assessed.

TeraWulf's record describes a company that raised capital on a large plan and then spent roughly two years refinancing it in smaller pieces. The 2021 capital stack was conventional for a pre-revenue miner: a $30M private placement (claim #14098), $50M of Series A preferred (claim #14099), a $124M term loan (claim #14100), and $76M of December private placements (claim #14101), with a $200M ATM (claim #14102) and a $500M shelf (claim #14103) in place behind them. After early 2022 the raises became smaller and more frequent. There were $15M of direct offerings (claim #14104), a $10M convertible preferred (claim #14276), and a $20M public offering paired with a $5M insider purchase (claims #14066, #14048). A $50M Yorkville standby facility followed with a $15M convertible advance (claims #13900, #13901), and its conversion price was later amended (claim #13463). By October the company was using single-digit-million unit placements and term-loan draws (claims #13554, #13555).

The mix reads less as a sequenced strategy than as successive responses to liquidity pressure. A working capital deficiency of $112M (claim #12850) preceded a February 2023 offering priced at $0.68 per share (claim #13404). Lender concessions were paid for in equity-linked instruments. The Fifth Amendment's amortization relief depended on raising $33.5M of equity (claim #13210). It also issued penny and dollar warrants, recorded at 26,666,669 and 13,333,333 (claims #13211, #13212) and later restated at 27,759,265 and 13,879,630 (claims #12159, #12151). Issuing 8.5M shares to Somerset for a lease amendment (claim #13700) and shares to Beowulf for services (claim #12155) also suggests equity was being used in place of cash. Insider activity on record consists entirely of purchases and exchanges by affiliated entities, including the CEO's participation (claim #13976), the Prager-entity warrant exchange (claim #13429), and the Allin WULF and Bayshore placements (claims #13534, #13535). No insider sales appear.

The record does not let a reader reconcile stated use of proceeds with actual deployment. The intended use for the Series A raise is noted (claim #14277) without detail. The core commitments it would have funded are all marked delayed: the $169M and $33M Bitmain agreements (claims #14137, #14136), the $156M Talen joint venture contribution (claim #14138), and the $17M Nautilus cash contribution (claim #13712).

The record ends in January 2024. Two items fall due after that point. The first is the $40M repayment by April 1, 2024 that conditions the maturity extension to December 2024 (claim #13199). The second is how much of the $200M ATM, reaffirmed repeatedly (claims #12255, #12129), was ultimately drawn. Lender warrant exercises are also worth tracking, since warrants became exercisable in late 2023 (claim #12132).

TeraWulf's record from March 2022 through January 2024 shows dilution that accelerated rather than tapered. Each financing was followed by another within weeks. The early capital stack was broadly conventional: a $124M December 2021 term loan (claim #14134), $76M of private placements (claim #14101), a $200M ATM (claim #14102), and a $500M universal shelf (claim #14103). By mid-2022 the mix had shifted to smaller and more structured instruments. These included a $50M Yorkville standby equity agreement with a $15M convertible advance (claims #13900, #13901), the Yorkville terms amended within four months (claims #13463, #13487), and direct unit placements of roughly $1.75M to $2M each in October 2022 (claims #13533 to #13535). This reads less as a staged strategy than as sequential patching of near-term liquidity, a reading reinforced by the stated working capital deficiency of $112M (claim #12850).

The term loan is the throughline. Warrants were granted to lenders repeatedly: 5,787,732 in mid-2022 (claim #13848), then penny and dollar warrants of 26,666,669 and 13,333,333 (claims #13211, #13212), later restated at 27,759,265 and 13,879,630 (claims #12150, #12151). Amortization relief was made conditional on a $33.5M equity raise (claim #13210). The maturity extension was tied to a $40M repayment by April 2024 (claim #13213). Equity issuance was therefore effectively being used to service lender covenants. The February 2023 offering at $0.68 per share (claim #13404) marks the low point in pricing.

Insider participation runs toward buying, not selling. The CEO participated in the April 2022 private investment (claim #13976), and related entities exchanged or purchased warrants (claims #13400, #13401, #13429). No insider sales appear in this record.

Use of proceeds cannot be tested well here. The intended use for the Series A is recorded (claim #14277) but not itemized. The Bitmain purchase commitments (claims #14136, #14137) and the Talen joint venture contribution (claim #14138) are marked delayed, and the Nautilus contribution (claim #13712) is likewise delayed. Deployment appears to have lagged the original plan.

A reader should watch whether the April 2024 $40M repayment condition was met (claim #13199). They should also watch the exercise of the outstanding penny warrants (claim #12105) and whether the renewed $200M ATM (claim #12129) becomes the primary funding channel.

TeraWulf's record describes a company that financed a capital-intensive buildout through 2022 and then spent most of 2023 managing the consequences, and the cadence shows that shift clearly. The early structure looks planned. The December 2021 term loan recorded at $124 (claim 14134), the $200 ATM program (claim 14135), and the $500 universal shelf (claim 14103) form a conventional base of debt anchored by shelf-backed equity. That plan was quickly supplemented. Between April and October 2022 the company layered on a $20,000,000 public offering (claim 14066), a $50,000,000 Yorkville standby facility with a $15,000,000 convertible advance (claims 13900, 13901), small direct unit placements (claims 13534, 13535), and lender warrants attached to an additional $50 term facility (claims 13873, 13874). The ticket sizes shrink and the instruments diversify, which reads more as patching than as a designed capital stack. The Yorkville conversion price was amended within months (claim 13463), which requires explanation. Dilution then accelerates through lender consideration rather than primary issuance. The penny and dollar warrants recorded at $26,666,669 and $13,333,333 (claims 13211, 13212) were conditioned on a $33,500,000 equity raise in exchange for amortization relief (claim 13210). Those figures were later restated upward to $27,759,265 and $13,879,630 (claims 13159, 12160). That revision should be reconciled against the original amendment terms. Insiders appear only as buyers in this record: the CEO's participation (claim 13976), the Prager-entity share-for-warrant exchange (claim 13429), and Allin WULF's warrant purchase (claim 13401). No insider sales are disclosed. On use of proceeds, the stated intent for the Series A (claim 14277) carries no figure that can be traced. Visible deployment is limited to miner deposits of $60 (claim 13878) and a $17 Nautilus contribution (claim 13712), set against a $156 Talen commitment marked delayed (claim 14138), so the tracking cannot be confirmed from this data. Several figures also appear to mix unit scales, and the record ends in January 2024 (claim 12105). A sophisticated reader should therefore watch for three things: whether the $40,000,000 repayment condition tied to the maturity extension was met (claim 13213), how heavily the reauthorized $200,000,000 ATM was drawn (claim 12254), and how the repeatedly delayed amortization suspension was finally resolved (claim 12122).

TeraWulf's record through early 2024 shows dilution that started as growth funding and became balance-sheet maintenance. The 2022 raises came at a steady monthly pace: a Series A convertible preferred (claim #14276), a $20M public offering with a concurrent $5M insider purchase (claims #14066, #14048), a $50M Yorkville standby equity facility with a $15M convertible advance (claims #13900, #13901), and small unit placements in October (claims #13554, #13534). By February 2023 the company was selling common stock at $0.68 per share (claim #13404) for roughly $27M including the over-allotment (claim #12907). Over the same period it reported a working-capital deficiency of $112M (claim #12850). The direction is clear: each raise was smaller relative to need and done at worse terms.

The mix reads less as a single plan than as successive patches around the $124M term loan (claim #14134). Lender compensation was paid largely in equity. Penny and dollar warrants were issued in tranches (claims #13211, #13212), and the counts were later restated upward (claims #12150, #12160). Amortization relief depended on raising $33.5M of equity (claim #13210). The Yorkville note's conversion price was amended within months of issue (claim #13463). Insider participation is consistent across the record, through Allin WULF, Lake Harriet and Stammtisch (claims #13401, #13176, #13400), and the record contains no insider sales. A $200M ATM is reaffirmed in at least four filings (claims #14135, #12931, #12255, #12129), but none of them discloses how much has actually been drawn.

Use of proceeds is hard to check against deployment. The intended use of the Series A proceeds is noted (claim #14277), but its substance is not captured here. The capital commitments that would show deployment are mostly marked delayed: the Bitmain miner agreements (claims #14136, #14137), the $156M Talen joint venture commitment (claim #14138) and the Nautilus cash contribution (claim #13712). Together these require explanation of how commitments were sequenced against cash actually raised.

The record thins after late 2023 and ends in January 2024 (claim #12105), so the following points are forward-looking only within that window. A careful reader should first watch whether the $40M repayment needed by April 2024 to extend maturity (claim #13199) was met, and from what source. They should also watch whether the April 2024 warrants (claim #12133) were exercised. Finally, two items are still open: actual ATM utilization, and the incentive-plan share pool, which increases automatically through fiscal 2026 (claim #12667).

TeraWulf's record shows a company that entered 2022 with a layered stack already in place, including the $124 December 2021 term loan (claim #14100), $76 of December 2021 private placements (claim #14101), a $500 universal shelf (claim #14103) and a $200 ATM (claim #14102). It then spent roughly two years adding smaller, more expensive instruments. The cadence of dilution accelerated rather than tapered. The April 2022 public offering of $20,000,000 (claim #14066) gave way within two months to a Yorkville standby equity facility of $50,000,000 (claim #13900) and a $15,000,000 convertible advance (claim #13901), whose conversion price was later amended (claim #13463). By February 2023 common stock was being sold at $0.68 per share (claim #13404). The mix reads less as a coherent plan than as sequential patching. Lender warrants accompanied the additional term facility at stepped dilution levels of 5.0%, 4.25% and 3.75% (claims #13874-13876). The Fifth Amendment then made amortization relief through April 2024 conditional on a $33.5 million equity raise (claim #13210) and issued penny and dollar warrants alongside it (claims #13211, #13212), with counts later restated upward (claims #12150, #12160). The $-112 working capital figure (claim #12850) is the clearest statement of why. Insiders and affiliated entities were consistent buyers rather than sellers: the CEO took part in the $5,000,000 existing-investor purchase (claim #13976), and affiliated entities participated through a $12,000,000 warrant exchange (claim #13429), the Lake Harriet notes of $853,912 (claim #13176), and the Allin WULF and Bayshore units (claims #13534, #13535). No insider sales appear in the record. On use of proceeds, the stated intent for the Series A (claim #14277) is not itemized, and the capital commitments it would plausibly fund are marked delayed. These include the Talen joint venture contribution of $156 (claim #14138), the Bitmain agreements of $169 and $33 (claims #14137, #14136) and the Nautilus cash contribution of $17 (claim #13712). That gap requires explanation. The $200 ATM is restated repeatedly through November 2023 (claim #12129) with no disclosed utilization in these claims, and several figures appear in inconsistent units, which limits precise reconciliation. A careful reader should watch whether the $40 million repayment by April 1, 2024 (claim #13199) was met and the maturity extended. They should also watch how many penny and dollar warrants are exercised, what the April 2024 warrants (claim #12133) add, and how far the omnibus plan's automatic increases through fiscal 2026 (claim #12667) compound the share count. The record after January 2024 is sparse, so later conclusions are not supported.

TeraWulf's record from March 2022 through January 2024 shows a capital-intensive buildout financed in small, frequent increments rather than through the layered plan first described. The early 2022 disclosures pair a $500 universal shelf (claim #14103), a $200 ATM (claim #14102), the $124 December 2021 term loan (claim #14134) and Series A convertible preferred (claim #14276) with Bitmain miner commitments of $169 and $33 (claims #14137, #14136) and a $156 Talen joint venture commitment (claim #14138). All three commitments are now marked delayed. Units vary across claims, so small figures should be read in context, generally as millions.

Within weeks the mix turns opportunistic. A $20,000,000 underwritten offering (claim #14066) was followed by a $50,000,000 Yorkville standby equity facility with a $15,000,000 convertible advance (claims #13900, #13901), lender warrants (claims #13874-13876, #13848), 8.5M shares issued to Somerset for a lease amendment (claim #13700), and sub-$10M placements and unit sales to affiliated vehicles such as Allin WULF and Bayshore (claims #13554, #13534, #13535). The Yorkville note was then amended on conversion price and maturity (claims #13463, #13464).

By early 2023, equity issuance is explicitly tied to lender relief. The Fifth Amendment conditions amortization suspension on $33.5M of equity by March 15, 2023 (claims #13210, #13197), and maturity extension on a $40M repayment by April 1, 2024 (claim #13199). The February 2023 offering at $0.68 per share (claim #13404) closed at roughly $27 with the over-allotment (claim #13309), which on its own sits below that threshold. A stated working capital deficiency of $112 (claim #12850) frames the period. Lender penny and dollar warrants (claims #12909, #12910) were later restated upward (claims #12150, #12160).

The stated Series A use of proceeds (claim #14277) is not quantified in the record. The miner, Talen and Nautilus commitments (claim #13712) are all delayed. How deployment tracked the original plan therefore requires explanation. The record contains no insider sales; insiders appear as buyers and warrant exchangers (claims #13429, #13401).

A careful reader should watch four things. The first is whether the $40M repayment condition was met. The second is how the roughly $27.8M penny warrant overhang (claim #12105) is exercised. The third is utilization of the renewed ATM (claim #12129). The fourth is the incentive plan's automatic share-pool increases through fiscal 2026 (claim #12667). The record ends in January 2024, so later events are not assessed here.

TeraWulf's record through early 2024 shows a capital-intensive buildout financed across nearly every available channel within about two years, with the mix drifting from growth capital toward lender accommodation as 2022 progressed. The early layer was conventional for a newly listed miner: the December 2021 term loan of $124M (claim #14134), a $200M ATM (claim #14102), a $500M universal shelf (claim #14103), and a Series A convertible preferred of roughly $10M (claim #14276). By mid-2022 the instruments become smaller and more bespoke. They include the Yorkville standby equity facility at $50M with a $15M convertible advance (claims #13900, #13901), whose conversion price was amended within five months (claim #13463), and direct unit placements of $1.75M to $2M with Allin WULF and Bayshore (claims #13534, #13535). That sequence reads more as incremental patching than as a planned capital stack.

Dilution accelerated into early 2023. The February 2023 offering priced at $0.68 per share for $25M (claim #13404). Lenders received penny warrants of roughly 26.7M shares and dollar warrants of 13.3M shares (claims #13211, #13212), later restated at about 27.8M and 13.9M (claims #12150, #12151). Those warrants accompanied the Fifth Amendment, which relieved amortization conditional on $33.5M of equity raised by March 15, 2023 (claim #13210) and extended maturity contingent on a $40M repayment by April 1, 2024 (claim #13199). The equity raise was therefore substantially tied to debt relief rather than new deployment, against a disclosed working capital deficiency of $112M (claim #12850).

On use of proceeds, the record is thinner than the financing record. The Series A purpose is stated (claim #14277), but the Bitmain purchase agreements of $33M and $169M (claims #14136, #14137), the $156M Talen joint venture commitment (claim #14138), and the $17M Nautilus contribution (claim #13712) are all tagged delayed. The mapping from capital raised to capacity deployed requires explanation from later filings. The record contains no insider sales; insider participation appears only as purchases, such as the $5M existing-investor placement including the CEO (claim #13976).

The claims end in January 2024, so they do not show whether the April 2024 repayment condition was met or how much of the ATM, re-registered repeatedly at $200M (claims #12254, #12129), was drawn. A sophisticated reader should watch whether the term loan was retired or refinanced, the exercise status of the lender warrants, and whether subsequent financings return to conventional instruments.

TeraWulf's record shows dilution that accelerated while the unit price of that dilution worsened. The early 2022 stack was institutional in form: a term loan (claim #14100), a $200M ATM backed by a $500M shelf (claims #14143, #14103), and a $10M Series A convertible preferred (claim #14276). By mid-2022 the company was layering on more expensive instruments. These included a $50M Yorkville standby facility with a $15M convertible advance (claims #13900, #13901), whose conversion price was later amended (claim #13463); lender warrants (claim #13848); and 8.5M shares issued to Somerset for a lease amendment (claim #13700). The February 2023 offering priced at $0.68 per share (claim #13404), roughly $27M including the over-allotment (claim #13309). That raise sits alongside the Fifth Amendment, which tied amortization relief to a $33.5M equity raise (claim #13210) and added 26,666,669 penny warrants and 13,333,333 dollar warrants (claims #13211, #13212). Later filings restate these warrant counts slightly higher (claims #12150, #12151), which requires explanation. The mix reads less as a planned capital structure than as sequential patching around the term loan. Each equity raise appears tied to lender concessions, and the $200M ATM is re-authorized in nearly every filing (claims #12870, #12254, #12129) without a disclosed drawdown figure. Insider activity runs one direction: participation by the CEO and affiliated entities in private placements and warrant exchanges (claims #13976, #13429, #13401, #13534). No insider sales are on record in this dataset. Use-of-proceeds tracking is thin. A stated intended use exists for the Series A (claim #14277), but its substance is not captured. Several deployment items carry delayed status, including the Bitmain purchase agreements (claims #14136, #14137) and the Talen joint venture contribution (claim #14138). The later Nautilus cash contribution figure (claim #13712) is materially smaller than the original commitment, and the record does not reconcile the two. A reported working capital deficiency (claim #12850) frames the period. Several values are recorded in inconsistent units, which limits precise comparison. A careful reader should watch whether the $40M repayment condition for the maturity extension (claim #13213) is met, how much of the ATM is actually drawn, and whether the automatic share-pool increases through fiscal 2026 (claim #12667) add a steady baseline of dilution. The record ends in January 2024, so later developments are not reflected here.

TeraWulf's disclosed financing record shows a company that raised capital in structured, conventional rounds through 2021 and then shifted into frequent, smaller raises from mid-2022 onward. The 2021 base consisted of private placements and preferred stock (the March 2021 placement, per claim #14098; the June 2021 Series A preferred, per claim #14099; the December 2021 placements, per claim #14101) plus a term loan reported at $124 (presumably millions), per claim #14134. After that, raises came roughly every one to three months. They included a $20,000,000 public offering alongside a $5,000,000 purchase by existing investors including the CEO (claims #14066 and #13976), a $50,000,000 Yorkville standby facility with a $15,000,000 convertible advance (claims #13900 and #13901), unit placements of $1.75M to $2M in October 2022 (claims #13534 and #13535), and a $25,000,000 offering priced at $0.68 per share in February 2023 (claim #13404). The falling offering price, and the shift toward related-party and small direct placements, point to rising dilution per dollar raised rather than a planned growth program.

The mix reads more as liquidity management shaped by lenders than as a coherent capital plan. The Yorkville note's conversion price was amended and its maturity became extendable (claims #13463 and #13464). Term-loan amortization relief was tied to a $33,500,000 equity raise (claim #13210) and paid for with penny and dollar warrants to lenders. Those warrant counts later appear as 27,759,265 and 13,879,630 (claims #12150 and #12151), up from 26,666,669 and 13,333,333 (claims #13211 and #13212), and the increase requires explanation. The $200,000,000 ATM was re-disclosed repeatedly (claims #13972, #12254, and #12129) without any utilization data on record. A working-capital deficiency reported as $-112 (claim #12850) provides context for these decisions.

Use of proceeds cannot be tested against deployment. The intended use of the Series A proceeds is noted (claim #14277) but not detailed, and the stated deployment targets are all marked delayed: the Bitmain purchase agreements (claims #14136 and #14137), the Talen joint venture commitment (claim #14138), and the Nautilus contribution (claim #13712). After early 2023, the record discusses lender covenants far more than these projects.

A careful reader should watch four things. The first is whether the $40,000,000 repayment that conditions the maturity extension to December 2024 was met (claim #13199). The second is the exercisability of the April 2024 warrants (claim #12133). The third is actual ATM draws. The fourth is the Omnibus plan's automatic share-pool increases through fiscal 2026 (claim #12667). The record ends in January 2024, so later developments are not reflected here.

TeraWulf's record through January 2024 describes a company financing its buildout in many small increments rather than a few large, planned raises. The early structure had some coherence. A $124M term loan closed in December 2021 (claim #14134), a $200M at-the-market agreement was in place by March 2022 (claim #14102), and a $500M universal shelf backed both (claim #14103). The ATM authorization is restated repeatedly through November 2023 (claims #12870, #12255, #12129), but the record does not show how much was actually drawn, so its role in funding requires explanation.

From mid-2022 the pattern shifts toward patching. A $50M Yorkville standby equity facility and a $15M convertible advance (claims #13900, #13901) were followed within months by an amended conversion price (claim #13463). Several sub-$10M private placements and direct unit sales followed (claims #13554, #13534, #13535). Each round of lender accommodation carried a warrant cost. That cost grew over time: the initial lender warrants (claims #13874 to #13876, #13848) were followed by penny and dollar warrants of roughly 26.7M and 13.3M (claims #13211, #13212). Those figures were later restated at about 27.8M and 13.9M (claims #12150, #12151), which suggests an adjustment the data does not explain. The February 2023 offering priced at $0.68 per share (claim #13404) marks the low point of that period.

The debt terms were revised repeatedly. Amortization relief was made conditional on a $33.5M equity raise (claims #13197, #13210). A maturity extension depends on a $40M repayment by April 2024 (claim #13213). Amortization suspension appears again in August and November 2023 (claims #12252, #12122). Many of these items carry delayed status, as do the Talen joint venture commitment (claim #14138) and the Bitmain purchase agreements (claims #14136, #14137).

A stated use of proceeds is on record for the Series A (claim #14277). The data does not provide enough deployment detail to test it against actual spending. The insider activity on record is participation rather than selling: the CEO joined a $5M private investment (claim #13976), and a Prager-affiliated entity took part in a $12M warrant exchange (claim #13429). No insider sales appear.

The record also shows unit inconsistencies. For example, the Series A appears as both $10,000,000 and $10,000 (claims #14276, #14243). A reader should watch four things next: whether the April 2024 repayment condition was met, how much of the ATM was actually used, whether the warrant overhang converts into shares, and whether any financings after January 2024 extend this incremental pattern or replace it.

TeraWulf's record through early 2024 describes a company financing its buildout in small, frequent increments rather than through a planned capital stack. The early framework looked orderly: a $124 term loan closed in December 2021 (claim 14134), a $500 universal shelf (claim 14103), and a $200 ATM (claim 14102). What followed was a sequence of patches. Within months the company added a $50,000,000 Yorkville standby equity facility and a $15,000,000 convertible advance (claims 13900, 13901). It then ran low-seven-figure unit and warrant placements in October 2022 (claims 13534, 13535, 13554) and a $25,000,000 offering priced at $0.68 per share in February 2023 (claim 13404). The declining offering sizes and the per-share price point to dilution that accelerated as the equity cheapened, not dilution that tapered off.

The mix reads as reactive. Lender warrants recur at nearly every credit amendment: $5,787,732 in mid-2022 (claim 13848), then penny and dollar warrants of $26,666,669 and $13,333,333 (claims 13211, 13212), later restated at $27,759,265 and $13,879,630 (claims 12150, 12151). This suggests the term loan's flexibility was bought with equity each time. The Fifth Amendment tied amortization relief to a $33,500,000 equity raise (claim 13210) and maturity extension to a $40,000,000 repayment (claim 13213), so the credit terms effectively dictated equity issuance. Insider participation, including the CEO in a $5,000,000 private investment (claim 13976) and the Prager-entity share-for-warrant exchange (claim 13429), reads as support capital rather than selling, though the record contains no discrete insider-sale data.

Use of proceeds is hard to verify. Claim 14277 records an intended use but does not detail it. The commitments that proceeds would fund, including the $156 Talen joint venture contribution (claim 14138), the Bitmain agreements (claims 14136, 14137), and the $17 Nautilus contribution (claim 13712), are marked delayed, and the working capital deficiency of $112 (claim 12850) requires explanation against the capital raised.

A careful reader should watch whether the $40,000,000 repayment condition (claim 13213) was met by April 2024, and whether the amortization suspension (claim 12122) became permanent. They should also watch whether the penny warrants were exercised in full (claim 12105). The inconsistent unit reporting across claims, such as $10 versus $10,000,000 for the same Series A (claims 14243, 14276), limits precise dilution arithmetic, and the record stops in January 2024.

TeraWulf's record through early 2024 shows dilution arriving in steady, small increments rather than through a single large raise. The March 2022 Series A preferred (claim 14276) was followed within a month by a priced common offering and a parallel placement from existing investors (claims 14066 and 14048). By June a Yorkville standby equity facility with a convertible advance had been added (claims 13900 and 13901). The size of each raise then shrinks: roughly $9.5M in October 2022 (claim 13554), low single-digit-million unit and warrant placements (claims 13534, 13535, 13435), and a $25M offering at $0.68 per share in February 2023 (claim 13404). Lower prices and smaller tickets point to a financing window that narrowed through 2022.

The mix looks more like balance-sheet management under lender pressure than a pre-set capital plan. The term loan (claim 14134) drives much of the later activity. Successive rounds of lender warrants were issued, including penny and dollar warrants whose reported figures rise from claim 13211 to claim 12150. The Fifth Amendment tied amortization relief to a $33.5M equity raise (claim 13210). The February 2023 offering therefore appears to have been sized partly to meet a covenant, not solely to fund operations. The $200M ATM appears repeatedly from March 2022 through November 2023 (claims 14135, 12129). The data does not show how much was actually drawn, which leaves its practical role unclear. Insiders consistently bought or exchanged into the company, including the CEO's participation (claim 13976) and the Prager-entity exchange (claim 13429). No insider sales appear in the record.

Use of proceeds is hard to test. The stated intent for the Series A (claim 14277) has no deployment figure attached. Several capital commitments are marked delayed, including the Bitmain purchase agreements (claims 14136, 14137), the Talen joint venture contribution (claim 14138), and the Nautilus cash contribution (claim 13712). The working capital deficiency of roughly $112M (claim 12850) indicates that early commitments ran ahead of available funding.

A careful reader should look next for how the term loan was resolved against the $40M paydown condition for the December 2024 maturity extension (claim 13199). The other open items are actual ATM utilization and the exercise pace of the outstanding penny warrants (claim 12105), which is the largest overhang still pending as of this record.

TeraWulf's disclosed record through early 2024 shows a financing cadence that accelerated as the build-out proceeded. The starting capital structure was conventional: a $124M term loan closed December 2021 (claim #14134) alongside $76M of December 2021 private placements (claim #14101). By mid-2022 the company had added several more sources within about three months. These were a $200M ATM (claim #13972), a roughly $20M underwritten offering paired with a $5M private investment by existing holders including the CEO (claims #14066, #13976), and a $50M Yorkville standby equity facility with a $15M convertible advance (claims #13900, #13901). The Yorkville note's conversion price was later amended and its maturity was given an extension option (claims #13463, #13464), which indicates the instrument was renegotiated rather than run off as first designed.

The mix reads less as a single capital plan than as a series of responses to the term loan. Lender warrants recur: 5,787,732 were issued in mid-2022 (claim #13848), followed by 26,666,669 penny warrants and 13,333,333 dollar warrants tied to the Fifth Amendment (claims #13211, #13212). Amortization relief was explicitly conditioned on a $33.5M equity raise (claim #13210), and extending maturity required a $40M paydown (claim #13199). The February 2023 offering priced at $0.68 per share (claim #13404) came alongside a stated $112M working capital deficiency (claim #12850). That places the period's equity issuance in the context of lender conditions rather than discretionary growth funding. Small related-party placements, warrant exchanges, and share issuances to Beowulf E&D for services (claim #12155) add dilution in increments that are individually minor but frequent.

Use-of-proceeds disclosure in this record is thin. The Series A intended-use claim (#14277) is recorded without detail. The Bitmain purchase agreements and the Talen joint venture commitment (claims #14137, #14136, #14138) are all marked delayed, so these claims alone cannot confirm deployment against the original equipment and joint venture plan.

A careful reader should watch three things. The first is whether the $40M repayment condition (claim #13199) was met by April 2024. The second is how much of the repeatedly referenced $200M ATM authorization (claims #12254, #12129) was actually drawn. The third is the exercise status of the penny warrants, which reappear in January 2024 (claim #12105). The record contains no insider sale transactions and ends in early 2024, so conclusions about later capital strategy would require data not shown here.

TeraWulf's disclosed record from March 2022 through January 2024 shows a company financing a capital-intensive buildout on progressively tighter terms. The early structure was conventional for a pre-revenue miner. It combined the $124 million term loan closed in December 2021 (claim 14134), private placements totaling $76 million (claim 14101), a $500 million universal shelf (claim 14103) and a $200 million ATM (claim 14135). Within three months the funding moves toward smaller and more expensive sources. These include a $50 million Yorkville standby equity facility with a $15 million convertible advance (claims 13900, 13901), whose conversion price was later amended (claim 13463), and lender warrants tied to an additional $50 million term loan facility (claims 13873, 13848). Dilution steepens through early 2023. The February 2023 offering priced at $0.68 per share for $25 million (claim 13404). Around the same time, lenders received 26.7 million penny warrants (claim 12909) and 13.3 million dollar warrants (claim 12910) in exchange for amortization relief. That relief depended on raising $33.5 million of equity by March 15, 2023 (claim 13197) and repaying $40 million by April 2024 (claim 13199), so equity issuance was effectively contractually required to service the debt. Warrant counts later rose to 27.8 million penny and 13.9 million dollar warrants (claims 12159, 12151), a change that requires explanation. Insiders repeatedly put money in, including the investment by existing shareholders including the CEO (claim 13976) and the Prager-entity exchange (claim 13429). This reads as sponsor support rather than distribution, and no insider sales appear on record. The mix looks more like sequential patching around the term loan than a pre-set plan. Each new instrument responds to a covenant or liquidity constraint created by the one before it, which the $112 million working capital deficiency disclosed in March 2023 (claim 12850) makes plain. On use of proceeds, the record is thin. Claim 14277 refers to an intended use for the Series A proceeds without detail. The Talen joint venture commitment (claim 14138) and the Bitmain purchase agreements (claims 14136, 14137) are marked delayed. Deployment therefore cannot be matched against stated purpose from this data. A sophisticated reader should watch three things: whether the April 2024 condition to repay $40 million was met, how much of the $200 million ATM (claim 12254) has been drawn, and when the lender warrants are exercised. Several claims mix unit conventions, and the record ends in January 2024, so later developments are not reflected here.

TeraWulf's disclosed record from March 2022 through January 2024 describes a company financing construction under pressure rather than on a set plan. The cadence is dense. The record shows the Series A convertible preferred at $10,000,000 (claim #14276), a priced public offering the following month at $20,000,000 (claim #14066), and a Yorkville standby facility with a $15,000,000 convertible advance by June (claims #13900 and #13901). Unit placements with related parties followed in October 2022 (claims #13534 and #13535), and an underwritten offering at $0.68 per share raised $25,000,000 in February 2023 (claim #13404). Moving from an April 2022 raise to a sub-dollar February 2023 raise indicates that dilution accelerated as the share price fell, with each dollar costing more shares.

The mix reads more as sequential patching than as a single strategy. A $200 million ATM (claim #13972) ran alongside repeated small private placements. The term loan relationship produced successive layers of lender warrants, first grants described at 3.75% to 5.0% dilution (claims #13875 and #13874). Later came penny and dollar warrants of 26,666,669 and 13,333,333 (claims #13211 and #13212), subsequently restated at 27,759,265 and 13,879,630 (claims #12159 and #12160). Amortization relief was conditioned on a $33.5 million equity raise (claim #13197), and maturity extension depended on a $40 million repayment (claim #13199). That made issuance a lender obligation rather than a discretionary choice. Insiders appear as buyers, including the CEO-linked investment (claim #13976) and the $1,250,000 Allin WULF warrant purchase (claim #13401). The record contains no insider sales.

Use of proceeds is difficult to trace. The Series A's intended use is noted (claim #14277) without deployment detail. The Bitmain purchase agreements (claims #14137 and #14136) and the Talen joint venture commitment (claim #14138) carry delayed status. A working capital deficiency of $112 million as of March 2023 (claim #12850) is the clearest sign that commitments outran funding.

Several amounts are recorded in inconsistent units. Claim #14243 lists the Series A at $10,000 against $10,000,000 elsewhere, so individual figures warrant checking against filings. A reader should watch whether the ATM, reaffirmed through November 2023 (claim #12129), becomes the primary funding channel. Other open items are whether the April 2024 warrants (claim #12133) and the conditional maturity extension are satisfied, and whether the delayed Nautilus and Talen contributions are later reported as completed or revised. The record ends in January 2024 and says nothing about subsequent events.

TeraWulf's record describes a company financing continuously rather than episodically. Between March 2022 and early 2024 there is no quarter without a new equity, warrant or debt-modification event. Many figures are recorded without a unit scale; they are read here as millions where adjacent claims make that explicit, as with the $200 million ATM in claim #13972.

The structure was built quickly in early 2022. It paired a $124 term loan (claim #14100) with the $200 ATM (claim #14102), a $20,000,000 public offering (claim #14066) and a $50,000,000 Yorkville standby facility with a $15,000,000 convertible advance (claims #13900, #13901). What followed looks more like patching than a coherent plan. Unit placements of a few million dollars in October 2022 (claims #13534, #13535) came alongside $7.5 million term loan draws (claim #13555). The Yorkville conversion price was amended (claim #13463). In February 2023 the company priced an underwritten offering at $0.68 per share for $25,000,000 (claim #13404).

The lenders' position is the most instructive thread. Amortization relief was made conditional on a $33.5 million equity raise (claim #13210). It was paid for with 26,666,669 penny warrants and 13,333,333 dollar warrants (claims #13211, #13212). Those counts were later revised upward to 27,759,265 and 13,879,630 (claims #12159, #12160). The relief itself was then carried forward through repeated amendments (claims #12260, #12122). This sat against a stated working-capital deficiency of $112 (claim #12850). Dilution here came less from priced offerings than from warrant compensation for debt accommodation.

Insiders consistently supplied capital rather than taking it out. The record shows a CEO-inclusive $5,000,000 investment (claim #13976) and a $12,000,000 warrant exchange with a Prager entity (claim #13429). No insider sales appear in the data.

Use-of-proceeds tracking is thin. Claim #14277 records an intended use without detail. The deployment commitments that are named all carry delayed status: the Bitmain agreements of $169 and $33 (claims #14137, #14136), the Talen joint-venture commitment of $156 (claim #14138) and the Nautilus contribution of $17 (claim #13712). The ATM authorization is re-stated repeatedly through November 2023 (claim #12129), but the record does not show how much of it was actually drawn.

A careful reader should watch four things. First, whether the $40 million repayment condition due by April 1, 2024 (claim #13213) is met. Second, the exercise pace of the lender warrants. Third, any disclosed ATM utilization. Fourth, the equity-plan share pool, which increases automatically through fiscal 2026 (claim #12667).

TeraWulf's disclosed record, which runs from March 2022 to January 2024, shows capital raised in frequent, small increments. There is no sign of a few planned tranches. By the March 2022 filing the company already carried a $124 term loan (claim #14100), $76 in December 2021 private placements (claim #14101) and a $200 ATM authorization (claim #14102). Against these sat miner purchase commitments of $169 and $33 (claims #14137, #14136) and a $156 Talen joint venture commitment (claim #14138), all marked delayed. Within three months the company added a $50,000,000 Yorkville standby facility with a $15,000,000 convertible advance (claims #13900, #13901). By October 2022 it was placing units of $1,750,000 to $2,000,000 with related holders (claims #13533 through #13535). The cadence reads as serial gap-filling rather than a single capital plan. Several figures on record carry no stated unit, which limits precise comparison.

Over time the mix shifted toward accommodating the term lenders. Lender warrants appear in mid-2022 (claims #13874 through #13876). The Fifth Amendment then made amortization relief conditional on $33.5 million of equity by March 15, 2023 (claim #13197). It paired that condition with penny and dollar warrants of 26,666,669 and 13,333,333 (claims #13211, #13212), later restated as 27,759,265 and 13,879,630 (claims #12159, #12160). The same filing disclosed a working capital deficiency of $112 (claim #12850). The February 2023 offering at $0.68 per share (claim #13404) closed at $27 with the over-allotment (claim #13309), which appears to satisfy the equity condition, although the related amendment claims remain marked delayed. No insider sales appear. Insider activity runs the other way, with the CEO participating in the April 2022 raise (claim #13976) and a Prager-entity share-for-warrant exchange (claim #13429).

On use of proceeds, the record states an intended use for the Series A (claim #14277) but contains few deployment claims. The main ones are $60 in miner deposits (claim #13878) and a $17 Nautilus contribution (claim #13712, delayed). Whether proceeds tracked their stated purpose cannot be established from this data and requires explanation.

There are four items to watch. The first is the $40 million repayment by April 1, 2024, on which the extension of maturity to December 2024 depends (claim #13199). The second is the April 2024 warrants (claim #12133). The third is draws on the $200 ATM, which was reiterated through November 2023 (claim #12129) with no disclosed utilization. The fourth is the incentive-plan pool, which auto-increases through fiscal 2026 (claim #12667). Events after January 2024 are not covered here.

TeraWulf's record through early 2024 describes a company that financed its build-out in many small, overlapping tranches. The cadence was steady rather than episodic. Capital was raised every few weeks, and the terms grew more dilutive as the equity price fell. The April 2022 offering was a $20M underwritten deal alongside a $5M insider-inclusive placement (claims #14066 and #13976). By February 2023, the company was selling stock at $0.68 per share for $25M (claim #13404). The trajectory runs from growth equity toward rescue equity. The mix reads as reactive patching rather than a planned capital structure. A $200M ATM (claim #14102) and a $500M shelf (claim #14103) were in place from March 2022. Even so, the company went on to add a Yorkville standby facility with a $15M convertible advance (claims #13900 and #13901), whose conversion price and maturity were later amended (claims #13463 and #13464). It also added small direct unit placements in the $1.75M–$2M range (claims #13534 and #13535) and repeated warrant issuances to lenders. That lender dilution is the most consequential thread. It began with warrant grants in mid-2022 (claims #13874–13876) and expanded to roughly 26.7M penny warrants and 13.3M dollar warrants (claims #13211 and #13212), with the penny count restated at about 27.8M later (claim #12150). These warrants were granted in exchange for amortization relief that was itself conditional on raising $33.5M of equity (claim #13210). The debt was being serviced with equity, and the lenders took equity-linked compensation for their patience. On use of proceeds, the record supports only a partial comparison. The stated intended use of the Series A proceeds (claim #14277) has no matching deployment claim. The large announced commitments were all delayed: the $156M Talen joint venture contribution (claim #14138), the Bitmain purchase agreements of $169M and $33M (claims #14137 and #14136), and the Nautilus cash contribution (claim #13712). A working capital deficiency of $112M (claim #12850) suggests commitments outran funding, though the data do not show how each shortfall was resolved. The record contains no insider sales. Insider activity appears only as participation in placements and warrant exchanges (claims #13429 and #13401). A careful reader should watch whether the company meets the $40M repayment that conditions its maturity extension (claim #13213). They should also watch how much of the $200M ATM (claim #12255) is drawn to meet it, and whether the auto-increasing incentive pool (claim #12667) adds a further layer of dilution.

The shape here is a company that front-loaded conventional project capital and then spent two years refinancing the consequences. The December 2021 term loan of $124M (claim #14100) plus the $200M at-the-market program and $500M shelf authorized in March 2022 (claims #14102, #14103) were sized against announced capex: the $169M and $33M Bitmain purchase agreements (claims #14137, #14136) and the $156M Talen joint venture contribution commitment (claim #14138). All three deployment commitments carry delayed status while the financing side kept executing, which is the central tension in the record.

On cadence, increments shrink and repeat through 2022 into 2023 — the $10M Series A (claim #14276), the $20M public offering with $5M concurrent insider placement (claims #14066, #14048), a $9.5M private placement (claim #13554), $1.75M and $2M unit placements (claims #13534, #13535), and by March 2023 convertible notes of $853,912 (claim #13164) converting almost immediately (claim #13177). The February 2023 offering priced at $0.68 (claim #13404) marks the trajectory's low point on this record.

The mix reads as covenant-driven more than opportunistic. The $33.5M equity commitment due March 15, 2023 (claims #13197, #13210) was the consideration for amortization relief under the Fifth Amendment (claim #13155), and the February raise with over-allotment, $27M (claim #13309), delivered it — announced use tracked actual deployment there. Less legible is the Yorkville $50M standby with its $15M convertible advance (claims #13900, #13901), subsequently amended on both conversion price and maturity (claims #13463, #13464); bridge capital that required renegotiation.

Lender warrant compensation escalates and is revised upward: percentage-of-dilution warrants mid-2022 (claims #13874 through #13876), penny warrants at 26,666,669 restated to 27,759,265 (claims #13211, #12150, #12105), dollar warrants from 13,333,333 to 13,879,630 (claims #13212, #12160).

Insider flow is one-directional purchase — CEO and existing holders (claim #13976), Allin WULF, Bayshore, Lake Harriet, the Stammtisch warrant exchange (claim #13400). No disclosed insider selling appears.

Watch three things: whether the $40M repayment by April 1, 2024 extending maturity to December 1, 2024 (claim #13199) occurred; whether the repeatedly re-announced $200M ATM (claims #12254, #12261, #12129) is fresh or re-disclosed capacity against the $112M working capital deficiency (claim #12850); and the record's silence after January 2024.

TeraWulf's record divides cleanly at mid-2022. The 2021 stack was discrete and build-sized: a March 2021 private placement of $30 million (claim 14098), a $50 million Series A in June (claim 14099), $76 million of December private placements (claim 14101) and a $124 million term loan closed December 1 (claim 14100). That is coherent project finance matched to a stated capex program — the Bitmain orders of $169 million and $33 million (claims 14137, 14136) and the $156 million Talen joint venture commitment (claim 14138).

What follows does not hold that shape. Ticket size collapses while frequency rises: $9.5 million in October 2022 (claim 13554), unit placements of $1.75 million and $2.0 million (claims 13534, 13535), a convertible note of $853,912 (claim 13164). The Yorkville standby facility of $50 million with a $15 million advance (claims 13900, 13901), later amended on conversion price (claim 13463), is a liquidity backstop rather than a capital plan. Consideration increasingly leaves in shares instead of cash — 8.5 million shares to Somerset on a lease amendment (claim 13700), 2,460,513 shares to Beowulf E&D for services (claim 12155). Most telling, lender warrants are denominated in percent of the company, at 5.0, 3.75 and 4.25 percent (claims 13874, 13875, 13876), and the penny warrant pool expands from 26,666,669 to 27,759,265 (claims 13211, 12105). Lenders were pricing ownership, not credit. Insider participation runs alongside throughout, including the CEO in the $5 million April 2022 placement (claim 13976) and the Allin WULF warrant purchase of $1.25 million (claim 13401); the marginal buyer is repeatedly a related party.

The use-of-proceeds question answers itself in March 2023, when the $33.5 million equity raise became an explicit condition of amortization relief (claims 13197, 13210, 13198). Equity was funding debt service while miner and joint-venture milestones sat delayed, against a $112 million working capital deficiency (claim 12850). That is redirection, not execution.

Three items require explanation. The $200 million ATM is authorized or re-announced across at least five dates (claims 14102, 13972, 12931, 12254, 12129) against disclosed net equity-and-debt proceeds of only $36 million (claim 12259) — unclear whether restated capacity or stacked. Whether the $40 million repayment extending maturity to December 2024 (claim 13199) occurred. And the record stops in January 2024; roughly twenty months of capital history is absent here, which bounds any conclusion about where the structure settled.

The shape here is a financing cadence that accelerated as ticket size collapsed. The 2021 stack was conventional for an infrastructure build — $30 million in March (claim 14098), $50 million of Series A preferred in June (claim 14099), $76 million of December placements (claim 14101) alongside a $124 million term loan (claim 14100) — and it was followed in March 2022 by the scaffolding of a standing program: a $500 million shelf (claim 14103) and a $200 million ATM (claim 14143). What follows does not read as that program being executed on a schedule. The raises fragment instead: $20 million public plus $5 million from existing holders including the CEO in April 2022 (claims 14066, 13976), a $50 million Yorkville standby facility drawn initially as a $15 million convertible advance in June (claims 13900, 13901), then a $9.5 million placement and $1.75 million and $2 million unit placements in October (claims 13554, 13534, 13535) against a term loan now disbursing in two $7.5 million tranches (claims 13555, 13556). A lender that funded $124 million in a single December 2021 draw was funding in $7.5 million increments ten months later; that compression is the most informative fact in the record.

The mix looks lender-directed rather than strategic. The February 2023 offering at $0.68 raising $25 million and upsized to roughly $27 million (claims 13404, 13309) sits against a $112 million working capital deficiency (claim 12850) and a Fifth Amendment conditioning amortization relief on $33.5 million of equity by March 15, 2023 (claims 13197, 13210) — equity raised to satisfy a covenant, not to fund a machine. The cost shows in warrants sized as percentages of the company (5.0%, 3.75%, 4.25%, claims 13874, 13875, 13876) and penny warrants ratcheting from 26,666,669 to 27,759,265 shares (claims 13211, 12150, 12105).

Notably, the funding-side claims are stated while the deployment-side claims are delayed: Bitmain at $169 million and $33 million (claims 14137, 14136), the $156 million Talen contribution (claim 14138), the $17 million Nautilus contribution (claim 13712). Raising tracked the plan; spending requires explanation.

Watch whether the $40 million April 2024 repayment condition (claim 13199) was met, whether amortization suspension repeated in August and November 2023 (claims 12252, 12122) became permanent, and share count rather than announcements given repeated $200 million ATM re-authorizations (claims 12254, 12129) and the omnibus plan auto-increase through 2026 (claim 12667). The record contains insider purchases only (claims 13976, 13401, 13395) and no dispositions, and ends January 2024, so the later period is unread here.

The record divides cleanly at mid-2022. Through the first half, the mix reads as coherent project finance: a $124 million term loan closed December 1, 2021 (claims 14100, 14134) alongside preferred equity and private placements, with a $500 million shelf (claim 14103) and a $200 million ATM (claim 14102) established in March 2022 as headroom rather than as a funding source. Most of what follows is patching. The $10 million Series A convertible preferred carrying a 10% dividend (claims 14276, 14278) is small against the buildout it was meant to support; the June 2022 Yorkville standby equity purchase agreement and its $15 million convertible advance (claims 13900, 13901) is the structure reached for when underwritten capacity has closed; and by October 2022 the company was taking $1.75 million and $2.0 million unit placements from named affiliates (claims 13534, 13535) while drawing term loan tranches in $7.5 million pieces (claims 13555, 13556).

Price marks the trajectory. The April 2022 underwritten offering raised roughly $20.6 million (claim 13975); the February 2023 offering cleared at $0.68 per share, $24.08 million net (claims 13404, 13405), reaching $27 million with the over-allotment (claim 13309). Dilution in that window came less from cash sales than from warrants issued to lenders as consideration for covenant relief — 26,666,669 penny and 13,333,333 dollar warrants under the Fifth Amendment (claims 13211, 13212), adjusted upward by September 2023 to 27,759,265 and 13,879,630 (claims 12159, 12160). Whether that ratchet runs again is the most consequential mechanical item on file.

Announced use and deployment diverge at the hardware and joint-venture commitments. The Bitmain purchase agreements (claims 14136, 14137), the $156 million Talen contribution (claim 14138), and the $17 million Nautilus contribution against a $25 million minimum ownership threshold (claims 13712, 13717) are all recorded as delayed, while $60 million in miner deposits (claim 13878) is stated.

Watch ATM utilization against the repeatedly re-authorized $200 million (claims 12255, 12129) versus the $36 million actually reported in August 2023 (claim 12259); whether the $40 million April 2024 repayment condition (claim 13199) was met; and the absence of any insider sale on record — participation is uniformly on the buy side, including the CEO's $5 million (claim 13976) and the $12 million Prager-entity share-for-warrant exchange (claim 13429), the latter requiring explanation given its timing. The record ends January 2024 (claim 12105); nothing here supports a current read.

The disclosed record runs from March 2022 to January 2024 and contains almost no quiet quarters. Within six weeks the company layered a $10M Series A convertible preferred (claim #14276) onto the $124M December 2021 term loan (claim #14100), a $200M ATM program (claim #14102) and a $15M March 2022 direct offering (claim #14104), then priced a $20M public offering alongside a $5M concurrent insider placement in April (claims #14066, #14023), and by June added a $50M Yorkville standby equity facility with a $15M convertible advance drawn against it (claims #13900, #13901). Four instrument types in one quarter is the signature of patching, not of a designed capital stack.

The early raises did track identifiable spend — the December 2021 Bitmain orders at $33M and $169M (claims #14136, #14137) and the $156M Talen joint venture commitment (claim #14138) — but all three are carried as delayed, and the instruments that follow are defensive rather than expansionary. The Fifth Amendment conditioned amortization relief on raising $33.5M of equity by March 15, 2023 (claims #13197, #13210) and made the maturity extension contingent on a $40M repayment (claim #13213). The February 2023 underwritten offering cleared at $0.68 per share for roughly $27M including the over-allotment (claims #13404, #13309), against a stated working capital deficiency of $112M (claim #12850). Lender warrants escalate in step: 3.75%, 4.25% and 5.0% dilution grants (claims #13875, #13876, #13874), then penny and dollar warrants of 26.67M and 13.33M shares (claims #13211, #13212), upsized in September 2023 to 27.76M and 13.88M (claims #12150, #12151). Dilution was the currency used to buy time on the term loan.

Two absences matter. There are no insider sales on this record; insider activity runs the other way, through CEO-affiliated participation in the April 2022 placement (claim #13976), the $12M Stammtisch warrant exchange (claim #13429), and the Allin WULF and Lake Harriet purchases (claims #13401, #13395). And the repeated $200M ATM entries (claims #14102, #12870, #12255, #12254, #12129) are re-authorizations of one program rather than stacked capacity; the only utilization figure visible is $36M of net equity and debt proceeds in mid-2023 (claim #12259).

Worth watching: whether the $40M April 2024 test was met or revised, actual ATM draw against authorized capacity, and share count once the roughly 41M warrant shares settle. The record stops in January 2024 (claim #12106); anything later is unobserved here.

TeraWulf's record through early 2024 shows dilution that was frequent, small in individual size, and increasingly shaped by lender terms rather than by the company's own plan. The 2022 opening stack was broad. It combined a term loan (the $124 figure in claim #14134), a $200 ATM authorization (claim #13982), a $10,000,000 Series A convertible preferred (claim #14276), and an April public offering of $20,000,000 (claim #14066) alongside a $5,000,000 insider purchase that included the CEO (claim #13976). A structure that broad can be read as deliberate. By mid-2022, however, the instruments became more expensive. A Yorkville standby equity facility of $50,000,000 (claim #13900) and a $15,000,000 convertible advance (claim #13901) arrived alongside lender warrants (claim #13848). Within months, the conversion price and maturity of that advance were amended (claims #13463, #13464). Raises in October 2022 were smaller still: a $9,500,000 private placement (claim #13554) and unit placements of $1,750,000 and $2,000,000 (claims #13534, #13535). This sequence reads more like successive patching than a planned capital structure. The February 2023 offering priced at $0.68 per share (claim #13404) and closed at roughly $25,990,000 net (claim #13384). It is best read as a condition of the lenders, not an independent choice. The Fifth Amendment tied amortization relief to $33,500,000 of equity by March 15, 2023 (claim #13210). The same record shows a working capital figure of -$112 (claim #12850), and lender warrant counts rose from 26,666,669 (claim #13211) to 27,759,265 (claim #12159). The insider activity on record consists of purchases and exchanges, not sales, including the Prager-entity exchange (claim #13429). That shows continued insider commitment, but it also means insiders captured part of the discounted issuance. Use of proceeds is difficult to reconcile. The Series A's stated purpose (claim #14277) has no later deployment claim. The Bitmain purchase agreements (claims #14136, #14137), the $156 Talen joint venture commitment (claim #14138), and the Nautilus contribution (claim #13712) are all marked delayed, so announced deployment and actual deployment cannot be matched from this record. Coverage thins after late 2023, which limits any view of the current trajectory. A careful reader should watch four items: whether the $40 million repayment that conditions the maturity extension to December 2024 was met (claim #13199), exercise of the April 2024 warrants (claim #12133), how heavily the renewed $200,000,000 ATM (claim #12255) is drawn, and share-pool growth under the auto-increasing incentive plan (claim #12667).

TeraWulf's record describes a company that raised capital almost continuously from March 2022 through early 2024. The trajectory moves from institutional-style instruments toward smaller, more contingent ones. The early sequence includes the $10,000,000 Series A convertible preferred (claim #14276), the $200,000,000 ATM (claim #13972), and the $20,000,000 public offering priced in April 2022 (claim #14066). By mid-2022 the company had added a $50,000,000 Yorkville standby facility with a $15,000,000 convertible advance (claims #13900, #13901). By October it was raising in single-digit millions through a $9.5 million private placement and unit sales of $2,000,000 and $1,750,000 (claims #13554, #13535, #13534). The February 2023 offering at $0.68 per share for $25,000,000 (claim #13404) marks the low point of pricing in this record.

The mix reads less as a planned capital structure than as sequential accommodation of the term lenders. Lender warrants recur throughout the record, first at 5,787,732 (claim #13848), then as 26,666,669 penny warrants and 13,333,333 dollar warrants (claims #13211, #13212). Those counts later appear as 27,759,265 and 13,879,630 (claims #12159, #12160), a revision that requires explanation. Amortization relief was made conditional on a $33.5 million equity raise (claim #13210), which tied equity issuance directly to debt servicing. Insider activity in this record runs toward buying rather than selling. The CEO participated in the $5,000,000 existing-shareholder placement (claim #13976), and a Paul Prager entity exchanged shares for warrants valued at $12,000,000 (claim #13429). No insider sales are on record.

Use of proceeds is difficult to reconcile. The stated intent for the Series A (claim #14277) carries no detail here. The Bitmain purchase agreements (claims #14136, #14137), the Talen joint venture contribution (claim #14138), and the Nautilus cash contribution (claim #13712) are all tagged delayed. Meanwhile, the reported working-capital deficiency of $112 (claim #12850, units as disclosed) suggests that later raises went substantially to liquidity and lender terms rather than the originally described build-out.

The record thins after late 2023 and ends in January 2024. A reader should look for how the $40 million repayment condition tied to April 1, 2024 was resolved (claim #13199), whether the amortization suspension (claim #12122) became permanent, and how much of the repeatedly re-referenced $200 million ATM (claim #12129) was actually drawn. The incentive-plan share pool, which auto-increases through fiscal 2026 (claim #12667), is an additional source of dilution that sits outside the financing ledger.

TeraWulf's financing record from March 2022 through January 2024 reads less as a planned capital structure than as a sequence of responses to a balance sheet under strain. The cadence is dense: a Series A convertible preferred of $10 million (claim #14276), a $20 million public offering (claim #14066) alongside $5 million from existing investors (claim #14048), a $50 million Yorkville standby facility with a $15 million convertible advance (claims #13900, #13901), a $9.5 million private placement (claim #13554), and a $25 million offering priced at $0.68 per share (claim #13404) that closed near $27 million after the over-allotment (claim #13309). Insider participation runs toward buying rather than selling, including the CEO's inclusion in the April 2022 raise (claim #13976) and the Prager-linked warrant exchange (claim #13429); the record contains no insider sales.

The mix points to patching rather than strategy. Much of the dilution comes from lenders, not the market. Warrants accompanied the term loan amendments, and the recorded penny-warrant figure grew from 26,666,669 (claim #13211) to 27,759,265 (claim #12159), with dollar warrants rising from 13,333,333 to 13,879,630 (claims #13212, #12160). Amortization relief under the Fifth Amendment was conditioned on a $33.5 million equity raise (claim #13210). The $200 million ATM was re-filed or reaffirmed at least four times (claims #13972, #12870, #12272, #12129), but no drawn amounts are disclosed. The ATM therefore reads as standing optionality, not a demonstrated funding channel.

Use of proceeds is hard to reconcile with deployment because the data rarely states either. The capital commitments that are recorded are marked delayed. These include the Bitmain purchase agreements of $33 million and $169 million (claims #14136, #14137), the $156 million Talen joint venture commitment (claim #14138), and the $17 million Nautilus contribution (claim #13712). A working capital deficiency of $112 million (claim #12850) suggests that proceeds went toward bridging obligations rather than new capacity. That link is inferred and requires explanation from later filings.

A careful reader should first check whether the $40 million repayment due by April 1, 2024 was met, since it conditions the maturity extension to December 2024 (claim #13199). Beyond that, the questions are whether the ATM is actually drawn, and whether the lender warrants are exercised as the share price moves. The record ends in January 2024, and dollar units are recorded inconsistently across claims, so these observations cover the distressed period only.

TeraWulf's record reads less as a planned capital stack than as a sequence of financings, each shaped by the constraints of the one before it. The early architecture was conventional: a $124M term loan (claim 14100), a $500M universal shelf (claim 14103) and a $200M ATM (claim 14102). The additions that followed were progressively more expensive. The April 2022 public offering of $20M (claim 14066) came with a $5M purchase by existing investors (claim 14048). Two months later the company added the Yorkville standby equity facility of $50M (claim 13900) and a $15M convertible advance (claim 13901), whose conversion price was amended by October (claim 13463). An additional $50M term facility (claim 13873) arrived with lender warrants. By February 2023 common stock was being sold at $0.68 per share for $25M (claim 13404), and dilution had shifted from primary issuance toward warrant layering. The penny and dollar warrants to lenders are recorded at 26,666,669 and 13,333,333 (claims 13211, 13212) and later restated at 27,759,265 and 13,879,630 (claims 12150, 12160). The mix reads as sequential patching rather than a coherent plan, and the lenders increasingly set the terms. The Fifth Amendment tied amortization relief to a $33.5M equity raise (claim 13210), which placed debt covenants directly behind equity timing. The recorded working capital deficiency of $112M (claim 12850) is consistent with that reading. Insider behavior runs the other way from what such stress often produces: the data records insider-affiliated purchases and warrant exchanges (claims 13976, 13401, 13429) and no insider sales. Tracking use of proceeds is harder, because the stated intent for the Series A (claim 14277) is on record without detail. The deployment items that can be traced, namely the Talen joint venture commitment of $156M (claim 14138), the Bitmain purchase agreements (claims 14136, 14137) and the Nautilus contribution (claim 13712), are all classed as delayed, and that gap between announced and realized deployment requires explanation. A reader should next watch whether the $40M repayment needed by April 1, 2024 to extend maturity to December 2024 (claim 13199) is met, how much of the $200M ATM reauthorized in 2023 (claim 12272) is actually drawn, when the April 2024 warrants become exercisable (claim 12133), and how quickly the Omnibus plan's automatic share-pool increases (claim 12667) add to the dilution already recorded. The record ends in January 2024, so the outcome of these conditions is not yet visible.

TeraWulf's record reads as a company financing a capital-intensive buildout from a weak balance-sheet position, with the terms of each raise tightening as it went. The early stack combined private equity, preferred, and a term loan, including the December 2021 term loan of roughly $124M (claims #14100, #14134) and the $200M at-the-market program (claims #14102, #13972). By mid-2022 the mix had widened to a $50M Yorkville standby equity facility with a $15M convertible advance (claims #13900, #13901). Lender warrants followed, stepping from 3.75% to 5.0% dilution grants (claims #13875, #13874). The Yorkville note's conversion price was later amended (claim #13463). This sequence looks less like a planned capital structure than successive patches, each one layering dilution onto the last.

The early-2023 period is the clearest inflection. The underwritten offering priced at $0.68 per share for $25M (claim #13404), a level that compares poorly with the April 2022 public offering (claim #14066). The Fifth Amendment then tied amortization relief to a $33.5M equity raise (claims #13210, #13198). It also conditioned the maturity extension on a $40M repayment by April 2024 (claim #13199) and added penny and dollar warrants of about 26.7M and 13.3M to lenders (claims #13211, #13212), later restated at roughly 27.8M and 13.9M (claims #12150, #12160). A reported working capital deficiency of $112M (claim #12850) is the context for all of this. Insider participation appears as participation rather than selling: a $5M purchase by existing holders including the CEO (claim #13976) and a warrant exchange with a Prager-affiliated entity (claim #13429). The record contains no disclosed insider sales.

Use of proceeds cannot be fully tested here. The Series A intended use is on record (claim #14277), but its substance is not captured. Many dollar figures appear in inconsistent units, for example $10,000 versus $10 for the same Series A (claims #14243, #14244). The committed deployments, namely the Bitmain miner agreements (claims #14136, #14137), the Talen joint venture contribution (claim #14138), and the Nautilus contribution (claim #13712), are all marked delayed, which requires explanation against the capital raised.

A careful reader should watch four things: whether the $40M repayment condition (claim #13199) was met, how much of the recurring $200M ATM authorization (claims #12255, #12129) was actually drawn, whether the dilutive lender warrants were exercised, and whether the delayed deployment commitments were completed or quietly revised. The record ends in January 2024 (claim #12106), so later developments are not reflected here.

TeraWulf's record describes a company that financed a capital-intensive buildout during a period when the market kept narrowing its options. It started in early 2022 with a full shelf-and-ATM architecture: a $200 million at-the-market agreement (claim 13972), a $20 million public offering (claim 14066) with a $5 million side purchase by existing holders including the CEO (claim 13976), and a $50 million Yorkville standby facility carrying a $15 million convertible advance (claims 13900, 13901). By mid-2022 the instruments were smaller and more structured. The record shows an additional $50 million term loan facility sweetened with lender warrants (claims 13873, 13848), a $9.5 million private placement (claim 13554), unit placements of $1.75 million to $2 million (claims 13534, 13535), and a revised Yorkville conversion price (claim 13463). The February 2023 offering priced at $0.68 per share for $25 million (claim 13404), which puts the dilution trajectory in plain view. Across this period, raise size shrank while per-dollar dilution rose.

The mix reads less as a planned capital stack than as sequential accommodation of the term lenders. Amortization relief was made conditional on a $33.5 million equity raise (claim 13210), and the maturity extension on a $40 million repayment by April 1, 2024 (claim 13199). Both came with penny and dollar warrants, which the company first stated at 26,666,669 and 13,333,333 (claims 13211, 13212) and later revised to 27,759,265 and 13,879,630 (claims 12150, 12151). Non-cash share issuance also adds to the count: 8.5 million shares to Somerset for a lease amendment (claim 13700), shares for services to the related party Beowulf E&D (claim 12155), and a share pool that increases automatically through fiscal 2026 (claim 12667). The record shows insiders buying or exchanging into the company, and it contains no insider sales.

On use of proceeds, the record has the stated commitments: the Talen joint venture contribution (claim 14138), Bitmain purchase agreements (claims 14136, 14137), and the Nautilus cash contribution (claim 13712). It does not have a deployment ledger that would let anyone reconcile those commitments against actual spending. The intended use of the Series A proceeds (claim 14277) is not specified in the data. Several figures are recorded without consistent units.

The open item is whether the $40 million repayment condition (claim 13199) was met, since many related claims remain marked delayed. Readers should also look for how much of the $200 million ATM was drawn and for warrant exercise activity once exercisability began (claim 12132). The record is truncated at January 2024 and has nothing after that date, so the most recent two and a half years cannot be assessed from these claims.

The record runs from March 2021 through mid-January 2024. The last entry is cut off, so anything after that point is outside this read. Within that window, dilution moves through two phases. Through 2022 the company raised against a growth plan. By early 2023 it was raising against its lenders. The Series A preferred (claim 14276), the $200M ATM (claim 13972) and the $20M April public offering with a $5M insider placement alongside it (claims 14066, 14048) sat next to large forward commitments: the $169M and $33M Bitmain agreements (claims 14137, 14136) and the $156M Talen joint venture commitment (claim 14138). By mid-2022 the funding mix widens. It adds the $50M Yorkville standby facility and $15M convertible advance (claims 13900, 13901), lender warrants attached to an additional term loan (claims 13873, 13848), and 8.5M shares issued to Somerset for a lease amendment (claim 13700). That breadth reads more as patching than as a planned capital structure. The Yorkville conversion price was later amended (claim 13463). The $25M offering at $0.68 per share (claim 13404) marks the low point in pricing.

From that point, the structure is shaped by the Fifth Amendment. Amortization relief was conditioned on $33.5M of equity proceeds (claim 13210). A maturity extension to December 2024 depended on repaying $40M by April 1, 2024 (claim 13199). Lender penny and dollar warrants were sized at roughly 26.7M and 13.3M (claims 13211, 13212) and later recorded at about 27.8M and 13.9M (claims 12159, 12151). Several of these items carry a delayed status, which means their final terms need explanation. Insiders show up as buyers and warrant counterparties, including the CEO-linked placement and the Prager-entity exchange (claims 13976, 13429). The record shows no insider sales.

The stated use of Series A proceeds (claim 14277) cannot be checked against deployment from this data. The Nautilus cash contribution of $17M (claim 13712) sits well below the Talen commitment. A working capital deficiency of $112M (claim 12850) suggests that deployment ran ahead of funding. Several figures appear in inconsistent units, so a sophisticated reader should confirm them against the filings. The items to watch are whether the $40M repayment condition was met, how much of the ATM reauthorized in November 2023 (claim 12129) was drawn, how the lender warrants were exercised, and the dilution from the auto-increasing incentive plan pool (claim 12667).

TeraWulf's record shows a company that funded its initial build-out through private capital and a single large term loan, then leaned progressively harder on dilutive and lender-driven instruments as conditions tightened. The early stack combined the March 2021 private placement (claim #14098), the $50M Series A preferred (claim #14099), the December 2021 private placements (claim #14101) and the $124M term loan (claim #14100). A $500M universal shelf (claim #14103) and a $200M ATM (claim #14102) sat behind it. From that base the financings became smaller and more frequent. The $20M April 2022 public offering (claim #14066) was followed within two months by a $50M Yorkville standby equity facility and a $15M convertible advance (claims #13900, #13901), whose conversion price was later amended (claim #13463). By October 2022 the company was raising in single-digit millions through unit placements (claims #13554, #13534, #13535). The trajectory through early 2023 is one of accelerating dilution at lower prices. The February 2023 offering priced at $0.68 per share (claim #13404) came against a stated working capital deficiency of $112M (claim #12850). The authorized share count was later raised to 600 million (claim #12077).

The mix reads less as a planned capital structure than as sequential accommodation of the term lenders. Amortization relief was conditioned on a $33.5M equity raise (claim #13210), and a maturity extension on a $40M repayment (claim #13213). Both came with penny and dollar warrants to lenders, later restated at 27,759,265 and 13,879,630 (claims #12150, #12160). Insider-affiliated entities participated repeatedly, including the Prager-entity warrant exchange (claim #13429) and the Allin WULF purchase (claim #13401). No open-market insider sales appear in the record provided.

Use of proceeds is thinly documented. Miner deposits (claim #13878) and the Nautilus contribution (claim #13712) are visible against a $156M joint venture commitment (claim #14138), but most 2023 equity appears directed toward satisfying lender conditions rather than stated expansion uses. That link requires explanation in later filings. The $22M repayment is stated (claim #12093), while the $30M repayment remains marked delayed (claim #12014).

A careful reader should watch whether the delayed repayment is completed, how much of the repeatedly renewed $200M ATM (claims #12255, #12129) has actually been drawn, and when lender warrants are exercised. The data supplied ends in May 2024, so later events are not assessed here.

TeraWulf's record reads as two distinct regimes. The first, running from April 2022 through early 2023, is survival financing. A $200,000,000 ATM with Cantor (claim 13972) sat alongside a $50,000,000 Yorkville standby facility and a $15,000,000 convertible advance (claims 13900, 13901), a term loan expansion carrying lender warrants (claims 13873-13876), and a string of small direct placements such as the $9,500,000 October 2022 private placement (claim 13554). Dilution came from many small instruments rather than one large one, and pricing deteriorated: the February 2023 offering cleared at $0.68 per share (claim 13404). That offering came as a working capital deficiency of $-112 was recorded (claim 12850; several figures in this record carry no stated scale). The mix in this period looks like patching rather than design. Equity raises were driven by lender conditions: amortization relief required roughly $33,500,000 of new equity by March 2023 (claim 13210). That relief came with 26,666,669 penny warrants and 13,333,333 dollar warrants (claims 13211, 13212), later restated upward to 27,759,265 and 13,879,630 (claims 12159, 12160), a revision that requires explanation. Insiders consistently took the other side as buyers. Examples include the CEO-linked $5,000,000 placement (claim 13976), the Prager-entity exchange (claim 13429), and Lake Harriet notes of $853,912 converting into $2,157,006 of stock (claims 13176, 13177). No insider sales appear in the record. On use of proceeds, the 2023 money tracks its stated purpose closely, since it flowed into lender compliance and then into the $22 debt repayment (claim 12093) and a further $30 repayment logged as delayed (claim 12014). The second regime begins in 2024. Authorized shares rose to 600,000,000 (claim 12013), and the Riesling Power lease buyout was paid partly in 20 million shares (claim 5583). The October 2024 convertible was upsized from $350 to $425 (claims 5532, 5502) and netted $415 (claim 5504). Of that, $115 was earmarked for repurchases (claim 5505) and $51 for capped calls (claim 5506). This is a coherent pivot: debt-funded buybacks and dilution hedging replaced lender-mandated equity. However, the record does not show where the remaining proceeds went. The record also ends in late October 2024, with its final entry cut off, so nearly two years of subsequent activity are not captured here. A careful reader should check several things next. First, how much of the $200 repurchase authorization (claim 5531) was actually executed. Second, whether the repeatedly renewed ATM (claims 12870, 12255, 12007) was drawn against the new share headroom. Third, how the $13 capped-call ceiling (claim 5499) compares with where the stock trades as the notes approach conversion.

TeraWulf's financing record falls into two distinct regimes, and they should be read separately. From mid-2022 through early 2024 the company financed itself in small increments, mostly with lender concessions. Lender warrants (claims #13848, #13438, #13437), unit placements with related entities such as Allin WULF and Bayshore (claims #13534, #13535, #13401), and a $25,000,000 offering priced at $0.68 per share (claim #13404) sat alongside repeated amendments to the term loan. Amortization relief was conditioned on raising $33,500,000 in equity (claim #13210), and a maturity extension depended on repaying $40 million (claim #13199). Penny and Dollar Warrants of $26,666,669 and $13,333,333 (claims #13211, #13212) were later restated at $27,759,265 and $13,879,630 (claims #12105, #12106). A stated working capital deficiency of $-112 (claim #12850) is the context for all of it. This period reads less as strategy than as keeping the lenders in place, and the dilution arrived through warrants and exchanges rather than through a single visible raise.

The second regime begins in October 2024. It centers on convertible notes: the $500 offering with a $8 conversion price and a $13 capped-call cap (claims #5492, #5495, #5499), then the $850 offering with $86 allocated to capped calls (claims #4447, #4452). These are paired with a $200,000,000 ATM that recurs across filings (claims #12272, #12007) and whose sales agreement is later marked withdrawn (claim #4992). The change toward larger instruments, with capped-call hedging of dilution, is coherent. The authorized-share increase to 600 million (claim #12009) put the capacity in place beforehand.

On use of proceeds, the record is thin. The 2024 raise earmarked $115 for share repurchases (claims #5500, #5505), but the repurchase authorization is marked delayed (claim #5531). Nothing in this record confirms how much was executed. The 2025 deployment language on data center expansion (claim #4445) is cut off in the data provided, so whether it tracks actual spending cannot be assessed here. No insider transactions appear in the record supplied, and this read makes no inference about them.

A careful reader should watch four items. The first is whether the delayed S-3 filing obligation tied to the Google warrants (claims #4503, #4510, #4459) is met. The second is the realized size of the repurchase against the stated $115. The third is whether the withdrawn ATM is replaced. The fourth is how two layers of convertible principal are refinanced or converted as the Cayuga and CBRE-linked commitments (claims #4499, #4512) come due.

TeraWulf's record reads as two distinct financing regimes, and the transition between them is the central fact. Through early 2023 the company funded itself in small, frequent increments. It used a $9.5 million private placement (claim #13554), $7.5 million term loan draws (claim #13555), unit placements of $1.75 million and $2.0 million with Allin WULF and Bayshore (claims #13534, #13535), and an amended Yorkville convertible (claim #13487). That period culminated in a stated working capital deficiency of $112 million (claim #12850) and a $25 million underwritten offering at $0.68 per share (claim #13404). The mix in this phase looks like patching rather than strategy. Equity raises were conditions of lender relief: amortization relief under the Fifth Amendment depended on a $33.5 million equity raise (claim #13210), and lenders received penny and dollar warrants (claims #12909, #12910). Several affiliated parties participated alongside outside capital, including the Stammtisch exchange (claim #13429) and the Lake Harriet notes (claim #13176). Many of these items carry a delayed status, and the $30 million repayment (claim #12014) and $40 million maturity contingency (claim #13213) are among those not shown as resolved. From late 2024 the company shifted to large convertibles paired with capped calls. The 2024 note grew from $350 million (claim #5532) to $425 million (claim #5502) to $500 million (claim #5492), with a $60 million capped call (claim #5498) and a $115 million planned repurchase (claim #5500). In August 2025 it raised $850 million (claim #4441) with an $86 million capped call (claim #4444). This is a more coherent capital structure: dilution is deferred and partly hedged rather than incurred immediately. On use of proceeds, the record states purposes but rarely shows deployment. No completed repurchase figure appears against the $115 million, and the broader $200 million authorization is marked delayed (claim #5531). The data center expansion attributed to the 2025 proceeds (claim #4445) has no matching spend on record beyond the $98 million Cayuga lease prepayment (claim #4499). The Google warrant is recorded at 41,011,803 (claim #4510) and 32,568,197 (claim #4459), a difference that requires explanation. A reader should watch several items next: whether the delayed S-3 registration tied to that warrant (claim #4503) is filed, the withdrawn 2025 ATM (claim #4992), repurchase execution, and the contingent CBRE commission (claim #4512). No insider sale transactions appear in the supplied data, and the record is truncated after claim #4434, so later events are not assessed here.

TeraWulf's record divides into two distinct regimes, and the transition between them matters more than any single transaction. Through late 2022 and early 2023 the company financed itself in small, overlapping pieces. It raised money through direct unit placements of $1.75M and $2M (claims #13534, #13535) and a Yorkville advance whose conversion price and maturity were both amended (claims #13463, #13464). It also issued penny and dollar warrants to lenders (claims #13211, #13212) in exchange for amortization relief. That relief depended on raising $33.5M of equity (claim #13210) and repaying $40M to extend maturity (claim #13213). Against a disclosed working capital deficiency of $112M (claim #12850), the $25M underwritten offering at $0.68 per share (claim #13404) reads as lender-driven patching rather than chosen strategy. Dilution in this period was high relative to dollars raised, because warrants and sub-dollar equity did most of the work. The $200M ATM (claims #12870, #12255, #12007) and the increase in authorized shares to 600M (claim #12009) extended that equity-first posture into 2024, alongside debt paydowns of $22M and $30M (claims #12093, #12014). From October 2024 the mix changed to large convertibles with capped calls. The $500M issue carried a conversion price near $8 and a $13 cap (claims #5492, #5495, #5499). It was followed by the $1B 1% notes, with net proceeds of $975M and $101M spent on capped calls (claims #4433, #4434, #4435). Withdrawing the $200M ATM (claim #4992) and pairing the first convertible with a $115M repurchase (claim #5500) suggests a deliberate shift from spot dilution toward deferred, hedged dilution. On use of proceeds, the 2025 notes were earmarked for data center expansion (claim #4445), and the record shows related outflows such as the $98M Cayuga lease prepayment (claim #4499) and the $30M CBRE commission (claim #4511). The record does not, however, show deployment in detail, and whether the $115M repurchase was executed is not confirmed. The broader $200M repurchase authorization is marked delayed (claim #5531). Readers should watch the September 2025 charter amendment on authorized shares (claim #4414), whose entry is truncated here. They should also watch the Google warrants (claims #4510, #4459), the delayed S-3 registration (claim #4503), and the $20M contingent CBRE commission (claim #4512). The record contains no insider transactions and nothing after September 2025. Some amounts are recorded in inconsistent units, so conclusions about the most recent year require explanation.

TeraWulf's financing record falls into three distinct phases, and each one changes both the scale and the mix. In 2023 the company was managing constraint. It disclosed a working capital deficiency (claim #12850), negotiated amortization relief through the Fifth Amendment (claim #12851), and paid lenders in equity-linked form, with penny warrants of 27,759,265 and dollar warrants of 13,879,630 (claims #12150 and #12160). A $200M ATM ran alongside (claim #12255). That period reads as patching rather than strategy. The 2024 debt repayments of $22M and $30M (claims #12093 and #12014) closed it out, along with the increase in authorized shares to 600 million (claim #12077).

From October 2024 the mix becomes more deliberate. The $500M convertible (claim #5492) was paired with a $115M share repurchase (claim #5500) and capped calls costing $60M (claim #5498), a structure meant to soften dilution at issuance. The 2025 converts followed the same template, including the $1,000M 1% notes with $101M in capped calls (claims #4433 and #4435). The $3.2B secured notes at WULF Compute (claim #4187) then moved leverage down to the project level. Dilution was not removed, only relocated. The Google warrants (claims #4510 and #4459) now sit as a liability reported at $844,698, apparently in thousands (claim #11683). The charter increase to 950 million authorized shares is marked delayed (claim #4414). The $800M common stock offering (claim #2101) sits uneasily beside the buyback logic of 2024.

On use of proceeds, the record does not show whether announced uses were carried out. The only stated purpose for the August 2025 raise is general data center expansion (claim #4445). No claim on record confirms that the $115M repurchase was executed. A separate $200M repurchase authorization is marked delayed (claim #5531). These gaps require explanation rather than any inference.

Several items are worth watching next. The Lake Hawkeye financing has been withdrawn while the parallel Flash Compute and Chesapeake financings remain stated (claims #11743, #11742 and #11744), which leaves the reason for the difference open. The $500M Hawesville bridge (claim #2239) will need take-out financing. The $100M completion guarantee and $75M letter-of-credit collateral (claims #3955 and #3953) tie parent-level cash to project debt. The data contain no insider transactions, so this read offers no view on insider behavior, and the record appears truncated after claim #2103.

TeraWulf's financing record divides into two distinct phases. Through 2023 and early 2024 the company relied on a renewed $200M at-the-market program (claims 12272 and 12007), penny and dollar warrants (claims 12150 and 12160), and a debt amendment suspending fixed amortization (claim 12252). That is balance-sheet patching: buying time on the legacy loan while paying it down in increments of $22M and $30M (claims 12093 and 12014). Beginning with the $500M convertible in October 2024 (claim 5492), the pattern becomes a deliberate capital-stack build. The $1B 1% convertible (claim 4433) was followed by the $3.2B 7.75% senior secured notes at WULF Compute (claim 4187), a further $1.025B of convertibles (claim 1057), and $1.3B of secured notes (claim 3951). The move from parent-level equity-linked paper to project-level secured debt, backed by Google credit support on the Fluidstack leases (claims 3636 and 3647), reads as a coherent strategy rather than opportunism. Dilution has been deferred rather than avoided. Authorized shares rose to 600M (claim 12009), and a further increase to 950M is on record as delayed (claim 4414). Series A preferred was mandatorily converted (claim 4038). The April 2026 common offering was the first large straight equity raise. Its reported size moves across $800M, $900M, $1,004M and $1,170.78M in successive claims (claims 2101, 2081, 2064 and 10301), and that sequence requires reconciliation. Google warrant liability of $844,698 thousand (claim 11683) adds further overhang. Use of proceeds is only partly traceable. The $115M share repurchase tied to the 2024 convertible (claim 5500) sits under a $200M authorization marked delayed (claim 5531), and no later claim confirms execution. The April 2026 equity was earmarked for repaying the $500M Hawesville bridge (claims 2103 and 2239), but the repayment claims state intent rather than completion. The data contains no insider transaction records, so that dimension cannot be assessed, and the final claim (5374) is truncated. A careful reader should watch four items. The first is whether the $1,479M planned financings for Flash Compute and Chesapeake Data (claims 11742 and 11744) close, and why Lake Hawkeye's was withdrawn (claim 11743). The second is the status of the 950M authorization and the delayed S-3 filing (claim 4503). The third is confirmation that the bridge was retired. The fourth is how the $450M Abernathy joint venture sale proceeds (claim 8913) are deployed against a secured debt load of $3,052,240 thousand (claim 11682).

TeraWulf's record shows a company that has moved in roughly three years from small-balance-sheet patching to project-scale capital formation. The 2023 entries point to constraint. The $200M ATM (claim #12254), the suspension of fixed principal amortization (claims #12252, #12122), and the penny and dollar warrants (claims #12150, #12160) read as measures to preserve liquidity rather than to fund a defined build. The shift arrives with the October 2024 convertible. It was sized at $350M (claim #5532) and closed at $500M (claim #5492), paired with a capped call costing $60M (claim #5498) and a $115M planned repurchase (claim #5500). That structure tries to raise capital while visibly managing dilution. From there the instruments escalate. The $1,000M of 1% convertibles (claim #4433) came alongside Google warrants (claim #4510). The $3.2B of 7.75% secured notes at WULF Compute (claim #4187) were followed by a $1.025B convertible (claim #1057) and the $1.3B secured notes (claim #3951), which carried a $100M completion guarantee (claim #3955). Taken together, the mix looks coherent. Secured debt sits at the project level against Google-backstopped leases (claims #3648, #3653), while holding-company converts and, from April 2026, straight equity fund the equity layer. Dilution has been deferred rather than avoided. Authorized shares rose to 600 million (claim #12009) and a further increase to 950 million is recorded as delayed (claim #4414). The first large common offering, $800M (claim #2083), appears in the record at figures up to $1,170,780,000 (claim #10301), and those figures require reconciliation. The record also diverges on convertible net proceeds, with $878M (claim #4173) against $999.7M (claim #4168). On use of proceeds, the stated purposes include data center expansion (claim #4445) and repayment of the $500M Hawesville bridge (claim #10320). Nothing on record confirms deployment, confirms that the bridge was retired, or shows whether the 2024 repurchase was executed; the related authorization is marked delayed (claim #5531). No insider transactions are on record, so that dimension cannot be assessed. A sophisticated reader should watch four things. The first is how the $3.5B Anthropic-related raise (claim #5374) is structured, and whether it is secured at the project level or reaches the parent. The second is the status of the $1,479M planned financings, now that the Lake Hawkeye tranche is withdrawn (claim #11743). The third is the delayed S-3 filing (claim #4503). The fourth is whether the $450M Abernathy joint venture sale (claim #8913) signals asset monetization becoming a recurring funding source.

TeraWulf's record describes a company that moved from small patching to capital raising at infrastructure scale in about eighteen months. In 2024 it relied on warrants (claims 12105, 12106), a $200M ATM (claim 12007), a share-based lease termination (claim 5583) and an increase in authorized shares to 600 million (claim 12009). From October 2024 onward, convertibles became the main instrument, and each one was upsized during marketing: $350 to $500 (claims 5532, 5492), $400 to $1,000 (claims 4454, 4433), and $500 to $1,025 (claims 4180, 4167). Every one was paired with a capped call (claims 5498, 4435), which suggests a deliberate effort to limit conversion dilution rather than improvisation. Late 2025 added a second layer of asset-level secured debt: the $3.2B WULF Compute notes (claim 4187) and the $1.3B notes (claim 3951), backed by completion guarantees and cash collateral (claims 3955, 3953). April 2026 then returned to straight equity. That offering is recorded at $800M (claim 2101), with later figures of $900M and $1.17B (claims 2081, 10301), several of them marked delayed. The final size requires reconciliation. The proposed increase in authorized shares to 950 million (claim 4414) is also marked delayed, and the ATM sales agreement was withdrawn (claim 4992). Taken together, the mix reads as a coherent project-finance structure, with equity and convertibles at the parent and secured debt at the project entities, even though individual tranches were sized to meet demand. On use of proceeds, stated purposes were specific: a $115 repurchase alongside the 2024 notes (claim 5500), data center expansion (claim 4445), and Hawesville plus repayment of the $500M bridge (claims 2239, 2103). This record contains no follow-up confirming that the repurchase was executed or the bridge retired, and the related repurchase authorization is marked delayed (claim 5531). Deployment therefore cannot be verified from these claims. The record also contains no insider transactions, so that dimension cannot be assessed. Several items deserve attention next. The three planned $1,479 project financings (claims 11742 to 11744) and the $3.5B Anthropic-linked raise (claim 5374) have no disclosed terms. The Abernathy stake sale appears at $450M and later $530 (claims 8913, 11408), which needs explanation. Google warrant liability is recorded at $844,698 (claim 11683). All of this sits against Q2 2026 revenue of $45 with a reported year-over-year decline (claims 2804, 2805). The record as supplied is truncated after claim 748.

TeraWulf's capital structure over the past 20 months reveals an escalating and layered financing strategy rather than a coherent plan. Between October 2024 and August 2026, the company raised approximately $10 billion through overlapping instruments: convertible notes ($488 million due 2030, per claim 4991; $1 billion at 1% due 2031–2032, per claim 4433; and tranches totaling $2 billion through late 2024), senior secured debt ($3.2 billion at 7.75% due 2030, per claim 4187; and $1.3 billion, per claim 3951), a $500 million bridge facility (claim 2084), and equity ($800–900 million public offering in April 2026, per claims 2083 and 2101; $1.8 billion Google investment in August 2026, per claim 7811). The progression from debt-first (Oct 2024–Dec 2024) to equity-dependent (Apr–Aug 2026) suggests capital capacity tightened. Notably, a $200 million ATM offering was withdrawn in May 2025, and a Charter Amendment to increase authorized shares was delayed in September 2025, both indicating market or internal constraints.

Proceeds tracking tells a different story. Management stated deployment for data center expansion (claim 4445) and Hawesville construction (claim 2103), yet Q2 2026 results show revenue of $45 million against a net loss of $940 million (claim 769) and negative working capital of $957 million (claim 760). The company monetized its Abernathy joint venture stake for $450–530 million in August 2026, precisely when announcing a major $19 billion lease agreement with Google (claim 8306). This suggests neither organic cash generation nor deployed capital is yielding expected returns. The debt-to-equity ratio sits at 27x (claim 742), an extreme leverage profile for a company burning cash operationally. Google's involvement deepened structurally: beyond equity and the lease, Google provided credit support for multiple facility leases (claims 3646–3657) and TeraWulf issued warrants totaling $844.7 million in liability (claim 11683).

The maturity wall arrives in 2030 across convertibles and senior secured notes ($3.5+ billion due). A sophisticated reader should track whether the $19 billion Google lease contract delivers cash before maturity; whether additional Google capital infusions become necessary; and why operational losses persist despite $10 billion in capital raised. The bridge facility repayment and asset sale suggest liquidity stress occurred mid-2026, contradicting any narrative of smooth deployment. Watch for revised guidance, delayed data center completions, or further dilutive financings—each would signal execution risk rather than strategy.

TeraWulf's capital narrative from October 2024 through August 2026 reveals a company shifting instruments in response to market access rather than executing a unified strategy. The company opened with convertible notes, issuing approximately $425 million stated in October 2024 (per claim 5532, net proceeds $487 million after $60 million in capped call costs per claim 5498) with a $3 interest rate (claim 5493) and $8 conversion price (claim 5495). This approach recurred in May 2025 ($488 million in 2030 notes per claim 4991) and again in August 2025 ($850 million offering per claim 4441, net proceeds $829–975 million per claims 4443 and 4434). The conversion profile shifted materially: later convertibles (2031–2032) reached $1.58 billion outstanding by March 2026 (claim 11681), representing substantial future dilution assuming equity appreciation.

The pivot to secured debt appears opportunistic. Starting October 2025, the company accessed senior secured markets aggressively—$3.2 billion at 7.75% due 2030 (claim 4187), followed immediately by another $1.025 billion convertible (claim 1057), then $1.3 billion in senior secured notes by December 2025 (claim 3951). This debt load—$3.05 billion 2030 secured notes outstanding by March 2026 (claim 11682)—now dominates the liability structure. A common equity offering of $1.17 billion closed in April 2026 (claim 10301), suggesting debt serviceability concerns prompted equity raises.

Stated use of proceeds centered on data center expansion at multiple sites (Hawesville per claim 2103, Somerset, and three subsidiary vehicles per claims 11741–11744). Yet actual deployment appears misaligned. Q2 2026 results showed revenue of only $45 million against a $940 million net loss and negative $1 million free cash flow (claims 2803, 756, 759), indicating either massive pre-revenue buildout or operational challenges at scale. The company also monetized the Abernathy joint venture stake for $530 million (claim 11408), an asset sale inconsistent with internally robust capital deployment.

Three developments require scrutiny. First, the Google relationship—$1.8 billion equity investment in August 2026 (claim 7811) plus $3.5 billion Anthropic data center financing (claim 5374)—signals TeraWulf shifted from credit market access to customer-backed capital structures. Second, the cash balance collapsed to approximately $3 million by August despite $6+ billion raised (claim 757), requiring explanation of cash timing and covenant pressure. Third, the 2030 maturity wall ($3.05 billion secured debt) arrives without demonstrated revenue scale, creating material refinance risk.

TeraWulf's capital strategy through mid-2026 reflects aggressive expansion financing outpacing revenue generation, with a layered, opportunistic structure rather than a coherent sequencing. The company staged multiple convertible offerings beginning October 2024, ranging from $350M to $1B tranches through October 2025 (per claims 5532, 5492, 4454, 4447, 4441, 4433, 4180, 4167), each marginally larger than the prior, suggesting repeated market testing and investor appetite assessment rather than premeditated sizing. Convertibles dominated the first year, but the company subsequently layered senior secured debt ($3.2B in October 2025, per claim 4215; $1.3B in December 2025, per claim 3951), signaling a shift toward fixed obligations once convertible appetites saturated. The April 2026 equity offering ($1.17B raised, per claim 10301) and July 2026 asset sale ($450M-$530M from Abernathy interest, per claims 8913, 11408) suggest capital had tightened; announcing equity after secured debt refinancing typically indicates debt markets had begun pricing in execution risk.

The announced use—data center capacity for AI workloads, with explicit Anthropic-leased facilities receiving $3.5B financing in July 2026 (claims 5374, 2)—aligned with capital deployment through lease prepayments and facility commitments. However, Q2 2026 results (reported August 5, claims 1401, 2804) showed only $45M revenue against a net loss of $940M and negative free cash flow. This 20:1 loss-to-revenue ratio, even accounting for pre-revenue capex accounting, suggests either project delays, lower-than-expected utilization, or contracted terms disadvantageous to WULF. The company's cash balance had contracted to $3M by Q2 2026 (claims 757, 771), despite raising roughly $6B in net capital from debt and equity offerings over eighteen months—indicating substantial ongoing burn or capex commitment drawdowns.

A sophisticated investor should distinguish between committed capex and discretionary cash reserves. The Google warrant liability of $844.7M (claim 11683) represents potential equity dilution at future valuations. Delayed announcements on the charter amendment increase (claim 4414) and preferred stock conversion (claims 4038, 4039) warrant clarification on authorization constraints. The reliance on asset monetization after Q2 losses and the timing of the $3.5B Anthropic financing suggest the company exhausted traditional refinancing channels and required strategic capital partnership to sustain burn.

TeraWulf's capital narrative over eighteen months reveals a sequential rather than coordinated strategy: the company cycled through convertible notes, senior debt, equity, and strategic partnerships as capital markets conditions and internal needs dictated. The convertible issuances began modestly in October 2024 with a $350M offering (claim 5532) and expanded through August 2025 into two major tranches—$850M (claim 4441) and $1B at 1% coupon (claim 4433)—suggesting the company seized windows of investor appetite for equity optionality. Each convertible was paired with capped call transactions totaling $51-101M (claims 5506, 5498, 4444, 4435), a textbook dilution hedge that signals management anticipated conversion risk.

The funding mix shifted decisively in October 2025 toward fixed-rate debt: $3.2B in senior secured notes (claim 4187) followed by $1.3B in additional senior notes in December (claim 3951). This pivot from optionality to certainty likely reflects rising equity issuance costs or covenant constraints. Convertibles continued in October with a $1.025B offering (claim 1057) netting $999.7M (claim 4168), but debt dominance suggests equity dilution had become prohibitive. An aborted ATM offering in May 2025 (claim 4992) preceded the shift to debt, indicating equity markets were unreceptive.

The company's stated use of proceeds—data center expansion, Hawesville construction, and infrastructure financing—tracks actual deployment evident from Q2 2026 results. Yet capital deployment reflects deteriorating unit economics: despite raising approximately $3-4B through April 2026, the company reported $45M quarterly revenue (claim 2804) against $940M net loss and negative $1B free cash flow (claims 756, 759). The subsequent $3.5B Anthropic lease financing (claim 5374) and $1.8B Google investment (claim 7811) in July-August 2026 are revenue-backed partnerships rather than traditional capital raises, signaling exhaustion of conventional markets.

Sophisticated readers should monitor covenant compliance and refinancing capacity. Working capital deteriorated to negative $957M by Q2 2026 (claim 774), and the Abernathy joint venture stake sale for $450M (claim 8913) in July indicates liquidity pressure. The company raised capital opportunistically—each tranche executed when windows permitted—rather than executing a premeditated plan. Debt-to-equity ratio stands at 27x (claim 742), and cash burn exceeds capital inflow velocity. Positive unit economics remain undemonstrated.

TeraWulf's capital trajectory from late 2024 through August 2026 reveals a company navigating an increasingly constrained funding environment against aggressive data center buildout. The financing mix escalates from convertible notes in Q4 2024 (totaling approximately $500M announced, per claim 5492) through a dramatic shift toward senior secured debt in late 2025: $3.2B in October (claim 4215) and $1.3B in December (claim 3951), both structured around the Hawesville completion guarantee (claim 4189). This progression—convertibles to senior secured to equity—suggests not strategic design but successive recourse to deeper capital pools as prior tranches tighten or approach maturity.

The convertible positioning attempted to defer dilution through conversion prices around $8 (claim 5495), but by mid-2025 the company abandoned this approach entirely. The $200M ATM offering (claim 4992) was withdrawn, and instead WULF issued $488M convertibles (claim 4991) followed by a $1.025B convertible close in October (claim 1057). The company then proceeded to $1.17B in direct equity in April 2026 (claim 10301), framed around repaying the $500M Hawesville bridge facility (claim 2103). The stated use of convertible proceeds tracked deployment—data center expansion (claim 4445)—but the escalation from convertibles into $3.2B senior secured into equity within eighteen months indicates the capital structure ladder was not pre-planned but traversed under pressure.

The Google relationship encapsulates the coherence question. Beginning with warrant grants totaling approximately $73.6M in August 2025 (claims 4510 and 4459), it escalates to credit support for lease arrangements and culminates in an $1.8B equity investment by August 2026 (claim 7811). This is less diversified financing and more concentrated dependency: Google is simultaneously tenant, equity holder, and credit supporter. The near-simultaneous monetization of the Abernathy joint venture stake for $450–$530M (claims 8913 and 11408) within one quarter of announcing $3.5B Anthropic data center financing (claim 2) invites scrutiny regarding portfolio composition and cash velocity.

By Q2 2026, the balance sheet reflected operational leverage reversed: $45M revenue (claim 2804) against a 27x debt-to-equity multiple (claim 742), negative working capital of $957M (claim 760), and free cash flow of -$1M (claim 775). Revenue declined year-over-year (claim 2805). The company has secured substantial capital for build but requires significant revenue scale to service it. Watch whether the $19B contracted revenue lease announced August 2026 (claim 8306) represents durable Google commitment or embedded optionality that could shift if utilization diverges from underwriting.

The pattern shows a company executing a deliberate capital structure transformation across eighteen months. From October 2024 through mid-2026, TeraWulf cycled through convertible notes ($488M due 2030 per claim 4991), then pivoted decisively to $4.5B in senior secured debt ($3.2B in October 2025 per claim 4215, $1.3B in December 2025 per claim 3951), attracting Google warrants ($41M per claim 4510, $32.5M per claim 4459) and eventually a $3.5B financing commitment for an Anthropic data center (claim 5374). The April 2026 equity offering ($800M per claim 2101, expanding to $1.17B via greenshoe per claim 10301) arrived only after the debt base was established, suggesting capital market access was tied to secured assets and strategic partnerships rather than equity strength. This is coherent strategy, not opportunistic patching.

The sequencing reflects deliberate positioning. Convertible notes early provided equity optionality for an operator without sustained revenue; secured debt followed, collateralized by operating lease commitments; equity issuance only after Google's $19B contracted lease commitment (claim 8306) de-risked the asset base. The use of proceeds tracks the thesis: lease terminations and establishment costs in 2024, data center construction throughout 2025, the April 2026 bridge facility ($500M per claim 2102) explicitly for Hawesville completion, culminating in the Anthropic deployment in July 2026. No misalignment between stated purposes and capital sequence.

However, Q2 2026 financial results require immediate scrutiny and explain the capital intensity. Net income of -$940M, revenue of only $45M, and debt-to-equity of 27x (claims 769, 765, 742) reveal that contracted leases have not yet matured into material revenue. Negative working capital of -$957M (claim 760) confirms cash constraints during deployment. The Anthropic facility itself funded only in July 2026, meaning the largest revenue-generating asset remains pre-deployment as of the latest reported quarter. The gap between committed lease revenue and recognized revenue is the material fact.

A sophisticated reader should watch three things: whether Google lease revenues appear meaningfully in Q3-Q4 2026 results; whether debt service requirements force additional capital raises before Anthropic and Google leases generate sufficient cash; and whether the company can sustain operations at the announced utilization rates without demand shocks. The Google partnership is structural, not merely financial—credit support for multiple leases signals strategic co-investment in WULF's infrastructure. If that partnership cools, refinancing becomes acutely difficult at current leverage.

TeraWulf has pursued a distinctly debt-heavy capital strategy since late 2024, moving from hybrid instruments to increasingly senior-secured leverage. Early convertible offerings in October 2024—including $500M in convertible notes per claim 5492 and associated capped-call transactions of $60M per claim 5498—were hedged instruments, suggesting confidence in equity appreciation and intent to minimize near-term dilution. However, this strategy shifted decisively in October 2025 with the $3.2B senior secured notes per claim 4215, followed by an additional $1.3B senior secured tranche in December per claim 3968. Over the same interval, the company issued nearly $3.4B in convertible notes (including the October $1.025B closing per claim 1057), suggesting a mixed approach once equity markets tightened. The April 2026 common stock offering of $800M per claim 2101, subsequently increased to $1.17B per claim 10301, indicates a return to equity markets after a period of debt reliance.

Stated use of proceeds tracks consistently to data center infrastructure. The October 2025 senior secured notes included a completion guarantee for data center expansion per claim 4189. The $3.5B Anthropic financing per claim 5374 was earmarked specifically for an Anthropic-leased facility. Google credit support facilities announced in February 2026 per claims 3636, 3648, 3647 structured around specific lease commitments. This suggests disciplined capital allocation aligned to revenue generation, though the financing mix evolved opportunistically rather than according to a preset strategy.

Financial metrics as of Q2 2026 expose substantial execution risk. Net income was -$940M per claim 769, with reported EPS of -$2 per claim 1400, against revenue of only $45M per claim 2804. Yet the company disclosed contracted revenues of $19B per claim 8306—a gap indicating revenue is contractually committed but not yet recognized. Working capital stands at -$957M per claim 774, and debt-to-equity reached 27x per claim 742, meaning the company is funding infrastructure deployment with borrowed capital while betting entirely on future revenue realization. The Google $1.8B equity investment per claim 7811 and Anthropic partnership suggest institutional validation, but also raise execution stakes. A sophisticated reader should monitor whether subsequent disclosures show revenue tracking toward the $19B contracted base or whether capacity constraints, deployment delays, or customer softening materialize. The capital structure leaves minimal margin for error.

TeraWulf raised capital across five sequential phases from October 2024 through July 2026, with instrument selection reflecting narrowing credit access. The first phase consisted of $350 million and $500 million convertible offerings in October 2024 (claims 5532, 5492) with $487 million in net proceeds stated (claim 5497). A parallel $12 million lease termination payment to Riesling Power (claim 5584) and 20 million share issuance (claim 5583) addressed operational needs. Authorization of a $200 million ATM offering in May 2025 (claim 4992) was withdrawn, signaling either market dislocation or preference for alternative structures.

The August 2025 phase introduced overlapping convertible offerings with conflicting disclosure: a $400 million convertible stated (claim 4454), an $850 million offering initially delayed (claim 4447), and a $1 billion convertible at 1% coupon (claim 4433). Net proceeds were reported as $829 million to $1 billion (claims 4443, 4434), reflecting amendments or filing lag. October 2025 marked the shift to senior secured debt—$3.2 billion priced (claim 1058)—accompanied by a $1.025 billion convertible (claim 1057). December added $1.3 billion in secured notes (claim 3951). This progression from convertibles to senior debt to common equity (April 2026: $800 million to $1.17 billion, claims 2083, 10301) indicates tightening credit markets or internal capital constraints forcing costlier instruments.

Disclosed use of proceeds tracks data center expansion at Hawesville and an Anthropic-leased facility requiring $3.5 billion (claim 2). Google provided $1.8 billion in direct equity investment (claim 7811), $73 million in warrants (claims 4510, 4459), and credit support across multiple lease agreements. A $450 million joint venture stake sale in July (claim 8913) suggests earlier capital pressure or valuation timing.

Yet Q2 2026 results expose acute misalignment: $45 million in revenue (claim 765) against $940 million in net loss (claim 756), with free cash flow at negative $1 million (claim 759) and cash on hand at $3 million (claim 771). A 27-to-1 debt-to-equity ratio (claim 742) and negative $957 million working capital (claim 760) eliminate margin for execution risk. The 2030 maturity wall on $2.8 billion in convertibles and $4.5 billion in senior notes requires refinancing or asset sales in thirty-six months. Google's concentration across equity stakes, warrants, and the $19 billion contracted revenue lease (claim 8306) creates both validation and dependency—any material shift in Google's AI data center strategy would pressure WULF's capital structure and deployment thesis.

TeraWulf's financing arc divides into three distinct phases, each revealing progressively constrained capital conditions. Through mid-2025, the company issued convertible notes in a measured cadence—$488M in May, followed by $850M and $1B tranches in August, each paired with capped call hedges—establishing a pattern of equity-light capital raising. This shifted abruptly in October 2025 when TeraWulf accessed the senior secured market, issuing $4.5B in two tranches ($3.2B in October, $1.3B in December, per claims 4187 and 3951), signaling major capex deployment for data center buildout across Hawesville, Abernathy, Akela, and Cayuga facilities. The announced use of proceeds explicitly linked these borrowings to construction, supported by completion guarantees and collateral reservations totaling $175M by year-end 2025 (claims 4189, 3955, 3953). Yet deployment has substantially outpaced operating cash generation. By Q2 2026, TeraWulf reported $45M revenue against a $940M net loss and negative $1B free cash flow (claims 1401, 756, 759), while carrying debt-to-equity of 27x (claim 742)—an extreme leverage position for a nascent operator. This mismatch forced a strategic pivot: an April 2026 public equity offering of $1.17B (claim 10301) to service maturing obligations, followed in July by asset monetization of its Abernathy joint venture stake for $450M (claim 8913) and a $3.5B financing facility from Google (claim 5374) paired with a $1.8B Google equity investment (claim 7811). The Google facilities include substantial credit support for multiple leases (claims 3648, 3657, 3632), revealing an operational dependency that extends beyond customer concentration into credit provision. What distinguishes this pattern from disciplined leverage is the absence of revenue trajectory justifying the debt stack: $19B in contracted lease revenue announced in August (claim 8306) remains forward-looking against current negative operating cash flow. The mandatory preferred stock conversion (claim 4038) adds further dilution pressure. A sophisticated reader should monitor two critical metrics going forward: whether the company achieves positive operating cash flow within twelve months to justify its $8B-plus total liabilities, and whether Google's growing financial entanglement—credit guarantor, primary customer, majority equity investor—becomes a structural constraint on management autonomy or signals a potential restructuring catalyst.

TeraWulf raised approximately $12 billion in gross capital over eighteen months via accelerating capital layers. The October 2024 settlement with Riesling Power involved $12M cash and 20M shares (per claims 5583, 5584), followed by an initial convertible offering announced at $350M and revised upward to $425M-500M (per claims 5532, 5502, 5492), yielding $415M-487M in net proceeds (claims 5504, 5497). A tactical $115M share repurchase accompanied the offering (claim 5505). The pace then accelerated: August 2025 saw convertible issuances of $400M-1,000M (per claims 4454, 4441, 4433) with reported net proceeds of $829M (claim 4443), followed by a major senior secured debt placement of $3.2B at 7.75% due 2030 (per claim 4187) in October and a further $1.3B senior secured tranche in December (claim 3951). April 2026 introduced an $800M-900M common equity offering (per claims 2101, 2083, with greenshoe per claim 2082), $500M bridge financing (claim 2067), and $250M revolving credit (claim 9297). Strategic partnerships included Google warrants (claims 4510, 4459) and a $1.8B equity co-investment (claim 7811); the company also monetized an Abernathy joint venture stake for $450M (claim 8913).

The capital structure strategy is opportunistic layering: convertibles when equity was undervalued, senior secured debt once credit credibility improved, equity and credit facilities for operating flexibility, and Google's partnership for balance-sheet support. Announced uses align operationally—data center construction and lease commitments across Somerset, Cayuga, Akela, Hawesville, and an Anthropic facility, including a $3.5B financing announced in July 2026 (claim 5374).

Yet Q2 2026 results create material tension. The company reported $45M in revenue (claim 765), a net loss of $940M (claim 756), only $3M in cash (claim 757), negative working capital of -$957M (claim 774), and debt-to-equity of 27x (claim 742). This is a pre-revenue infrastructure vehicle burning cash into multi-year capex. The critical watch is whether the announced $19B lease agreement (claim 8306) generates meaningful utilization before the senior secured notes mature in 2030. Any capex overrun or lease ramp delay materially pressures refinancing, particularly if cloud capex cycles soften. Google's sustained co-investment signals conviction in the model but creates concentration exposure should market conditions shift.

TeraWulf's financing pattern through mid-2026 reveals a coherent but financially strained capital strategy: build dedicated AI/GPU data center capacity at scale to secure long-term hyperscaler leases, funded through layered and increasingly aggressive dilution. The timeline shows this progression clearly. Early phases (Oct 2024–May 2025) relied on convertible notes with capped calls—traditional equity-hedging tools suggesting management viewed dilution as a calibrated cost of growth. A failed ATM attempt in May 2025 marked the pivot. By October 2025, the company raised $3.2B in senior secured notes (claim 4187), followed by another $1.3B tranche in December (claim 3951), signaling debt markets' appetite for infrastructure-backed capital. Concurrent convertible issuances—$850M in August (claim 4441) and $1.025B closed in October (claim 1057)—layered junior capital atop senior obligations in a structure typical of levered growth companies.

The strategic inflection arrived with Google warrants totaling roughly $74M (claims 4510, 4459) and a subsequent $1.8B Google equity investment (claim 7811). This partnership reframes the dilution narrative: not distress financing but contract-backed expansion. The $19B lease agreement (claim 8306) reads as external validation of the thesis.

Execution reveals material strain. Q2 2026 produced only $45M revenue (claim 765) against -$2 EPS (claim 764) and -$940M net income (claim 756). Free cash flow stands negative (claim 759), and working capital deteriorated to -$957M (claim 760). Meanwhile, capital deployment continues: a $3.5B Anthropic data center facility (claim 2) and Justified Data Campus investment estimates of $4–4.5B (claim 6724). The company sold Abernathy JV stakes for $450M (claim 8913)—asset monetization typically signals liquidity pressure beneath announcements. Debt-to-equity expanded to 27x (claim 742), assuming rapid revenue acceleration.

Sophisticated readers should monitor whether the $19B lease translates to material quarterly bookings within 2–3 quarters. If revenue growth lags leverage expansion, additional dilution becomes nearly certain despite recent capital raises. The Google equity stake functions as confidence signal but also as contingent call option on the company's solvency structure. Delayed claims around charter amendments and offerings suggest market friction. The trajectory now demands execution rather than capital flexibility.

TeraWulf's capital pattern from October 2024 through August 2026 reflects a coherent buildout strategy rather than opportunistic funding. The company moved through three distinct phases: early convertible notes (launching with $350M in October 2024, per claim 5532, escalating through convertibles worth $500M, $850M, and $1B in late 2024–2025), then a strategic pivot to secured debt once customer relationships solidified ($3.2B senior secured notes in October 2025, per claim 4215, followed by $1.3B in December, per claim 3951), and finally mixed equity and strategic capital ($800M public offering in April 2026, per claim 2101, and a $1.8B Google equity investment in August 2026, per claim 7811). The convertible overhang was substantial—multiple offerings closed within weeks (October 2024 saw $425M, $500M, and $850M tranches), suggesting either aggressive demand or pre-arranged syndication. Dilution accelerated markedly: 20 million Riesling shares issued October 2024 (claim 5583), followed by warrant issuances to Google totaling $73.6M (claims 4510, 4459), and public equity raising that introduced material new share count in Q2 2026. The strategy's coherence becomes apparent when tracking deployment: convertibles funded early capacity (Hawesville, per claim 2102 bridge facility; Akela, per claim 3632 Google credit support); secured debt arrived only after Google committed to both credit support (February 2026, claims 3646–3657) and a $19B lease agreement (August 2026, claim 8306); and the $3.5B Anthropic data center financing (July 2026, claim 5374) represents customer-backed capital, not speculative raise. The timing of the withdrawn $200M ATM facility (May 2025, claim 4992) suggests management chose secured paths once available. Q2 2026 results reveal the leverage inherent in this model: $45M revenue (claim 765) against $940M net loss (claim 756) and a 27x debt-to-equity ratio (claim 742), yet negative working capital of $957M (claim 760) and free cash flow loss (claim 775) did not derail a $4–4.5B Justified Data Campus commitment (claims 1398, 1408). The Abernathy JV monetization for $450M (August 2026, claim 8913) suggests strategic asset rotation. A sophisticated reader should monitor: whether the $19B lease contract and Anthropic relationship survive industry shifts; whether Google's $1.8B equity stake signals long-term anchor support or signals eventual exit; and whether current construction-phase losses on that scale can compress once utilization ramps. The data shows no use-of-proceeds disconnect—announced capital is visibly deployed into real capacity—but the burn rate against early-stage revenue requires either accelerating customer ramp or refinance at rates assuming near-term EBITDA inflection.

TeraWulf has raised approximately $13 billion over twenty-one months through a series of escalating financing windows that reflects tactical access to capital rather than coherent capital-structure design. The initial phase introduced convertible debt with 3 percent coupons and conversion prices near $8, per claims 5492 and 5495, yielding roughly $2–3 billion in proceeds by May 2025. That was followed by $4.5 billion in senior secured notes due 2030 at 7.75%, per claims 4215 and 4187, issued across two tranches in late 2025. The company then accessed public equity markets for $800 million in common stock and secured $1.8 billion from Google and $578 million from BlackRock, while also raising $3.5 billion tied to capacity commitments for Anthropic. Each window opened sequentially—convertibles when the market priced them, debt when rates permitted, equity and strategic capital once partnerships formed. This pattern suggests financing decisions followed operational cashburn and available windows rather than a premeditated strategy.

Announced use of proceeds has remained consistent: data center construction and lease prepayment. The company committed $4–4.5 billion to Justified Data Campus development, per claims 1408 and 6724, and deployed capital to Hawesville through a $500 million bridge facility, per claim 2102. Yet the operating picture creates immediate tension. Q2 2026 revenue reached only $45 million, per claim 2804, while the company reported a $940 million net loss and negative free cash flow, per claims 769 and 775. Cash balances fell to $3 million, per claim 771, against a debt-to-equity ratio of 27, per claim 742, and negative working capital of $957 million, per claim 774. The capital raised has almost certainly been capitalized into construction assets rather than consumed in operations, and the company now faces substantial debt servicing against unprofitable operations.

The outcome depends entirely on timing of customer deployment and cash capture. The Google partnership includes a $19 billion revenue commitment, per claim 8306, but cash realization timing and margin profiles remain undisclosed. The recent sale of Abernathy joint venture stake for $450 million and the influx of strategic capital must sustain the company and complete Justified development until customer revenue scales to cover operating losses and debt service. Monitor whether the current cash position and burn rate provide adequate runway, or whether near-term refinancing pressures emerge before customer contribution margins prove sufficient.

TeraWulf has executed a capital structure pivot from equity dilution through convertible debt to secured leverage, now requiring customer partnerships to fund continued expansion. Through October 2025, the company issued approximately $3.3 billion in convertible senior notes (per claims 5492, 4991, 4441, 1057, 4167), with net proceeds consistently 90-97% of stated amounts. Simultaneously, it raised $4.5 billion in senior secured debt at 7.75% (per claims 4187, 3951), due 2030-2031. This two-tranche debt issuance within two months—rather than a unified capital raise—suggests neither the convertible market nor the secured market alone could absorb the capital requirement, a sign of stretched capacity at each instrument tier. The announced use of proceeds remained largely generic (data center expansion, per claim 4189; capped call hedges consuming $86-101 million per claims 4444, 4435) until Q2 2026, when proceeds allocation became specific: Hawesville facility (claimed Apr 2026), Anthropic-leased facility ($3.5 billion, per claims 5374, 2), Justified Data Campus ($4-4.5 billion, per claims 1398, 6724). By August 2026, TeraWulf had pivoted entirely to customer-anchored financing, accepting $1.8 billion in Google equity (claim 7811) and $578 million from BlackRock (claim 1159) while reporting Q2 revenue of $45 million and net losses of $940 million (per claims 1400, 756). The balance sheet shows a debt-to-equity ratio of 27x and negative working capital of $957 million (per claims 742, 760), yet the company announced $19 billion in contracted revenue from a major lease (claim 8306). What requires scrutiny: whether contracted revenue translates to recognized revenue, given the company's operating loss profile; whether the Google and BlackRock checks represent genuine co-investment or customer development financing; and whether the equity dilution from these strategic rounds (compounded by convertible conversion risk) leaves sufficient equity cushion to absorb execution risk on four data center projects totaling $8-9 billion in capex announcements.

TeraWulf's financing sequence from October 2024 through August 2026 reveals a company executing rapid infrastructure buildout while absorbing acute liquidity pressure. The cadence moved from exploratory convertible issuances—a $350 million offering in October 2024, per claim 5532—toward aggressive debt monetization starting in fall 2025, when the company completed a $3.2 billion senior secured notes offering (October, claim 4215) followed by a $1.3 billion tranche (December, claim 3965). Simultaneously, convertible offerings escalated from $500 million to $850 million to $1 billion in 2025 (claims 4492, 4441, 4167), creating a layered dilution structure. The authorized share amendment, delayed as of September 2025 (claim 4414), suggests capacity constraints.

The capital mix reads less as coherent strategy and more as sequential pressure relief. The 2024 convertible base was modest; the 2025 pivot toward $4.5 billion in total senior secured debt (including the $1.3 billion tranche) indicates debt became the primary lever as equity dilution faced market skepticism. By April 2026, the company returned to direct equity with an $800 million offering (claim 2120), supplemented by strategic partnerships: Google warrant issuances totaling $41 million (claim 4510), a BlackRock investment of $578 million (claim 1159), and a $3.5 billion facility specifically for Anthropic-leased capacity (claims 2, 5374). This progression—debt, then equity, then partnerships—suggests each bucket served a tactical need rather than a premeditated architecture.

Announced use of proceeds nominally tracks deployment: data center expansion recurs in claim 4445 (August 2025) and claim 4189 (October 2025), with Hawesville explicitly named by April 2026 (claim 2103). Yet the arithmetic deteriorates on inspection. Q2 2026 reported revenue of $45 million (claim 2804) against a net loss of $940 million (claim 769), with free cash flow of negative $1 billion (claim 759). The Justified Data Campus carries a $4-4.5 billion investment estimate (claims 6724, 6706), still in planning as of August 2026. The Abernathy joint venture sale for $450-530 million (claims 8913, 8411) in July appears to be liquidity generation rather than portfolio optimization; this read is reinforced by negative working capital of negative $957 million (claim 760) and a debt-to-equity ratio of 27x (claim 742).

A sophisticated reader should monitor whether the $3.5 billion Anthropic facility closes materially and on what terms, as it appears contingent on unfinished campus development. The Google relationship—evidenced by lease credit supports and warrant holdings—may be the marginal source of forbearance on the $4.5 billion debt stack given negative cash generation. Most pressingly, the gap between capital deployed and revenue realization has widened into double-digit billions; execution risk on return timelines now supersedes financing risk.

TeraWulf's capital structure reflects a progression from opportunistic equity raises through convertibles to structured debt, culminating in its aggressive 2025-2026 buildout. From October 2024 through August 2025, the company executed five separate convertible offerings totaling roughly $2.7 billion (per claims 5532, 5502, 5492, 4454, 4433, 4441), each paired with capped-call hedges ($51M–$86M per claims 5506, 4452, 4444, 4435) designed to suppress dilution but implicitly acknowledging near-term equity weakness. The withdrawal of its $200M ATM facility (per claim 4992) in May 2025 signals either constrained market access or a deliberate pivot away from equity. The company shifted sharply toward leveraged debt, issuing $3.2 billion in senior secured notes (per claim 4215) and $1.3 billion additional (per claim 3968) within six weeks. This stack—convertibles plus $4.5 billion in secured borrowings plus $500M bridge facilities (claims 2102, 2084, 2067)—reads as construction financing rather than working-capital management.

The stated use of proceeds remained vague until late 2025. Early convertible issuances carried no clear allocation; the May offering mentioned data-center expansion only indirectly (claim 4445). By October, the secured issuance explicitly referenced "completion guarantee" language (claim 4189), and the April 2026 public offering targeted "Hawesville and bridge repayment" (claim 2103). A Google warrant grant (claims 4510, 4459) and later an $1.8 billion equity investment (claim 7811) suggest a customer-financed consumption model, reinforced by a major lease commitment (claim 8306: $19 billion stated). Yet Q2 2026 results expose the underlying strain: revenue of only $45 million against net losses of $940 million and negative free cash flow (claims 2804, 756, 775), alongside negative working capital of $957 million (claims 760, 774). The company is burning capital faster than deployed capacity can absorb it.

A reader should monitor three constraints. First, debt service on negative EBIT of $883 million (claim 768) against $4.5B+ in secured obligations leaves minimal covenant headroom. Second, the dilution stack—multiple convertibles at strike prices of $8 or lower (claim 5495), plus Google warrant overhang—will materialize upon stock appreciation, adding pressure precisely when capital efficiency is most critical. Third, the $4 billion Justified Data Campus investment (claims 1408, 1414, 6706) must produce revenue sufficient to service this structure within 18–24 months. Revenue ramp from current utilization levels, not further refinancing, is the sole pathway to sustainability.

TeraWulf's capital strategy reflects a company building data-center infrastructure at scale while opportunistically accessing multiple funding sources—a fundamentally sound approach that nevertheless shows increasingly pronounced strain. Beginning with $350–500 million in convertibles in October 2024 (per claims 5532, 5502, 5492), the company escalated to a $3.2 billion senior secured debt offering in October 2025 (claim 4215) and an additional $1.025 billion convertible tranche in late October 2025 (claim 1057), with a further $1.3 billion secured offering in December 2025 (claim 3951). The stated uses tracked the capex narrative: data-center expansion (claim 4445), Hawesville facility construction (claim 2239), and ultimately $3.5 billion earmarked for Anthropic-leased capacity (claim 5374). Google's warrant issuance in August 2025 (combined $73.6 million across claims 4510 and 4459) preceded a $1.8 billion equity investment in August 2026 (claim 7811), suggesting a genuine partnership rather than mere financing.

Yet coherence frays when examined closely. Multiple convertible offerings show revised figures (claims 5502 and 4502 marked as "delayed"), and a massive charter amendment for 950 million additional authorized shares (claim 4414, delayed September 2025) signals dilution beyond current iterations. The April 2026 $800 million public equity offering (claims 2101, 2083) arrived alongside a $500 million bridge facility (claim 2067)—suggesting capital-markets stress mid-cycle rather than planned funding. Most tellingly, TeraWulf sold its Abernathy joint-venture stake for $450 million in July 2026 (claim 8913), a mid-cycle asset sale inconsistent with orderly capital planning. By Q2 2026, the company reported $45 million revenue against a $940 million net loss (claims 765, 769), a debt-to-equity ratio of 27 (claim 742), negative working capital of $957 million (claim 774), and negative free cash flow (claim 775). The announced $4–4.5 billion Justified Data Campus remains unfinanced. The company has accessed all levers simultaneously—debt, convertibles, equity, strategic capital—which suggests not strategy but mounting urgency. Watch whether utilization ramps fast enough to sustain this burn profile, and observe whether additional capital raises arrive in Q4 2026 or beyond. The Google partnership may be essential support rather than optional advantage.

TeraWulf's capital trajectory from mid-2024 through mid-2026 reveals opportunistic financing rather than coherent strategy. Initial convertible offerings in October 2024—ranging from $350 million to $500 million across multiple tranches, per claims 5532, 5502, and 5492—were modest relative to what followed. By October 2025, the company had issued approximately $3.2 billion in senior secured notes (claim 4187) and $1.025 billion in convertibles (claim 1057), signaling a marked pivot toward debt. The April 2026 public equity offering of $800 million (claim 2101), followed within months by a $1.8 billion Google equity investment (claim 7811) and $3.5 billion Anthropic facility financing (claim 2), suggests exhaustion of conventional capital markets and shift to strategic partnerships. Dilution accelerated sharply: the company issued 20 million shares to Riesling in October 2024 (claim 5583) and pursued public equity after relying on convertibles, indicating compressed borrowing capacity or deteriorating equity terms.

Stated use of proceeds does not cleanly track actual deployment. Early filings emphasized data center expansion at Hawesville, Cayuga, Akela, and Abernathy, with relatively modest lease commitments. By August 2026, the disclosed $19 billion lease agreement with Google (claim 8306) and Justified Data Campus commitments of $4 billion to $4.5 billion (claims 6724, 6706) dwarf prior stated plans. The company simultaneously divested the Abernathy joint venture for $450 million (claim 8913), suggesting asset rotation rather than pure growth deployment. This revision in scale and mix requires explanation: did earlier forecasts dramatically underestimate Google demand, or did the company opportunistically capitalize on a partner's credibility to unlock capital markets access?

The balance sheet shows material stress. Q2 2026 financials disclose operating losses of $140 million on $45 million revenue (claims 767, 765) and net losses of $940 million (claim 756). Negative working capital of $957 million (claim 760) paired with a 27-to-1 debt-to-equity ratio (claim 742) indicates heavy leverage on a pre-revenue-scale platform. Current ratio of 1.0 (claim 744) warrants liquidity monitoring. Sophisticated readers should track: whether the Google partnership and Anthropic commitments materialize as disclosed; the feasibility of deploying $4 billion-plus in capital while operating at nine-figure quarterly losses; and what delayed claims around preferred stock conversion and charter amendments (claims 4039, 4038) signal about operational friction. The company has demonstrated access to capital but not yet a revenue model that justifies the scale of leverage incurred.

TeraWulf's financing trajectory from October 2024 through August 2026 charts a coherent but accelerating response to hyperscale customer demand, with the capital structure shifting decisively from balanced instruments to debt-heavy deployment. October 2024 opened with measured positioning: a $500M convertible offering (claim 5492) paired with $115M share repurchase authorization (claim 5500), signaling management confidence. By spring 2025, this confidence wavered—the company withdrew a $200M ATM offering (claim 4992), abandoning the public equity path. The pivot became explicit through the second half of 2025: $3.2B in senior secured 7.75% notes due 2030 (claim 4187), $1.025B convertible offering (claim 1057), and $1.3B additional senior secured notes (claim 3951) completed the debt-dominant mix. This instrument selection was strategically coherent: convertibles allowed upside capture if execution succeeded; senior secured notes funded tangible assets at lower cost. A mandatory preferred stock conversion in November (claim 4038) cleared the cap table before April 2026's $900M common offering (claim 2083), signaling confidence had returned.

Use of proceeds consistently tracked stated deployments. The October 2024 Riesling facility ($12M cash plus 20M shares, claims 5584, 5583), August 2025 Google warrant issuance ($41M, claim 4510), and April 2026 Hawesville bridge facility ($500M, claim 2102) were each tied to specific capacity additions. The July 2026 $3.5B raise explicitly funded an Anthropic-dedicated data center (claim 2), immediately followed by Google's $1.8B equity investment (claim 7811). Deployment discipline appeared intact through placement announcements.

Q2 2026 results, however, exposed material disconnect. Revenue of $45M (claim 2804) against $940M net losses (claim 756) and $957M negative working capital (claim 774) suggests either massive pre-revenue depreciation on newly-built capacity or that major leases booked slower than capital outflow. The debt-to-equity ratio reached 27x (claim 742), extreme even for asset-heavy businesses. The August announcement of a $19B contracted revenue lease (claim 8306) and $4-4.5B Justified Data Campus investment estimate (claim 1414) indicate future conviction, but near-term cash burn is unsustainable without swift lease activation.

Key watches: whether Q3-Q4 2026 shows revenue inflection as Google and Anthropic facilities generate revenue; refinancing risk on the 2030 convertible and senior note maturity walls; whether the $4.5B+ Justified campus requires incremental capital or if July-August raises ($5.3B combined) suffice; and working capital trajectory—at negative $957M, cash sustainability hinges entirely on incremental facility cash flow commencement.

TeraWulf's capital progression from 2024 through August 2026 reveals a company cycling through funding sources as prior channels closed, rather than executing a coherent long-term strategy. The pattern began light: October 2024 saw convertible notes aggregating roughly $1.3 billion (claims 5532, 5502, 5492) paired with the Riesling Power transaction (claim 5583, 20 million shares and $12 million cash for operational facilities). By May 2025, a planned $200 million ATM offering was withdrawn (claim 4992), signaling market access deterioration. Management then accelerated convertible issuance dramatically—August 2025 brought $850 million and $400 million convertible offerings (claims 4447, 4454), followed by $1 billion and $500 million tranches in August and October (claims 4433, 4180), accumulating over $2.2 billion in converts. This escalation suggests converts became necessary as equity dilution reached ceiling. The capital mix then shifted sharply: October 2025 introduced $3.2 billion in senior secured notes (claim 4215), followed by $1.3 billion more in December (claim 3951). These debt tranches—at 7.75% interest on the October notes—indicate management expecting sufficient cash generation to service ~$350 million in annual debt service. April 2026 brought the first public equity offering at $900.6 million (claim 2066), suggesting even debt capacity tightened. Most recently, July 2026's $3.5 billion financing explicitly for Anthropic-leased data centers (claim 5374) and August's $1.8 billion Google equity investment (claim 7811) show dependency on either strategic partners or alternative capital structures.

Yet operational deployment lags capital raises substantially. Q2 2026 revenue was $45 million (claim 765), declining year-over-year (claim 1402), while net losses reached $940 million (claim 756). The company carries negative working capital of $957 million (claim 774) and a debt-to-equity ratio of 27x (claim 742). The $19 billion in contracted revenue from a major lease (claim 8306) remains undeployed cash. This disconnect—massive capital raises funding multi-billion-dollar data center construction while operational metrics deteriorate—requires scrutiny. Google's deepening involvement, from warrant holder to credit supporter for leases (claims 3636, 3647) to $1.8 billion equity investor, signals either substantial confidence in the underlying lease economics or consolidation of influence. The August 2026 announcement of $4-4.5 billion expected investment in the Justified Data Campus (claim 1414) compounds capital intensity: the company has now committed to projects exceeding available liquidity within any near-term horizon.

A sophisticated reader should monitor whether contracted lease revenues—particularly the Anthropic relationship—begin converting to actual cash before debt maturity dates arrive (2030 on the senior secured notes). Construction timelines and cost escalation at Justified warrant attention; data center buildouts routinely exceed budgets. Most critically, the $957 million negative working capital position with a current ratio near 1.0 (claims 760, 761) leaves no margin for deployment delays or revenue disappointments. Refinancing risk in 2028–2029 is material if operating leverage has not materialized by then.

TeraWulf raised approximately $7 billion in convertibles between October 2024 and December 2025, clustering issuances in October 2024 and August–October 2025, with conversion prices near $8 per share (claim 5495) and capped call strikes at $13 (claim 5499). Alongside convertibles, the company issued $4.5 billion in senior secured debt—$3.2 billion at 7.75% due 2030 (claim 4187) and $1.3 billion in December 2025 (claim 3951). The cadence reveals opportunistic patching rather than a locked master plan. Multiple tranches launched within weeks, several marked delayed in announcement, indicating equity discipline was secondary to cash acquisition speed.

Financing mix shifted materially in 2026. Google provided warrant issuance (~$32.6M, claim 4459) and $1.8 billion in equity investment (claim 7811), alongside credit support for data center leases, signaling deepening strategic alignment around capacity. BlackRock contributed $578M in equity (claim 1159). A public offering closed in April 2026 at $900M gross, though net proceeds claims carry delayed status (claims 2081, 2066). This pivot from convertible-heavy to equity-inclusive reflects prior headwinds: a $200M ATM offering initiated May 2025 was withdrawn (claim 4992), suggesting equity market receptivity questions at that moment.

Announced uses strain coherence with actual deployment. Proceeds targeted data center expansion and the Anthropic facility ($3.5B, claim 5374), with Kentucky's Justified Data Campus requiring $4–4.5 billion in development (claim 1398). Q2 2026 results, however, showed $45M revenue (claim 2804), negative $6M year-over-year decline (claim 1402), negative $2 EPS (claim 1400), and $940M net loss (claim 756). Free cash flow was negative (claim 759); working capital is negative $957M (claim 760). This gap—massive capital deployed against operating losses and minimal revenue—requires explanation. Either capex is still upstream with revenue ramp delayed, or deployment has missed projections. A $500M bridge facility in April 2026 (claim 2102), after $4.5B+ already raised, signals cash urgency.

Forward watch: (1) Series A preferred conversion completion status (delayed as of November 2025, claim 4039); (2) Form S-3 registration status (delayed August 2025, claim 4503); (3) debt maturity profiles and Google's quasi-anchor tenant role, likely entailing asymmetric covenants; (4) sequential revenue and free cash flow trends. Operating losses at 27x debt-to-equity leverage (claim 742) become unsustainable if revenue inflection does not arrive within 12–18 months.

TeraWulf has pursued a capital-intensive, infrastructure-centric dilution strategy that evolved coherently from 2024 through mid-2026. The company began with modest convertible issuances—$425 million and $500 million tranches in October 2024 (claims 5502, 5492)—paired with operational consolidation: the $12 million cash settlement to Riesling Power and 20 million share issuance (claim 5583) to resolve lease arrangements. This initial equity dilution was minimal and strategic. The mix then shifted systematically. By August 2025, TeraWulf layered additional convertibles ($850 million, claim 4441) alongside warrant issuances to Google ($32.6 million, claim 4459), signaling a shift toward anchor-tenant partnerships. The strategic inflection arrived in October 2025 when the company issued $3.2 billion in senior secured debt at 7.75% due 2030 (claim 4187) immediately followed by $1.025 billion in convertible notes (claim 1057). This debt-first posture continued in December with a further $1.3 billion senior secured issuance (claim 3951). The progression from convertibles to secured debt reflects escalating capital intensity as the company committed to the Justified Data Campus in Kentucky—an estimated $4 to $4.5 billion deployment (claims 1398, 1408). Only when debt markets tightened did pure equity emerge: an $800 million common offering in April 2026 (claim 2101) followed by strategic equity checkpoints—Google's $1.8 billion investment (claim 7811) and BlackRock's $578 million position (claim 1159) in August 2026. This is capital strategy, not opportunism. However, the announced deployment trajectory now requires scrutiny. The company committed $3.5 billion toward an Anthropic-leased data center in July 2026 (claim 2), yet Q2 2026 results revealed net losses of $940 million (claim 769), negative working capital of $957 million (claim 760), and a debt-to-equity ratio of 27x (claim 742). More concerning: revenue declined year-over-year (claim 2805) while capital raises accelerated. The company's Form S-3 registration filing has been delayed (claim 4503), and its prior $200 million ATM offering was withdrawn (claim 4992). Sophisticated investors should monitor whether Kentucky capex actually deploys on the disclosed timeline, whether the Anthropic transaction represents genuine TeraWulf deployment or financing pass-through, and whether rising debt service (at 7.75% due 2030) becomes unsustainable if revenue growth remains inverted relative to capex.

TeraWulf has pursued an increasingly aggressive capital-raising cadence that reveals a coherent if capital-intensive strategy. The company began with modest convertible issuances in October 2024—$487 million in net proceeds (claim 5497)—but by mid-2025 had shifted to secured debt. The October 2025 issuance of $3.2 billion in senior secured notes (claim 4187), followed by $1.025 billion in convertibles (claim 1057) and a $1.3 billion senior secured raise in December 2025 (claim 3951), established debt as the dominant financing vehicle. The July 2026 $3.5 billion capital raise explicitly tied to Anthropic data center leasing (claim 2) marks a maturation of this approach: customer-backed financing rather than issuer-only capital raising.

The equity dilution from announced share issuances remains modest in percentage terms—20 million shares to Riesling in October 2024 (claim 5583) and an $800 million equity offering in April 2026 (claim 2083)—but is dwarfed by convertible-embedded dilution and debt accumulation. The $200 million share repurchase authorization (claim 5531) was delayed and appears abandoned, which is telling given the scale of capital raises.

Announced use of proceeds tracks stated deployment: Hawesville and Justified Data Campus buildout (claim 2103, claim 1414). The company guided to $4.0–4.5 billion in initial Justified investment (claim 1398), consistent with the overall raise trajectory. However, Q2 2026 results show a $940 million net loss on $45 million in revenue (claim 756, claim 765), with negative working capital of $957 million (claim 760) and a debt-to-equity ratio of 27x (claim 742). Free cash flow was negative $1 billion (claim 759). This implies either severe execution delays, pricing pressure on capacity, or both.

The critical watch: revenue visibility against committed capex. Anthropic's multi-billion-dollar commitment reduces refinancing risk but introduces single-customer concentration. The company's ability to scale utilization and pricing from the Hawesville and Justified builds will determine whether the debt structure is sustainable or requires restructuring. The absence of an active ATM offering (claim 4992 was withdrawn) suggests capital markets may be pricing execution risk. Monitor the cadence of quarterly utilization announcements and lease-rate guidance.

TeraWulf's capital record shows a single disclosed financing: $3.5 billion (per claim #2, July 13, 2026), explicitly dedicated to Anthropic-leased data center infrastructure. The precision of stated use-of-proceeds is atypical and useful. Yet the absence of other capital claims in the record—prior equity, convertibles, debt—creates ambiguity: either TeraWulf operated without external capital until now, or earlier rounds are missing. Without financing history, whether this $3.5B represents coherent strategy or opportunistic response to a major customer opportunity cannot be assessed.

Insider transactions suggest shifting management conviction. Through mid-2025, insiders bought steadily and sold minimally—just $83K in Q2 2025 and $624K in Q3 2025. This posture reversed sharply in 2026: insiders sold $14.2 million in Q1, peaked at $16.5 million in Q2 across 24 transactions, and continued $2.3 million in Q3. This sustained selling aligned with or preceded the July $3.5B announcement. Such profit-taking at the moment of massive capital infusion typically reflects either conviction that stock has appreciated ahead of dilution, or concern that dilution will be material.

The Anthropic deployment claim is verifiable in real time, unlike generic "general corporate purposes" language. Quarterly capital outflows and infrastructure buildout become observable immediately.

Three items warrant close watch. First, track actual capital deployment into Anthropic infrastructure starting Q4 2026; delays or revised allocation signal execution or commitment risk. Second, establish the company's complete prior financing history to assess whether $3.5B is truly the first major raise; that context affects dilution trajectory assessment. Third, monitor insider selling post-close; sustained elevated selling after capital raise closure implies management skepticism about the company's ability to generate returns sufficient to offset shareholder dilution.

The record contains no signed customer agreements in the conventional sense. Through early 2023, TeraWulf's disclosed contracts are almost entirely supply-side: power, equipment, custody, execution, pool and joint-venture arrangements. Revenue comes from self-mined bitcoin routed through Foundry's pool (claims 13440, 13335, 13162), so Foundry functions as the de facto revenue counterparty without being a customer. The one oblique reference to a demand-side relationship, bitcoin 'earned from hosting arrangements' (claim 12852), names no hosting client. Claim 21075 concerns IKONICS, the predecessor registrant, and should be excluded from TeraWulf's commercial history. Most of the remaining language is executed rather than aspirational. The Minerva purchase agreement is 'executed' and the Bitmain contracts are four signed sales and purchase agreements (claims 14028, 14051, 13977). The NYDIG custody and execution agreements carry specific signing dates (claims 13412, 13414), and the NYPA and Susquehanna power arrangements are described as secured (claims 12842, 14129). The softer elements sit around them. The 58,000-miner figure is framed with 'Despite securing', which implies delivery shortfalls (claims 14123, 14247). Bitmain pricing is 'non-fixed' (claim 14051). NYPA capacity carries only the 'potential to expand' (claim 12842), and the Cumulus 100 MW expansion is an option that expires May 13, 2024 (claim 12922). Every material counterparty is named, with no unidentified 'strategic partners'. Disclosure is uneven on two points, however: Cumulus Member is identified only as a Talen affiliate (claims 14120, 13714), and two counterparties are related parties. Beowulf E&D is owned by the CEO (claims 12845, 12942), and NovaWulf is described as an affiliate (claim 13849). Concentration has narrowed rather than diversified. TeraWulf's Nautilus stake fell from 33.33% to 25%, with board control 4-to-1 for Cumulus (claims 13714, 12872, 12874), while NYDIG consolidated custody, execution and DACA roles (claims 13344, 13345, 12905). Little of the revenue arithmetic can be reconciled from this record. The only hard figures are a $19,072 (thousands) joint-venture investment (claim 13918) and the $339,200 transition fee (claim 12942). Neither ties to hashrate, miner deliveries or bitcoin revenue, and the claim list provided is truncated at claim 12891, so later hosting or HPC customer contracts are not assessed here.

TeraWulf's record through early 2023 contains almost no signed customer agreements in the conventional sense. As a self-mining bitcoin operator, its revenue comes from block rewards paid through a pool rather than from named buyers. Nearly every counterparty on file is a supplier, power provider, joint-venture partner or service vendor. The most concrete executed agreements are on the supply side: an equipment purchase agreement with Minerva and four non-fixed-price sales and purchase agreements with Bitmain (claims 14028, 14027, 14051, 13977). These were tied to delivery of 58,000 miners during 2022 (claims 14247, 14123). The 'despite securing' framing in claim 14123 hints that delivery fell short of that commitment. Power is similarly documented. Nautilus draws Susquehanna nuclear output under a five-year term with renewal options (claims 14113, 14129), and Lake Mariner has an initial 90 MW from NYPA (claims 12842, 12890). The earlier 'competitive power supply agreements already in place' (claim 21144) names no counterparty and should be read as less verifiable than these later disclosures. Every operational counterparty is named, and nowhere does the company describe an unidentified 'strategic partner' or 'anchor customer'. These include Talen, its affiliate Cumulus, Foundry as pool operator, NYDIG as custodian and NYDIG Execution as selling agent (claims 14120, 13440, 13412, 13414). Concentration has shifted from dependence on two equipment vendors in 2022 toward a narrow operating stack. Talen holds the Nautilus JV, and TeraWulf's interest in it fell from 50% of 300 MW (claim 14112) to 33.33% (claim 13714) and then to 25%, with one of five board seats (claims 12872, 12921, 12874, 12920). Foundry and NYDIG remain single points of dependence for revenue collection and custody. The affiliate relationships with NovaWulf and Beowulf E&D, including the $339,200 transition fee, are related-party arrangements rather than arm's-length commercial wins (claims 13849, 13718, 12942). Revenue arithmetic cannot be reconciled from this record. No revenue figures, hosting contracts or offtake prices are disclosed. The only quantified flow is a $19,072 thousand JV investment outflow (claim 13918), and Nautilus distributions are pro rata to hash rate net of costs (claim 12939) rather than contracted. Claim 21075, a 2006 IKONICS auditor ratification, is predecessor-shell noise and should be excluded. The claim set appears truncated at claim 12919, so later disclosures, including any hosting or HPC customer agreements, are not assessed here.

The record supplied for TeraWulf contains almost no customer agreements in the commercial-diligence sense. Nearly every executed contract is upstream (power, equipment, custody, brokerage) or structural (joint venture, related-party services). Revenue counterparties are essentially absent. One entry, the 2006 auditor ratification (claim 21075), belongs to IKONICS, the legacy shell, and should be disregarded.

The signed agreements are well identified and carry concrete terms. They include Bitmain's four non-fixed-price purchase agreements and the Minerva equipment purchase agreement (claims 14051, 14028, 13977), which together provide for 58,000 miners in 2022 (claims 14247, 14123). The Talen Energy joint venture dated May 13, 2021 (claims 14120, 14112) is also signed. So are the Susquehanna nuclear power arrangement, a five-year term with two three-year renewals (claims 14113, 14129), and the NYPA supply of an initial 90 MW at Lake Mariner (claims 12842, 12890). The NYDIG custody and execution agreements (claims 13412, 13414, 13333, 13334) and the Foundry pool service agreement (claim 13335) complete the executed set. The only soft language is the November 2021 statement that 'competitive power supply agreements' were 'already in place' without naming a counterparty (claim 21144). That gap is resolved by later filings naming Talen/Susquehanna and NYPA. No MOU or LOI language appears, but no unnamed 'anchor customer' or 'strategic partner' appears either. The one reference to 'hosting arrangements' (claim 12852) names no hosting client.

Counterparty concentration has increased over the period. Economically, the business depends on Foundry for block rewards and on NYDIG for custody, sale and a deposit control agreement with Wilmington Trust (claim 12905). Operationally, much of the build-out runs through Beowulf E&D, which is controlled by the CEO (claims 12845, 13718, 12942), and through the affiliate NovaWulf (claim 13849). TeraWulf's position in Nautilus also weakened. Its stake fell from 33.33% (claim 13714) to 25%, with one of five board seats (claims 12872, 12874). Talen's Cumulus retains a unilateral 100 MW expansion option (claim 12922).

The revenue arithmetic cannot be reconciled from this record. No revenue figure, hosting fee or customer contract value is disclosed against which the deals could be tested. The only quantified flows are outbound: the $19.1 million JV investment in Q1 2022 (claim 13918) and the $339,200 transition fee paid to the related party (claim 12942). Revenue is self-mined bitcoin distributed biweekly net of costs (claim 12939), so it has no contracted-customer basis to reconcile. The record supplied ends mid-entry at claim 12891 in March 2023. Any later customer or hosting agreements are therefore outside this assessment.

TeraWulf's record through March 2023 contains no signed revenue customer at all, and that is the central finding. Every named counterparty is a supplier, power provider, service vendor, joint-venture partner or related party. The revenue model is self-mining: bitcoin is earned through a pool run by Foundry Digital (claims 13440, 13335, 12854), held by NYDIG under a custodial agreement dated March 10, 2022 (claims 13412, 13344), and sold through NYDIG Execution under a September 16, 2022 agreement (claims 13414, 13334). These are executed contracts, but Foundry is a pool operator, not a buyer of capacity, so no third party is contractually committed to pay TeraWulf for anything. Claim 21075, an IKONICS auditor ratification from 2006, is shell-predecessor housekeeping and should be disregarded.

On naming, the record improves over time. The November 2021 statement that 'competitive power supply agreements' were already in place named no counterparty (claim 21144). Later filings identify NYPA for an initial 90 MW at Lake Mariner (claim 12890) and Susquehanna Station nuclear power at about 2.0 cents/kWh for Nautilus (claims 14113, 12859). Equipment supply is identified with Bitmain, under four non-fixed-price agreements, and Minerva (claims 14051, 14028). However, the 58,000-miner delivery commitment is framed defensively ('Despite securing...', claim 14123), which points to a delivery shortfall against the stated volume.

Concentration is high and has tilted toward Talen. TeraWulf's Nautilus stake fell from a 33.33% equity interest (claim 13714) to 25% with one of five board seats (claims 12872, 12874), so economics at the flagship nuclear site are now minority and dependent on a partner. Related-party exposure is material. Beowulf E&D, controlled by the CEO, provides infrastructure and administrative services (claim 12845) and a $339,200 transition arrangement to Nautilus (claim 12942). NovaWulf is an acknowledged affiliate (claim 13849).

Revenue arithmetic cannot be reconciled from this record because no contracted revenue is disclosed. The only quantified cash flows are outflows: $19.07 million in JV investments in Q1 2022 (claim 13918) and the Beowulf transition fee. Any later narrative of named hosting or compute customers falls outside these claims. The record is truncated at claim 12848 and should be tested separately rather than inferred from the infrastructure contracts above.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, and no MOU, LOI, pilot or 'discussions' language either. Every named counterparty sits on the supply, financing or service side of a self-mining business whose revenue is block rewards, so the classification question is moot. The record holds executed equipment contracts with Bitmain and Minerva (14051, 14028, 13977, 12847). It holds a joint venture with a Talen affiliate (14120, 12848) and nuclear power sourced from Susquehanna Station (14113, 12859). Foundry runs the mining pool (13335, 12854), NYDIG provides custody and execution (13412, 13414), and Cantor, B. Riley and D.A. Davidson act as sales agents (13973). Claim 21075 refers to the IKONICS shell's 2006 auditor and is not relevant. The record has no unnamed 'strategic partners' or 'anchor customers.' Disclosure is specific, but it identifies vendors, not buyers. The closest thing to a revenue counterparty is Foundry, which credits daily pool earnings (13441), so pool-level concentration is total.

The main shift in concentration is the shrinking Talen joint venture. TeraWulf was initially framed as owning 150 of 300 MW (14112). That fell to a 33.33% stake with 20,000 contributed miners (13714, 13713), then to 25% with one of five board seats (12872, 12874, 12860). Governance now rests with Cumulus, and bitcoin distributions are made net of each member's share of power and operating costs (12939). Related-party exposure is also material. Beowulf E&D, a Prager-linked affiliate, carries an $8.46 million annual base fee (12867) and a $339,200 transition services agreement with Nautilus (12942). NovaWulf is also an affiliate (13849), and the insider voting agreement in 13391 ties up shareholder votes.

The revenue arithmetic cannot be reconciled from this record because it discloses no revenue figures. What it does show raises execution questions. The 58,000-miner delivery commitment for 2022 (14247) is later qualified with 'despite securing' (14123), which implies delivery or deployment fell short. The Bitmain contracts are non-fixed-price (14051), so unit economics were not locked in. The $19.1 million joint venture investment in Q1 2022 (13918) predates the stake reductions, and the record does not show how capital contributions were re-marked. Any later hosting or HPC customer contracts fall outside this record and would need separate verification.

TeraWulf's claim record contains no signed customer agreements in the conventional sense. Every named counterparty is a supplier, power provider, joint-venture partner, service provider or financing agent, and that fits a self-mining model in which revenue comes from the company's own bitcoin production (claim 13441). One record should be set aside entirely: claim 21075 is a 2006 IKONICS auditor ratification inherited from the predecessor shell, and it has no commercial bearing on TeraWulf. Within what remains, the language is mostly executed-agreement language rather than MOU, LOI or pilot framing. The Bitmain and Minerva purchase agreements are described as executed (claims 14028, 13977, 12847), although the Bitmain contracts are explicitly non-fixed price and were later supplemented by future sales agreements. The 58,000-miner delivery commitment (claims 14247, 14123) is introduced with "despite securing," which signals that deliveries fell short of the contracted schedule. The Foundry pool, NYDIG custody and NYDIG Execution arrangements are dated, named contracts (claims 13335, 13412, 13414). The 2021 statement that power agreements were "already in place" named no counterparty (claim 21144), and only the later Nautilus filings identified Susquehanna Station and Talen (claims 14113, 12859). None of the counterparties is hidden behind phrases like "strategic partner" or "anchor customer." The real disclosure issue is related-party exposure. Beowulf E&D provides operating services and receives an $8.46 million annual base fee plus a $339,200 transition payment (claims 12867, 12942), and NovaWulf is an acknowledged affiliate (claim 13849). Concentration has moved away from Talen over time. TeraWulf's attributable stake in Nautilus started at 150 of 300 MW, or 50 percent (claim 14112). The August 2022 amendment cut it to 33.33 percent (claim 13714), and the March 2023 terms set it at 25 percent with one of five board seats (claims 12872, 12874, 12860). Operational dependence, by contrast, has stayed concentrated in Foundry for pooling and NYDIG for custody and execution. On revenue, the record does not support a reconciliation. The only figures provided are the $19.072 million Q1 2022 joint-venture investment outflow (claim 13918) and the Beowulf fees. Bitcoin distributions from Nautilus depend on hash-rate contribution net of costs (claim 12939), so revenue is a function of hashrate, bitcoin price and power cost rather than contracted customer value. Readers should treat any future hosting or compute-customer announcements as a separate category that this record does not cover.

TeraWulf's record through early 2023 contains no signed revenue customer in the conventional sense, and that is the central diligence point. The company was a self-miner, so its named counterparties are suppliers, a power and joint-venture partner, service providers and related parties, not buyers of capacity. Revenue comes from block rewards earned through the Foundry Digital pool (claims 13440, 13335, 12854), and Foundry is a pool operator rather than a customer under contract. The executed agreements are firm and named. They include Bitmain and Minerva equipment purchases (claims 14051, 14028, 13977, 12847), NYDIG custody and NYDIG Execution sales agreements (claims 13412, 13414), a DACA with NYDIG and Wilmington Trust (claim 12905), and the Talen joint venture with its Susquehanna nuclear power arrangement at about 2.0 cents/kWh (claims 14120, 14113, 12859). The record contains no MOU, LOI or 'discussions' language. The one unidentified reference is the November 2021 statement that 'competitive power supply agreements' were in place for New York and Pennsylvania without naming the New York counterparty (claim 21144). Several claims are not commercial relationships at all. Claim 21075 is a 2006 auditor ratification inherited from the IKONICS predecessor. Claim 13973 is an at-the-market sales agreement with underwriters, and claim 13391 is a shareholder voting agreement. Related-party exposure deserves attention. Beowulf E&D, an affiliate, receives an $8.46 million annual base fee (claim 12867) and a $339,200 transition fee (claim 12942), and NovaWulf is disclosed as an affiliate (claim 13849). Concentration has moved in one direction. TeraWulf's attributable share of Nautilus fell from 150 of 300 MW, or 50% (claim 14112), to 33.33% (claim 13714), and then to 25% with one of five board seats (claims 12872, 12874, 12860). Over the same period, monetization, custody and pool access consolidated around Foundry and the NYDIG entities. The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed. The only quantified items are the $19.1 million joint-venture investment in Q1 2022 (claim 13918) and the 58,000-miner delivery commitment (claims 14247, 14123). The record also shows only 20,000 miners contributed to Nautilus (claim 13713), so deployment of the remaining 38,000 is unaccounted for here. Bitcoin distributions are netted against power and operating costs on a hash-rate basis (claim 12939), which means gross JV economics will not appear directly as TeraWulf revenue. Any later hosting or capacity-customer narrative should be tested against documents outside this claim set.

TeraWulf's record through early 2023 contains no named revenue customers. The company describes itself as a self-miner whose revenue "comes from our self-mined bitcoin" (claim 13441), so every named counterparty is a supplier, power provider, JV partner, service provider or financing agent. Claim 21075, a 2006 auditor ratification for IKONICS Corporation, belongs to the pre-merger shell and should be excluded from any reading of TeraWulf's commercial book. The claims run only to March 2023, so nothing here speaks to later hosting or compute arrangements.

Most counterparties are actually named and tied to executed instruments. These include the Talen joint venture dated May 13, 2021 (claim 14120), the NYDIG custodial agreement of March 10, 2022 (claim 13412), the NYDIG Execution agreement of September 16, 2022 (claim 13414), the Foundry pool service agreement (claim 13335), the Minerva purchase agreement (claim 14028) and the NYPA power arrangement (claim 12842). The softer language sits around those contracts. The four Bitmain agreements are explicitly "non-fixed price" (claim 14051), which limits what "secured" in claim 14246 can mean. The later "strategic relationship" with Bitmain (claim 12891) is narrative, not a new contract. NYPA's 90 MW carries only a "potential to expand" (claim 12890), and the November 2021 "power supply agreements already in place" (claim 21144) names no counterparty. Beowulf E&D (claims 13718, 12867) and NovaWulf (claim 13849) are affiliates, and their fees are related-party flows rather than arm's-length commercial validation.

Concentration has narrowed rather than diversified. Equipment depends on Bitmain and Minerva. Custody and sales run entirely through NYDIG, and pool payout runs entirely through Foundry. Exposure to Talen was cut in steps. TeraWulf was initially to be attributed 150 of 300 MW (claim 14112). Its equity then fell to 33.33% (claim 13714) and then to 25%, with one of five board seats (claims 12872, 12874, 12920).

The revenue arithmetic cannot be reconciled from this record because no revenue, hashrate-to-revenue or bitcoin-production figures are disclosed. One volume check is possible. The 58,000 miners contracted for 2022 delivery (claims 14247, 14123) compare with 20,000 committed to Nautilus (claim 13713), a commitment later "right-sized" (claim 12920). The fixed cash items on record are the $19.072 million JV investment in Q1 2022 (claim 13918), the $8.46 million annual Beowulf fee (claim 12867) and the $339,200 transition payment (claim 12942). They describe cost and capital outlay, not contracted income, so any diligence view of top-line durability rests on bitcoin price and network economics rather than on signed customer agreements.

TeraWulf's record contains almost no customer agreements in the conventional sense, because the company is a self-miner whose revenue comes mainly from bitcoin it mines itself (claim 13441). The named counterparties are suppliers, service providers, joint-venture partners and related parties, not buyers of a product. Most of them are documented as executed contracts rather than MOU or LOI language. Minerva is covered by an executed equipment purchase agreement (14028), and Bitmain by four sales and purchase agreements (14051, 14027, 13977). Those Bitmain agreements are expressly non-fixed price, so the 'competitive' pricing framing (14246) is softer than it sounds. Foundry (13335), NYDIG custody (13412, 13344) and NYDIG Execution (13414, 13345) are dated, signed agreements. The Talen joint venture is dated May 13, 2021 (14120), and the Nautilus power arrangement carries a defined five-year term and a fixed $0.02 price (14113, 12838). The vaguer items are the November 2021 statement that unnamed 'competitive power supply agreements' were 'in place' (21144) and the later description of Bitmain as a 'strategic relationship' supporting 'preferred access' (12891), which is characterisation rather than a contract. Claim 12852 mentions 'hosting arrangements' but names no hosting customer, which is the one place an unidentified revenue counterparty appears. Claim 21075 is a 2006 IKONICS auditor ratification inherited from the predecessor shell and has no bearing here. Concentration has shifted in a telling direction. In March 2022 the Pennsylvania build depended heavily on Talen, and TeraWulf expected 150 MW of a 300 MW venture (14112). By August 2022 its Nautilus stake was set at 33.33% against Cumulus's 66.67%, with 20,000 miners contributed (13714, 13713). By March 2023 the stake had fallen to 25%, with one of five board seats (12872, 12874). Related-party exposure runs through Beowulf E&D, which is controlled by the CEO (12845, 12942), and through the affiliate NovaWulf (13849). The revenue arithmetic cannot be reconciled from this record. The 58,000-miner delivery commitment (14247) is introduced with 'despite securing' (14123), which suggests delivery shortfalls. The only quantified flow is a $19,072 joint-venture investment outflow (13918), and no revenue figure is disclosed that would tie mined output to the contracted fleet. The claim list is also cut off at 12841, so later disclosures may change this read.

TeraWulf's record contains no conventional customer agreements. The company describes itself as a self-miner whose revenue comes mainly from its own mined bitcoin (13441). Nearly every named counterparty is therefore a supplier, service provider, financing agent, related party or joint-venture partner, not a buyer of its output. Claim 21075 is an IKONICS auditor ratification from 2006 that predates the merger, and it should be disregarded. The record supplied also ends partway through, at claim 12922, so later entries could not be reviewed.

The signed, named agreements are concrete. They are the Talen joint venture dated May 13, 2021 (14120), the NYDIG custodial agreement of March 10, 2022 (13412, 13344), the NYDIG Execution agreement of September 16, 2022 (13414), a DACA with NYDIG and Wilmington Trust (12905), the Foundry pool service agreement (13335), a Minerva purchase agreement (14028), the Cantor, B. Riley and D.A. Davidson sales agreement (13973), and the NYPA power agreement (12842). Softer language deserves a discount. The November 2021 'competitive power supply agreements already in place' named no counterparty (21144). The four Bitmain agreements are explicitly 'non-fixed price' (14051, 13977), which weakens the 58,000-miner delivery figure (14247, 14123). The later Bitmain 'strategic relationship' promising 'preferred access' is relational framing, not a contract term (12891). NYPA's 90 MW carries only a 'potential to expand' (12842).

Concentration has moved toward Talen and toward insiders. TeraWulf's attributable Nautilus share fell from 150 of 300 MW (14112) to 33.33% (13714) and then to 25%, with one of five board seats (12872, 12874, 12920). The miner commitment fell to 20,000 (13713). Related parties are a material cost line. Beowulf E&D receives an $8.46 million annual base fee (12867) and $339,200 for transition services (12942), and NovaWulf is disclosed as an affiliate (13849). Operationally, custody, execution and pool access all run through only two firms, NYDIG and Foundry.

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-to-revenue or per-counterparty figures are disclosed. The only cash data points are the $19,072 thousand Q1 2022 joint-venture investment (13918) and the Beowulf fees. Nautilus bitcoin is distributed in proportion to hash contribution, net of power and operating costs (12939). A reader therefore cannot test announced capacity against reported results without the omitted filings.

TeraWulf's record through early 2023 contains no signed customer agreements in the conventional sense, because the company sells nothing to identified buyers. Its revenue is self-mined bitcoin, routed through Foundry Digital's pool (claims 13440, 13335) and custodied or sold through NYDIG and NYDIG Execution (claims 13412, 13414, 13333, 13334). These are executed, dated service contracts, but they are vendor relationships, not customers. The same holds for the equipment side. Bitmain's four non-fixed-price purchase agreements and the Minerva purchase agreement are executed (claims 14027, 14028, 13977), though non-fixed pricing weakens the 58,000-miner delivery figure (claims 14247, 14123) as a firm commitment. The 2023 description of a 'strategic relationship' with Bitmain offering 'preferred access' (claim 12891) is promotional language with no separate contract behind it. Claim 21075 concerns IKONICS's 2006 auditor ratification and should be excluded as legacy shell-company noise.

On naming, the counterparties are unusually well identified. Talen, Susquehanna, NYPA, Foundry, NYDIG, Bitmain, Minerva and the related-party Beowulf E&D are all named. The one early exception is the November 2021 reference to 'competitive power supply agreements already in place' (claim 21144), which named no supplier until the NYPA 90 MW agreement surfaced in 2023 (claims 12842, 12890). The NYPA agreement's 'potential to expand' language is optionality, not contracted capacity.

Concentration has shifted toward Talen and toward related parties. The Nautilus joint venture moved from a 33.33% TeraWulf stake (claim 13714) to 25% with one of five board seats (claims 12872, 12874, 12920). That leaves TeraWulf a minority economic and governance position in its nuclear-powered site. Beowulf E&D, an insider-affiliated operator, draws a reduced $8.46 million annual base fee (claim 12867) plus a $339,200 Nautilus transition payment (claim 12942). NovaWulf is likewise disclosed as an affiliate (claim 13849).

Revenue arithmetic cannot be reconciled from this record. No contracted revenue, pricing, or hash-rate-to-revenue figures are disclosed. The only quantified flows are outflows: a $19.1 million joint-venture investment (claim 13918) and the Beowulf fees. Revenue depends on bitcoin price, network difficulty and actual miner deliveries, none of which the announced agreements fix. The claim set supplied was truncated after claim 12922, so later developments, including any hosting or HPC customer contracts, are not assessed here.

TeraWulf's record through early 2023 contains almost no customer agreements in the conventional sense. Its disclosed revenue is self-mined bitcoin, and the named counterparties are suppliers, service providers and a joint-venture partner rather than buyers. Most of these arrangements are described as executed contracts, not MOU or framework language. Foundry Digital operates the mining pool under a Service Agreement (claims 13335, 13440). NYDIG holds the bitcoin under a Digital Asset Custodial Agreement dated March 10, 2022 (13412, 13344), and NYDIG Execution handles sales under a September 16, 2022 Execution Agreement (13414, 13345). Bitmain and Minerva supply equipment under executed purchase agreements (13977, 14028), although the four Bitmain contracts are expressly non-fixed price (14051). NYPA supplies an initial 90 MW to Lake Mariner (12842). The earliest power claim names no one, referring only to 'competitive power supply agreements already in place' (21144). Hosting is mentioned in passing, as 'bitcoin earned from hosting arrangements' (12852), but no hosting customer is identified. Claim 21075 relates to the IKONICS predecessor's choice of auditor and has no commercial relevance.

Counterparty concentration has become heavier around Talen, and TeraWulf's economic share has shrunk. The 2021 joint venture was announced as up to 300 MW with 150 MW attributable to TeraWulf (14112). By August 2022 TeraWulf's equity interest had been reset to 33.33% (13714), and by March 2023 to 25%, with one of five board seats (12872, 12874). Cumulus also holds a unilateral option to add 100 MW, funded solely by Cumulus (12922). Related-party exposure is significant as well. Beowulf E&D, which is controlled by the CEO, provides infrastructure and administrative services (12845, 13718) and a $339,200 transition arrangement (12942). NovaWulf is disclosed as an affiliate (13849).

The revenue arithmetic cannot be reconciled from this record. No revenue figure, hashrate output or bitcoin-produced total is disclosed against which the deals could be tested. The only hard numbers are a $19,072 thousand joint-venture investment (13918) and the miner volumes. Of the 58,000 miners contracted for 2022 delivery (14247, 14123), 20,000 were earmarked for Nautilus (13713). Whether the balance was actually delivered and energized is not shown. The claim list provided was truncated at claim 12838, so later disclosures may change this picture.

The striking feature of TeraWulf's record is that it contains no signed customer agreement in the conventional sense. Every named counterparty is a supplier, service provider, financing agent, joint-venture partner or related party, not a buyer of TeraWulf's output. The one outlier, claim 21075, is an IKONICS auditor ratification from 2006 that reflects the reverse-merger shell's history. It has no bearing on TeraWulf's commercial book. The named and executed relationships are concrete. They include equipment purchase agreements with Bitmain and Minerva covering 58,000 miners (claims 14246, 14247, 14028, 14027, 13977), a joint venture with a Talen Energy affiliate (claims 14120, 14112), and a Susquehanna nuclear power supply with a five-year term and renewal options (claims 14113, 14129). They also include an NYPA agreement for an initial 90 MW (claims 12842, 12890), a dated NYDIG custody agreement and Execution Agreement (claims 13412, 13414, 13345), and a Foundry pool service agreement (claim 13335). Softer language does appear. The NYPA deal carries only the 'potential to expand' (claim 12842), and the 2021 'power supply agreements already in place' (claim 21144) names no counterparty. The 2023 'strategic relationship' with Bitmain promising 'preferred' terms (claim 12891) describes a relationship rather than a contract. Unidentified 'hosting arrangements' surface once, in claim 12852, without customers or terms being disclosed. Counterparty concentration has run in the opposite direction from dilution. TeraWulf's stake in Nautilus fell from 33.33% (claim 13714) to 25%, with Cumulus holding 75% and four of five board seats (claims 12872, 12874, 12921). That shift deepens dependence on Talen while reducing control. Related-party exposure also persists through Beowulf E&D, which is owned by the CEO and provides infrastructure and transition services, including a $339,200 transition fee (claims 13718, 12845, 12942). NovaWulf is another disclosed affiliate (claim 13849). Revenue is effectively single-channel. Self-mined bitcoin flows through one pool, Foundry, into one custodian and one execution venue, all NYDIG (claims 13440, 13441, 13442). On revenue arithmetic, there is nothing to reconcile against customer deals because none are announced. The only disclosed figure is a $19,072 thousand joint-venture investment in Q1 2022 (claim 13918), which is a cash outflow rather than revenue. Nautilus distributions are tied to hash-rate contribution net of power and operating costs (claim 12939). Given that formula, TeraWulf's economics there should scale with miners actually contributed, initially 20,000 (claim 13713), rather than with headline JV capacity. Diligence should test the reported revenue against that formula and against the unnamed hosting income.

TeraWulf's record contains no customer agreements in the conventional sense, and no MOU, LOI, framework or pilot language either. As a self-mining bitcoin operator, its revenue comes from pool rewards rather than contracted offtake. The disclosed counterparties are suppliers, service providers and a joint-venture partner, and almost all are named and described as executed. The one exception is the November 2021 statement that 'competitive power supply agreements' were 'already in place' in New York and Pennsylvania without naming any utility or generator (claim 21144). That gap closes for Pennsylvania once the Susquehanna Station supply is specified, first as a five-year term with renewal options (claims 14113, 14129) and later as a fixed price of about 2.0 cents/kWh (claims 12838, 12859). The New York supplier is never identified in this record. Equipment counterparties are named and progressively firmed up. Bitmain and Minerva appear first as 'secured' supply arrangements (claim 14246), then as an executed Minerva purchase agreement and four non-fixed-price Bitmain contracts (claims 14028, 14051, 13977), and finally with five additional Bitmain future purchase agreements (claim 12847). Revenue-side intermediaries are also named with dated contracts: Foundry runs the mining pool (claims 13440, 13335), NYDIG provides custody under a March 10, 2022 agreement (claims 13412, 13333), and NYDIG Execution handles sales under a September 16, 2022 agreement (claims 13414, 13345). Claim 21075 is a legacy IKONICS auditor ratification from the predecessor shell and has no commercial bearing. Concentration has moved in a dilutive direction. The Talen joint venture was first framed as 150 of 300 MW attributable to TeraWulf (claim 14112), effectively half. It was cut to a 33.33% equity interest in August 2022 (claim 13714) and then to 25% (claims 12872, 12860), with TeraWulf holding one of five board seats (claim 12874). Dependence on Foundry, NYDIG and Bitmain remains single-name at each layer of the stack. The revenue arithmetic cannot be reconciled from this record. The 58,000-miner delivery commitment (claims 14247, 14123) is introduced with 'despite securing', which implies a shortfall, and only a 20,000-miner contribution to Nautilus is quantified (claim 13713). The figures that are disclosed, $19.1 million of joint-venture investment (claim 13918) and a $339,200 transition-services fee (claim 12942), are cost-side data. No hashrate-to-revenue bridge is provided against them. The claim list provided ends mid-entry at claim 12854, so later disclosures may change this read.

TeraWulf's record through March 2023 contains almost no customer agreements in the conventional sense. The company is a self-mining bitcoin operator, and it says itself that the majority of its revenue comes from self-mined bitcoin (claim 13441). What the record documents is a set of signed supplier, service, and joint-venture contracts rather than offtake or customer contracts. That distinction matters for any reader expecting a customer book. Claim 21075 should be set aside entirely. It is a 2006 auditor ratification by IKONICS, the predecessor public shell, and it has no bearing on TeraWulf's commercial position.

The disclosed counterparties are unusually well identified, and most of the language describes executed agreements rather than MOUs or discussions. Bitmain and Minerva are named as equipment suppliers under executed purchase agreements (claims 14246, 14028, 13977, 12847). NYDIG holds custody under a dated March 10, 2022 agreement (claims 13412, 13333), and NYDIG Execution handles sales under a September 16, 2022 agreement (claims 13414, 13345). Foundry operates the mining pool under a service agreement (claims 13440, 13335). Talen, through its subsidiary Cumulus Coin, is the joint-venture partner at Nautilus (claims 14120, 12841). The one vague item is the November 2021 reference to "competitive power supply agreements already in place" in New York and Pennsylvania, which names no counterparty (claim 21144). The Susquehanna-sourced power at roughly 2.0 cents/kWh is described more concretely (claims 14113, 12838, 12859). The 58,000-miner delivery figure is a contractual delivery schedule, not confirmed receipt (claims 14247, 14123).

The most material change in concentration is TeraWulf's shrinking share of Nautilus. The partnership began as a 300 MW project with 150 MW attributable to TeraWulf (claim 14112). It was then restructured to a 33.33% stake with a 20,000-miner contribution (claims 13714, 13713), and later to 25% (claims 12872, 12860, 12840). Governance sits with Talen, which appoints four of the five managers (claim 12874). Operationally, the company relies heavily on single providers, with Foundry for hashrate monetization and NYDIG for custody and liquidation.

The revenue arithmetic cannot be reconciled from this record. No revenue totals, hashrate-to-revenue bridges, or per-counterparty figures are disclosed. The only quantified items are $19,072 (presumably thousands) of joint-venture investment (claim 13918) and a $339,200 transition-services fee (claim 12942). Nautilus economics flow back as biweekly bitcoin distributions net of power and operating costs (claim 12939), not as contracted revenue. The record appears truncated at claim 12854 and does not cover later periods, so no conclusions about subsequent customer contracts should be drawn from it.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, and also no MOU, LOI or discussion-stage language. The disclosed agreements are executed supplier, financing, service and joint-venture contracts. That reflects the self-mining model: revenue comes from bitcoin earned through a pool operated by Foundry Digital (13440, 13335, 13162). The counterparties are vendors and service providers, not purchasers of a contracted product. Claim 21075, an IKONICS auditor ratification from 2006, comes from the pre-merger shell and has no commercial significance. The equipment commitments name their counterparties. Bitmain is named under four non-fixed-price agreements, later joined by five future sales agreements, and Minerva under an executed purchase agreement (14246, 14051, 14028, 13977, 12847). Custody and execution go to NYDIG and NYDIG Execution, with dated agreements of March 10 and September 16, 2022 (13412, 13414). The one material unnamed item is the 'competitive power supply agreements already in place' for New York and Pennsylvania (21144). The Pennsylvania supply was later identified as Susquehanna Station nuclear power at a fixed 2.0 cents/kWh over a five-year term with renewal options (14113, 12838, 12859). The New York counterparty is never named in this record. Counterparty concentration has narrowed around Talen Energy. The Nautilus joint venture began as a 300 MW partnership with 150 MW attributable to TeraWulf (14112, 14120). The August 2022 restatement cut the stake to 33.33% (13714). A later disclosure put it at 25% against Cumulus Coin's 75%, with one of five board seats (12872, 12860, 12874, 12841). TeraWulf's operating exposure at Nautilus therefore became more dependent on a single partner while its governance rights shrank. Several service relationships sit inside the issuer's affiliate orbit, notably NovaWulf (13849) and Beowulf E&D, which provides build-out and operating services and received a $339,200 transition fee (13718, 12942). The revenue arithmetic cannot be reconciled from these claims, because no revenue figures are disclosed. Only the $19,072 joint-venture outflow appears (13918). The 58,000-miner delivery commitment (14247, 14123) is not shown to tie to energized capacity, and the record does not say how the 20,000 miners contributed to Nautilus (13713) fit within it. The record ends in March 2023, so it reflects none of TeraWulf's later hosting or HPC tenant contracts.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, because the business it describes has no customers. The company states that the majority of its revenue comes from self-mined bitcoin (claim 13441), so the counterparties disclosed are suppliers, service providers and a joint-venture partner rather than buyers of a product. The oldest entry, an auditor ratification for IKONICS Corporation (claim 21075), is a legacy shell-company item and has no commercial relevance.

On naming, the record is largely specific and executed. Equipment supply is attributed to named vendors, Bitmain and MinerVA, through executed purchase agreements (claims 14246, 14051, 14028, 13977, 12847). Pool services from Foundry Digital (claims 13440, 13335) and custody and execution from NYDIG (claims 13412, 13414, 13333, 13345) are dated, signed contracts. The Nautilus joint venture is with Cumulus Coin, an identified Talen subsidiary (claims 14120, 12841, 12848). The main exception is the November 2021 reference to competitive power supply agreements already in place for the New York and Pennsylvania sites, which names no counterparty (claim 21144). Nautilus power was later tied to Susquehanna Station at a fixed rate of about $0.02/kWh (claims 14113, 12838, 12859). The record contains no framework, MOU, LOI or pilot language, and no unnamed anchor customers.

Concentration moved toward Talen, not away from it. TeraWulf's attributable share of Nautilus began at 150 of 300 MW (claim 14112). It fell to 33.33% in August 2022 (claim 13714) and to 25% after the restructuring, with Talen's side holding four of five board seats (claims 12872, 12860, 12874). Upstream, the record shows dependence on one pool operator and one custodian and execution venue, NYDIG. The 58,000-miner delivery commitment (claims 14247, 14123) narrowed into a 20,000-miner contribution to Nautilus (claim 13713).

The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed. The only cash-flow data points are a $19.1 million joint-venture investment in Q1 2022 (claim 13918) and a $339,200 transition-services payment to the affiliate Beowulf E&D (claim 12942). Related-party exposure through Beowulf E&D and NovaWulf (claims 13718, 13849) deserves scrutiny. The record is also truncated at claim 12854 in March 2023, so any later hosting or compute customer agreements fall outside this read.

TeraWulf's record contains no customer agreements in the conventional sense, and readers should not expect one. The company's revenue comes from bitcoin it mines itself, so its named counterparties are suppliers, a joint-venture partner, a mining pool, a custodian and financing agents rather than buyers. Claim 21075 is a legacy IKONICS auditor ratification from the pre-merger shell and has no bearing on the current business. Most of the named relationships are signed contracts. The Bitmain and Minerva equipment agreements (14051, 14028, 13977, 12847) are executed purchase contracts, although the Bitmain contracts are described as non-fixed price and later as 'future' sales agreements, so their volume and cost were not locked. The NYDIG custodial agreement dated March 10, 2022 and the NYDIG Execution agreement dated September 16, 2022 are both dated and signed (13412, 13414, 13333). The Foundry relationship is a service agreement formed by using the pool (13335), which is a thinner commitment than a negotiated contract. The one counterparty left unidentified is the supplier behind the New York 'competitive power supply agreements' (21144); the claims never name it, while the Nautilus power arrangement is tied explicitly to Susquehanna Station at about 2.0 cents/kWh (14113, 12838, 12859). There is no MOU, LOI or 'anchor customer' language anywhere in the record. Concentration in the joint venture has moved steadily away from TeraWulf. Its share of Nautilus was about half of 300 MW in early 2022 (14112), fell to 33.33% after the August 2022 amendment (13714), and then to 25%, subject to capital true-ups (12872, 12860). TeraWulf appoints one of five managers (12874), which gives it limited control over a co-owned asset. Operationally, the company relies on a single pool operator, Foundry, and a single custodian and execution venue, NYDIG (13440, 13441, 13442). The disclosed figures do not let anyone reconcile revenue with the announced deals. The 58,000-miner delivery schedule (14247, 14123) is introduced with the word 'despite', which signals a delivery shortfall. The 20,000 miners contributed to Nautilus (13713) are not tied to any hashrate or revenue figure in the claims. The only dollar amounts on record are a $19.07 million joint-venture outlay in Q1 2022 (13918) and a $339,200 transition-services fee owed to Beowulf E&D (12942). Neither amount maps to mined-bitcoin revenue, and the record is truncated after claim 12854.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, because a self-mining bitcoin operator sells its output into the network rather than to named buyers. Nearly every counterparty on file is a supplier, venture partner, or service provider. This record also ends before any later hosting or compute contracts, so it says nothing about them. Claim 21075 refers to IKONICS, the legacy shell, ratifying an auditor in 2006, and should be excluded from any commercial read.

The agreements themselves are concrete and executed rather than MOU or LOI language. They include the Minerva purchase agreement and the four non-fixed-price Bitmain agreements (13977, 14051, 14028), later joined by five future Bitmain agreements (12847). They also include the Talen joint venture dated May 13, 2021 and amended and restated on August 27, 2022 (14120, 12848), the Susquehanna-connected power supply at about 2.0 cents/kWh (14129, 12859), the NYDIG custody and execution agreements (13412, 13414), and Foundry pool participation (13335). The main exception is the New York power arrangement. Claim 21144 cites "competitive power supply agreements already in place" without naming a counterparty, and that is the only material unidentified relationship. Two other relationships are related parties rather than arm's-length deals: Beowulf E&D, under a reduced $8.46 million annual base fee (12867) and a $339,200 transition services agreement (12942), and NovaWulf (13849).

Concentration has tightened over time, and TeraWulf's control has weakened. The Talen venture was first framed as up to 300 MW with 150 MW attributable to TeraWulf (14112). It was then cut to a 33.33% stake (13714) and later to 25%, with Talen appointing four of five managers (12872, 12874, 12860). The record has an internal tension here. The August 2022 disclosure states 33.33%, while March 2023 filings describe the same amended agreement as producing 25%, "subject to adjustment based on relative capital contributions." Readers should treat the stake as variable. Execution flows run through one pool operator and one custodian-broker, NYDIG.

The revenue arithmetic cannot be reconciled from this record, because no revenue, hashrate-realized, or bitcoin-produced figures are disclosed. The investable inputs also fail to tie cleanly. The company announced 58,000 miners for 2022 delivery (14247, 14123) but contributed only 20,000 to Nautilus (13713), and the single disclosed cash figure, a $19.1 million joint venture investment in Q1 2022 (13918), has no matching output metric. Diligence should request realized deliveries, energized capacity by site, and the Nautilus distribution history implied by claim 12939.

TeraWulf's record through early 2023 contains no conventional customer contracts. As a self-mining bitcoin operator, its revenue comes from block rewards earned through a pool, so nearly every named counterparty is a supplier, service provider, JV partner or affiliate rather than a buyer. Claim 21075, the 2006 auditor ratification, belongs to the IKONICS legacy shell and is irrelevant to commercial diligence. The executed, named agreements are substantial and specific: equipment purchase agreements with Bitmain and Minerva (claims 14027, 14028, 13977), the Foundry pool Service Agreement (claim 13335), the NYDIG Custodial Agreement of March 10, 2022 (claim 13412), the NYDIG Execution Agreement of September 16, 2022 (claim 13414), and the Talen joint venture dated May 13, 2021 (claim 14120). None of these is framed as an MOU, LOI or pilot. The softer language sits elsewhere. The November 2021 statement that 'competitive power supply agreements' were 'already in place' for the New York and Pennsylvania sites names no New York counterparty (claim 21144). The Bitmain contracts are 'non-fixed price' (claim 14051), which leaves the 58,000-miner delivery figure (claims 14247, 14123) exposed to repricing. The concentration story is one of dilution. TeraWulf's economic exposure to Nautilus fell from 150 of 300 MW (claim 14112), to a 33.33% stake with a 20,000-miner contribution (claims 13714, 13713), to 25% with one of five board seats (claims 12872, 12874, 12920). Cumulus/Talen now controls governance and holds the expansion option (claim 12922). Related-party dependence also deserves attention. Beowulf E&D, an affiliated services provider, collects an $8.46 million annual base fee (claim 12867) plus a $339,200 transition fee from Nautilus (claim 12942), and NovaWulf is disclosed as an affiliate (claim 13849). Revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed, and the only quantified cash item is a $19.072 million joint-venture investment in Q1 2022 (claim 13918). Hash-rate-based bitcoin distributions from Nautilus (claim 12939) depend on miner deployment that has been cut sharply from the original 58,000-unit plan. The claim list provided ends mid-entry at claim 12860, so any later deals, including hosting or HPC tenancy, could not be assessed here.

TeraWulf's record contains no conventional customer contracts, and that is the central diligence point. As a self-mining bitcoin operator, its revenue comes from pool payouts. Foundry Digital operates the pool (claims 13440, 13335, 12854), NYDIG custodies the bitcoin (13412, 13333), and NYDIG Execution handles sales (13414, 13345). Each is a signed, dated service agreement with a named counterparty. None is a buyer of TeraWulf's output in the commercial sense, so the usual signed-customer versus MOU split applies here to suppliers, power and joint-venture partners. Claim 21075 relates to the IKONICS predecessor's 2006 auditor ratification and should be excluded as not relevant.

On naming, the record is unusually concrete. Every material counterparty is identified: Talen and its subsidiary Cumulus Coin (14120, 12841), Bitmain and Minerva (14246, 13977), Susquehanna Station as the power source (14129), and Cantor, B. Riley and D.A. Davidson on the sales agreement (13973). No unnamed 'strategic partners' or 'anchor customers' appear. The softer language sits elsewhere. Talen capacity is 'up to 300 MW' (14112). The Bitmain contracts are 'non-fixed price' (14051, 14027). Equity stakes are 'subject to adjustment based on relative capital contributions' (12872). The 2021 power claim (21144) asserts agreements 'in place' without naming a supplier.

Concentration in the flagship Nautilus venture has moved steadily against TeraWulf. The original framing attributed 150 MW of 300 MW to TeraWulf, which is 50% (14112). The August 2022 amendment cut that to 33.33% with a 20,000-miner contribution (13714, 13713). By March 2023 it was 25%, with one of five board seats (12872, 12874), described as 'right-sizing' (12860). Talen's control of the venture therefore rose as TeraWulf's economic share fell. Equipment supply stayed concentrated in Bitmain, which grew from four to nine agreements (12847).

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-to-revenue or bitcoin-production figures are disclosed. The 58,000-miner delivery commitment (14247, 14123) sits alongside only 20,000 miners contributed to Nautilus, and the record does not account for the remainder's deployment. The quantified items are all cost outflows: $19.1 million of joint-venture investment (13918), an $8.46 million annual Beowulf E&D fee (12867), and a $339,200 transition fee (12942). The record is truncated at claim 12905, so later disclosures may close these gaps.

The first thing to understand about TeraWulf's record is that it contains no customer agreements at all. Every counterparty on file is a supplier, joint-venture partner, service provider, financial agent or related party. That fits a self-mining model in which, as the company says, "the majority of our revenue comes from our self-mined bitcoin" (claim 13441). The record also does not use MOU, LOI or 'discussions' language. Its ambiguity comes from a different source: vague early framing that later filings resolve into named, executed contracts. The November 2021 reference to "competitive power supply agreements already in place" (claim 21144) names no counterparty. Neither does the unnamed Pennsylvania joint venture (claim 14141). Both are later resolved as Talen and its subsidiary Cumulus Coin (claims 14112, 14120, 12841), with power sourced from Susquehanna Station (claims 14113, 14129) at roughly 2.0 cents/kWh (claim 12859). The equipment agreements are executed and named, with Bitmain and Minerva (claims 14027, 14028, 13977, 12847). However, the Bitmain contracts are "non-fixed price" (claim 14051), so they set out a supply framework rather than a fixed cost commitment. The 58,000-miner figure is introduced with "despite securing" (claim 14123), which implies delivery fell short of contract. Counterparty concentration has deepened rather than diversified. Talen is the dominant relationship, but TeraWulf's position in it has weakened: its Nautilus stake fell from 33.33% (claim 13714) to 25% (claims 12872, 12860), with one of five board seats (claim 12874). Operating dependence has consolidated around Foundry for pool services (claims 13335, 12854) and NYDIG for both custody and execution (claims 13412, 13414). Related-party exposure is material. Beowulf E&D draws an $8.46 million annual base fee (claim 12867) plus a $339,200 transition payment from Nautilus (claim 12942), and NovaWulf is a disclosed affiliate (claim 13849). Claim 21075 refers to the 2006 IKONICS auditor ratification, a legacy item from the predecessor shell that is irrelevant to this analysis. Revenue arithmetic cannot be reconciled from the record. No contracted revenue exists to test, and the only figure disclosed is a $19,072 thousand joint-venture investment outflow (claim 13918). The record is also truncated at claim 12905, so any agreements disclosed after March 2023 are outside this read.

TeraWulf's record through March 2023 contains no revenue-paying customer agreements at all, so the usual split between signed customer contracts and MOU or pilot language does not really apply. Every named counterparty is a supplier, joint-venture partner, service provider or affiliate. Revenue is self-mined bitcoin routed through the Foundry Digital pool (13440, 13335, 12854), held in custody by NYDIG (13412, 13344) and sold through NYDIG Execution (13414, 13345). Those three are dated, executed service agreements, but they are infrastructure, not demand. Claim 21075 is an IKONICS-era auditor ratification from the predecessor shell and should be disregarded. The firmest named relationships are on the supply side. The Talen joint venture is executed and dated (14120, 12848), with Cumulus Coin identified as the Talen subsidiary (12841). The Minerva purchase agreement is stated as executed (14028). The Bitmain contracts are signed but explicitly "non-fixed price" (14051, 13977), which leaves the economics open. The one early unidentified-counterparty claim is the November 2021 reference to "competitive power supply agreements already in place" (21144). That was later resolved to the Susquehanna-connected nuclear supply at Nautilus at roughly 2.0 cents/kWh (14113, 12859). Related-party concentration deserves attention. Beowulf E&D provides operations and administrative services (13718) for an $8.46 million annual base fee (12867) plus a $339,200 transition arrangement (12942), and NovaWulf is disclosed as an affiliate (13849). Concentration in the flagship asset has moved against TeraWulf. Its Nautilus share was initially framed as 150 of up to 300 MW (14112), then 33.33% (13714), then 25% with one of five board seats (12872, 12874, 12860). The record is internally inconsistent here: 13714 dates the 33.33% split to an August 2022 filing, while 12860 says the August 2022 amendment produced 25%. The miner numbers do not tie either. Supply agreements covered 58,000 machines for 2022 (14247, 14123), yet the Nautilus contribution was set at 20,000 (13713), and the filings do not reconcile the remainder. On revenue arithmetic, there are no customer deals to reconcile against. The only quantified flow in this record is a $19,072 thousand joint-venture investment in Q1 2022 (13918), and Nautilus bitcoin distributions are tied to hash-rate share net of power and operating costs (12939). Revenue therefore depends on hash rate, bitcoin price and power cost, not on any contracted counterparty. The claim set is truncated at 12905, so later developments are outside this read.

TeraWulf's record through early 2023 contains no signed customer agreements in the conventional sense. The company's revenue is self-mined bitcoin, earned through a mining pool operated by Foundry Digital (claims 13440, 13335, 12854), so Foundry is a service provider rather than a buyer. The disclosed agreements are executed contracts, not MOUs or LOIs, but they are almost entirely supply-side. These include equipment purchase agreements with Bitmain and MinerVA (claims 14246, 14051, 13977, 12847), a custodial agreement with NYDIG dated March 10, 2022 (claims 13412, 13161), an execution agreement with NYDIG Execution dated September 16, 2022 (claims 13414, 13345), and an ATM sales agreement with Cantor, B. Riley and D.A. Davidson (claim 13973). The earliest item, claim 21075, is an IKONICS auditor ratification from 2006 that predates the reverse merger. It should be excluded as noise.

On naming, the record is unusually specific. There are no unidentified 'anchor customers' or 'strategic partners.' The one soft formulation is the November 2021 reference to 'competitive power supply agreements already in place' (claim 21144), which named no counterparty. That gap was later filled by the Susquehanna-sourced nuclear supply at roughly 2.0 cents/kWh (claims 14113, 14129, 12859).

Concentration is the more material issue, and it is structural rather than commercial. Talen, through its subsidiary Cumulus Coin, is the dominant counterparty. TeraWulf's Nautilus interest moved from a 50% framing (150 of 300 MW, claim 14112) to 33.33% in August 2022 (claim 13714). It then moved to 25%, with one of five board seats (claims 12872, 12874, 12860). That is a meaningful dilution of control over the Pennsylvania asset. Related-party exposure is also significant. Beowulf E&D, which is affiliated with management, receives an $8.46 million annual base fee (claim 12867) and a $339,200 transition payment from Nautilus (claim 12942), and NovaWulf is disclosed as an affiliate (claim 13849).

Revenue arithmetic cannot be reconciled from this record. No hashrate-to-revenue figures, bitcoin production totals or recognized revenue are cited. The only quantitative anchors are the 58,000-miner delivery commitment (claims 14247, 14123), the 20,000-miner Nautilus contribution (claim 13713), and $19,072 thousand of Q1 2022 joint-venture investment (claim 13918). Revenue from Nautilus arrives as bitcoin distributions netted against power and operating costs (claim 12939), so gross deal values do not map onto reported revenue. The input is also truncated at claim 12905, so later disclosures, including any HPC or hosting customer announcements, are not assessed here.

TeraWulf's record contains almost no customer agreements in the conventional sense, because during the period covered the company was a self-miner whose revenue came from bitcoin earned through a pool rather than from paying counterparties. Nearly every disclosed relationship is an executed supplier, power, joint-venture or service contract, and the record contains no MOU, LOI, framework or 'discussions' language. The one outlier, claim 21075, is a 2006 auditor ratification inherited from the IKONICS predecessor entity and has no commercial bearing.

Named counterparties are specific throughout. The equipment agreements identify Bitmain and Minerva and describe them as executed: four non-fixed-price purchase agreements with Bitmain plus a Minerva purchase agreement (claims 14051, 14028, 13977), later expanded to include five Bitmain future sales agreements (claim 12847). The 58,000-miner delivery figure (claims 14247, 14123) is a supplier commitment, not demand. The Talen joint venture is named and dated (claims 14112, 14120, 12848, 12841). The Susquehanna power supply carries a five-year term and a fixed rate (claims 14113, 12838). Foundry, NYDIG and NYDIG Execution are identified with agreement dates (claims 13335, 13333, 13414). The only vague language is the November 2021 reference to 'competitive power supply agreements already in place' (claim 21144), which predates the named Susquehanna and NYPA disclosures (claims 14129, 12842).

Concentration has narrowed in TeraWulf's disfavor at Nautilus. Its equity interest fell from 50% of a 300 MW plan (claim 14112) to 33.33% (claim 13714) and then to 25% (claims 12872, 12840), and it appoints one of five managers (claim 12874). Talen's Cumulus also held a unilateral 100 MW expansion option (claim 12922). Operational dependence is concentrated as well: Foundry handles all pool revenue, and NYDIG handles both custody and execution (claims 13440, 13441, 13442).

Revenue arithmetic cannot be reconciled from this record. No revenue line, hashrate-to-revenue bridge or contract value is disclosed. The only figures are the $19,072 joint-venture outflow, which is presented in thousands (claim 13918), and the $339,200 Beowulf transition-services fee (claim 12942). Both are costs or related-party payments, not customer revenue. The Beowulf and NovaWulf affiliate links (claims 13718, 13849) deserve related-party scrutiny. The record is truncated after claim 12842, so any later hosting or HPC customer contracts fall outside this read and should be assessed separately against signed-agreement evidence.

TeraWulf's record through early 2023 contains no signed customer agreements in the conventional sense. As a self-mining bitcoin operator, its revenue comes from pool participation rather than contracted offtake, and every named counterparty on file is a supplier, service provider, joint-venture partner or affiliate. Claim 21075, an IKONICS auditor ratification from 2006, is a legacy-shell artifact from the reverse merger and has no bearing on TeraWulf's commercial position.

The named relationships are documented as executed agreements, not MOU or framework language. Bitmain and Minerva appear as equipment suppliers under four non-fixed-price purchase agreements and one purchase agreement (claims 14246, 14051, 14028, 13977). Talen Energy is the joint-venture partner in Nautilus (claims 14120, 14112, 12848). Foundry Digital operates the mining pool (claims 13440, 13335), and NYDIG serves as custodian and execution venue under dated agreements (claims 13412, 13414, 12905). The main exception is claim 21144, which cites 'competitive power supply agreements already in place' for New York and Pennsylvania without naming a counterparty. Only the Nautilus nuclear supply from Susquehanna Station is later specified, at a fixed $0.02 rate (claims 14113, 12838). Beowulf E&D and NovaWulf are related parties, not arm's-length counterparties (claims 13849, 12867).

Concentration has deepened rather than diversified. The Talen venture began as up to 300 MW with 150 MW attributable to TeraWulf (claim 14112). The August 2022 amendment cut TeraWulf's stake to 33.33% (claim 13714), and the March 2023 restatement cut it again to 25% with one of five board seats (claims 12872, 12874). The headline commitment of 58,000 miners (claims 14247, 14123) contrasts with only 20,000 contributed to Nautilus (claim 13713). Pool revenue routes entirely through Foundry, and custody and sales route entirely through NYDIG, so the company depends on a single counterparty at each stage.

The revenue arithmetic cannot be reconciled from this record, because no revenue figure is disclosed alongside the agreements. The quantified items are all outflows: the $19.07 million joint-venture investment in Q1 2022 (claim 13918), the $8.46 million annual Beowulf fee (claim 12867) and the $339,200 transition-services payment (claim 12942). Readers should treat the 58,000-miner and 150 MW figures as capacity aspirations that were later revised down, not as contracted revenue. The record ends in March 2023, so later developments fall outside this read.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, because self-mining bitcoin produces revenue from network block rewards, not from contracted buyers. Most of what the record calls partnerships are supplier, power, joint-venture and service contracts. The earliest entry (claim 21075) concerns IKONICS's 2006 auditor ratification. It reflects the shell TeraWulf merged into and is not a commercial relationship.

The executed agreements are well identified. TeraWulf names Bitmain and Minerva as equipment counterparties, describing one executed Minerva purchase agreement and four non-fixed-price Bitmain agreements (claims 14028, 14051, 13977). It names NYPA for an initial 90 MW at Lake Mariner (claim 12842). It names Talen, through Cumulus Member, as its Nautilus joint-venture partner (claims 14120, 13714). Foundry and NYDIG are named under dated service, custody and execution agreements (claims 13335, 13344, 13345). The softer language sits around these deals. The 2021 reference to "competitive power supply agreements already in place" identifies no counterparty (claim 21144). The Bitmain "strategic relationship" offering "preferred access" is qualitative and has no contract behind it (claim 12891). NYPA expansion beyond 90 MW is framed only as "potential" (claim 12842). The 58,000-miner delivery schedule is introduced with "Despite securing," which signals shortfall against the commitment (claims 14123, 14247). The record also mentions "hosting arrangements" without naming any hosted customer (claim 12852).

Concentration has moved in the wrong direction on the Talen relationship. TeraWulf's attributable Nautilus share fell from 150 of 300 MW (claim 14112) to 33.33% in August 2022 (claim 13714), then to 25% (claims 12872, 12860), and that 25% remains "subject to adjustment." TeraWulf appoints one of five managers (claim 12874). Pool, custody and sales each run through a single provider (Foundry and NYDIG). Related parties recur: Beowulf E&D under a $339,200 transition services arrangement (claim 12942) and NovaWulf as an affiliate (claim 13849).

Revenue cannot be reconciled from this record. No revenue figures appear in it, and the only disclosed cash item is $19,072 of Q1 2022 joint-venture investment (claim 13918). Nautilus economics flow through biweekly bitcoin distributions net of power and operating costs (claim 12939), not through contracted revenue. The record is truncated after March 2023, so later developments are outside this assessment.

The record for TeraWulf through March 2023 contains no signed customer agreement. Every executed contract on file is a supply, power, service or joint-venture arrangement in which TeraWulf is the buyer, the venturer, or the user of a third party's services. That matters for diligence because this record cannot test customer quality or contract conversion at all. The company's revenue, as it says itself, came from self-mined bitcoin (claim 13441), not from counterparties paying for capacity.

The executed instruments are specific and named. They include equipment purchase agreements with Bitmain and Minerva (claims 14051, 14028, 13977), the Talen joint venture dated May 13, 2021 (claim 14120), the NYDIG custodial agreement of March 10, 2022 and the NYDIG Execution agreement of September 16, 2022 (claims 13412, 13414), the Foundry pool service agreement (claim 13335), and the NYPA supply for an initial 90 MW (claim 12842). The softer language is narrower than usual. 'Competitive power supply agreements already in place' (claim 21144) names no counterparty. The Bitmain 'strategic relationship' promising 'preferred access' (claim 12891) is qualitative framing layered over the signed purchase contracts. Nautilus being 'expected to be the first' behind-the-meter nuclear mine (claim 12923) is forward-looking. Claim 21075 concerns IKONICS's 2006 auditor and is a predecessor-shell artifact, not a TeraWulf relationship. NovaWulf (claim 13849) is a disclosed affiliate, not an arm's-length partner.

Concentration evolved toward dependence on a few named service providers and away from the flagship venture. The Talen economics shrank from a 50% share of up to 300 MW (claim 14112) to 33.33% (claim 13714) and then to 25%, with one of five board seats (claims 12872, 12874, 12860). Contributed miners fell to 20,000 (claim 13713) against 58,000 contracted units (claim 14247). Custody, sales execution and pool access all sit with two providers, NYDIG and Foundry.

Revenue arithmetic cannot be reconciled from this record. No revenue figure, hashrate-to-revenue bridge or per-contract value is disclosed. The only quantified items are cash outflows: a $19,072 thousand JV investment in Q1 2022 (claim 13918) and a $339,200 transition-services fee (claim 12942). The miner-delivery and equity-share figures move in different directions without a published reconciliation, so readers should treat capacity claims as unverified against the income line. The record is also truncated after claim 12838, and any later customer-style announcements fall outside this read.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. That fits a self-mining business whose revenue comes from block rewards rather than contracted offtake. What the record does show is a dense set of executed supplier, power and service contracts, nearly all with named counterparties. Bitmain and Minerva are identified as equipment vendors under executed purchase agreements (claims 14027, 14028, 13977) covering 58,000 miners for 2022 delivery (14247). The 'Despite securing' phrasing in claim 14123 implies delivery fell short of that figure. Foundry Digital operates the mining pool under a service agreement (13335, 13440). NYDIG provides custody under a dated March 10, 2022 agreement, and NYDIG Execution handles sales under a September 16, 2022 agreement (13344, 13345). Power comes from NYPA for an initial 90 MW at Lake Mariner (12842) and from the Susquehanna station for Nautilus (14129). Softer framing is confined to forward-looking upside: NYPA's 'potential to expand' (12890), and Bitmain described as a 'strategic relationship' conferring 'preferred access and advantaged cost structure' (12891), a claim of favourable terms beyond the four disclosed non-fixed-price agreements.

Identification is thinnest early in the record. The November 2021 statement of 'competitive power supply agreements already in place' (21144) names no counterparty. The Pennsylvania joint venture is first described without its partner (14141), although Talen is named elsewhere in the same filing (14112, 14120). Claim 21075 concerns IKONICS's 2006 auditor ratification and is not a TeraWulf commercial relationship; it should be disregarded.

Concentration moved in one clear direction: the Talen joint venture was restructured downward. TeraWulf's attributable Nautilus share fell from 150 of 300 MW (14112) to 33.33% (13714) and then 25% (12872, 12860). Its governance seat is one of five managers (12874). Exposure to the single partner-and-power relationship shrank, while operational dependence on NYDIG and Foundry for pool payouts, custody and sales became effectively total.

Revenue arithmetic cannot be reconciled from this record because no revenue figures appear in the claims. The only dollar amounts are $19,072 (in thousands) of joint-venture investment in Q1 2022 (13918) and a $339,200 transition-services fee payable by Nautilus to Beowulf E&D (12942). The record is also truncated at claim 12838, so later developments are outside the scope of this read.

This record contains almost no customer agreements in the conventional sense, and readers should weigh it that way. Through early 2023 TeraWulf disclosed that the majority of its revenue came from self-mined bitcoin (claim 13441), earned through participation in the Foundry Digital pool under a click-through Service Agreement (claims 13335, 13440). Every other named counterparty is a supplier, power provider, joint-venture partner, financial agent or affiliate rather than a buyer of TeraWulf's services. Claim 21075, an IKONICS auditor ratification from 2006, comes from the predecessor shell and has no bearing on TeraWulf's commercial book.

The named counterparties are reasonably well documented and appear to be executed contracts rather than MOU or framework language. These include Bitmain and Minerva equipment purchase agreements (claims 14027, 14028, 13977), the NYDIG custody and execution agreements with specific dates (claims 13412, 13414), the Talen joint venture dated May 13, 2021 (claim 14120), the Susquehanna-sourced nuclear power supply (claim 14113) and the New York Power Authority allocation (claim 12842). The softer language sits around these contracts rather than in them. The "strategic relationship" with Bitmain promising "preferred access" (claim 12891) is characterization layered on non-fixed-price agreements. The NYPA deal secures 90 MW with only the "potential to expand" (claim 12842). The 2021 reference to "competitive power supply agreements already in place" (claim 21144) names no counterparty. Bitcoin "earned from hosting arrangements" (claim 12852) implies hosting customers, but none is identified anywhere in the record.

Concentration has narrowed rather than diversified. The Talen venture was first framed as up to 300 MW with 150 MW attributable to TeraWulf (claim 14112). It was cut to a 33.33% stake (claim 13714) and then to 25% with one of five board seats (claims 12872, 12874, 12860). Of the 58,000 miners contracted for 2022 delivery (claims 14123, 14247), only 20,000 were committed to Nautilus (claim 13713). Revenue realization runs through a single pool operator and a single execution and custody provider. NYDIG, Foundry and the affiliated NovaWulf and Beowulf entities (claims 13849, 12942) represent counterparty and related-party exposure, not customer diversification.

The revenue arithmetic cannot be reconciled from this record. No revenue totals, hashrate-to-revenue bridges or hosting income figures are disclosed. The only dollar amounts are a $19.072 million joint-venture outflow (claim 13918) and a $339,200 related-party transition services fee (claim 12942). The record is also truncated at claim 12838, so any later shift toward hosted or HPC customer contracts falls outside what can be assessed here.

TeraWulf's record through March 2023 contains almost no customer agreements in the conventional sense. As a self-mining bitcoin operator, its revenue comes from block rewards paid through a mining pool, so the named counterparties are suppliers, service providers and joint-venture partners rather than buyers. The one entry that falls outside this pattern is not a TeraWulf relationship at all: claim 21075 is a 2006 auditor ratification by IKONICS, the shell TeraWulf later merged into, and it should carry no weight. Most counterparties are named and documented as executed contracts. These include the Talen joint venture dated May 13, 2021 (14120), the Minerva purchase agreement and four Bitmain purchase agreements (14028, 14027, 13977), the NYDIG custodial agreement of March 10, 2022 (13412, 13344), the NYDIG Execution agreement of September 16, 2022 (13414, 13345), the Foundry pool service agreement (13335), and the NYPA power supply agreement (12842). The earliest power claim is the softest. In November 2021 the company cited "competitive power supply agreements already in place" without naming any counterparty (21144), and only later disclosures tie this to Susquehanna and NYPA (14113, 12842). The Bitmain contracts are explicitly "non-fixed price" (14051), which means the "advantaged cost structure" described in claim 12891 is relational language rather than a locked commercial term. The phrase "Despite securing" agreements for 58,000 miners (14123) signals that delivery fell short of the commitment. Concentration has moved away from TeraWulf over time. The Talen venture was first presented as 150 MW of 300 MW attributable to TeraWulf (14112). It was then cut to a 33.33% equity interest with 20,000 contributed miners (13714, 13713), and later to 25% with Talen's Cumulus appointing four of five managers (12872, 12874, 12860). This shift is material: TeraWulf now holds a minority, non-controlling position in its flagship nuclear-powered site. Related-party links to Beowulf E&D (13718, 12942) and NovaWulf (13849) add affiliate exposure that should be watched. The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed. The only dollar amounts are a Q1 2022 joint-venture outflow of 19,072, presumably in thousands (13918), and a $339,200 transition-services fee (12942), and neither maps to the announced deals. The record is also truncated at claim 12838, so any later hosting or HPC customer contracts fall outside this assessment.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense, signed or otherwise. The company is a self-mining bitcoin operator. Its disclosed revenue comes from bitcoin earned through Foundry's mining pool and sent to a NYDIG-custodied wallet (claims 13440, 13441). The named counterparties are suppliers, service providers and a joint-venture partner, not buyers of a TeraWulf product. One stray item, claim 21075, concerns the auditor of the predecessor shell IKONICS and is not commercial evidence at all.

Within that supplier and partner set, the documentation is unusually concrete. Bitmain and Minerva are named under executed purchase agreements (claims 14051, 14028, 13977). The Bitmain contracts are explicitly non-fixed-price, which limits how far the 58,000-miner delivery figure (claims 14247, 14123) can be read as a firm commitment. The 'despite securing' phrasing in claim 14123 itself signals that deliveries fell short. NYDIG custody and execution agreements carry signing dates (claims 13412, 13414), as does the Foundry pool service agreement (claim 13335). The NYPA allocation is named but sized only as an initial 90 MW with expansion 'potential' (claim 12842). The later 'strategic relationship' with Bitmain (claim 12891) is promotional framing layered over the earlier contracts, not a new agreement. The only unnamed reference is to 'hosting arrangements' (claim 12852), which identifies no counterparty and attaches no economics.

Concentration has shifted toward related parties and away from TeraWulf's own economics. The Talen joint venture began as a 50/50 split of up to 300 MW (claim 14112). It was cut to a 33.33% stake (claim 13714) and then to 25%, with one of five board seats (claims 12872, 12874, 12860). Beowulf E&D, which is affiliated with management, provides operating services to Nautilus (claims 13718, 12942). NovaWulf is disclosed as an affiliate (claim 13849). Counterparty risk therefore sits with Talen, Foundry, NYDIG and insiders rather than with a customer base.

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-to-revenue or hosting-fee figures are disclosed. The only quantified flow is $19,072 thousand of joint-venture investment (claim 13918), and Nautilus distributions are formula-based and net of power costs (claim 12939). The record also ends at March 2023 (claim 12838 is truncated). Any later high-performance-computing or AI hosting contracts, which would be the first true customer agreements, are outside what can be assessed here.

TeraWulf's record through March 2023 contains no revenue customers in the conventional sense. Every named counterparty is a supplier, a service provider, a financing agent or a joint-venture partner. Revenue is self-mined bitcoin earned through Foundry Digital's pool (13440, 13335) and custodied and sold through NYDIG under dated, executed agreements (13412, 13414). The language throughout is signed-agreement language rather than MOU, LOI or 'discussions' phrasing. The Bitmain and Minerva purchase agreements are described as executed (14028, 13977), as are the NYPA power agreement (12842) and the Susquehanna-sourced Nautilus power supply with a stated five-year term (14113). The softest framing is Bitmain as a 'strategic relationship' offering 'preferred access' (12891), which describes the relationship rather than an additional contract. Earlier 'power supply agreements already in place' (21144) names no counterparty. 'Hosting arrangements' appear only in passing (12852), with no hosted client identified, so any hosting revenue cannot be attributed. Claim 21075 is a 2006 IKONICS auditor ratification inherited from the reverse-merger shell and has no commercial bearing. Claims 13973 (ATM sales agent) and 13391 (shareholder voting agreement) are capital-markets items, not commercial ones. Counterparty concentration has narrowed rather than diversified. The Talen joint venture began as a 50/50 split of up to 300 MW (14112). It was cut to 33.33% in August 2022 (13714) and then to 25% with a 1-of-5 board seat (12872, 12874, 12860), leaving Talen's Cumulus affiliate in control of Nautilus. Equipment supply rests on two vendors (14246), and monetization depends on a single pool operator plus a single custodian-broker. The revenue arithmetic cannot be reconciled from this record. The 58,000-miner commitment (14247, 14123) is introduced with 'despite securing,' which signals a delivery shortfall but quantifies none. The 20,000 miners contributed to Nautilus (13713) predate the stake reduction, so the post-amendment miner count is undisclosed. The disclosed dollar figures cover only the $19.1 million JV investment (13918) and the $339,200 transition-services fee (12942). No hashrate-to-revenue bridge appears. The claim list provided ends mid-entry at 12838, so later announcements, including any post-2023 hosting or compute customer contracts, fall outside this read and should be assessed separately before drawing conclusions about the current customer base.

TeraWulf's record contains no signed customer agreements in the conventional sense, because a self-mining bitcoin operator has no contracted buyer of its output. Revenue comes from block rewards earned through the Foundry Digital mining pool (claims 13440, 13335, 12854) and is converted to cash through NYDIG Execution under a September 2022 agreement (claims 13414, 13334). Every named counterparty on file is a supplier, service provider, joint-venture partner or affiliate. Claim 21075 concerns IKONICS Corporation's 2006 auditor ratification and predates the reverse merger, so it says nothing about TeraWulf's commercial base. The language is largely executed rather than aspirational. TeraWulf states that it executed an equipment purchase agreement with Minerva and four non-fixed-price sales and purchase agreements with Bitmain (claims 14028, 14027, 13977). Its custody and execution agreements with NYDIG carry specific dates (claims 13412, 13414), and its Talen joint venture was entered on May 13, 2021 and amended on August 27, 2022 (claims 14120, 12848). The one counterparty left unnamed is the early 'competitive power supply agreements' reference (claim 21144), which is only later tied to the Susquehanna Station arrangement for Nautilus (claims 14113, 12859). Concentration has moved toward fewer and more related counterparties. The Nautilus stake implied at 150 of 300 MW (claim 14112) fell to 33.33 percent (claim 13714) and then to 25 percent with one of five board seats (claims 12872, 12874, 12860). That leaves Talen's affiliate in control of the asset, while operations and administration run through the related party Beowulf E&D under a reduced annual base fee of $8.46 million (claims 13718, 12942, 12867). NYDIG serves as custodian, execution venue and DACA party (claim 12905), so treasury and liquidity counterparty risk sits with a single group. The arithmetic does not fully reconcile on the record provided. The claims cite delivery commitments for 58,000 miners in 2022 (claims 14247, 14123), but those commitments are later described with the qualifier 'despite securing,' which signals shortfalls. Only 20,000 miners were pledged to Nautilus (claim 13713). No hash-rate, revenue or delivered-unit figures appear that would tie these commitments to reported income. The only financial figure, $19,072 thousand of Q1 2022 joint-venture investment (claim 13918), measures capital outflow rather than earned revenue. The partially captured reference to five additional Bitmain future sales and purchase agreements (claim 12847) should be checked against the full filing.

TeraWulf's record contains almost no customer agreements in the conventional sense, and that is the central diligence point. As a self-mining bitcoin operator, its disclosed counterparties are suppliers, service providers, financing agents and joint-venture partners rather than buyers of a product. The agreements that are executed and named are concrete: equipment purchase agreements with Bitmain and MinerVA (claims 14051, 14052, 14028, 13977, later expanded to five further Bitmain future sales agreements in 12847), a Digital Asset Custodial Agreement with NYDIG dated March 10, 2022 (13412, 13333), a Digital Asset Execution Agreement with NYDIG Execution dated September 16, 2022 (13414, 13334), a pool Service Agreement with Foundry Digital (13346, 13335), an at-the-market Sales Agreement with Cantor, B. Riley and D.A. Davidson (13973), and the Talen joint venture, first signed May 13, 2021 and amended and restated August 27, 2022 (14120, 12848). None of this is MOU, LOI or 'discussions' language, and no counterparty is hidden behind an unnamed 'strategic partner' or 'anchor customer' label. One record should be excluded outright: claim 21075 is a 2006 IKONICS auditor ratification, a legacy item from the reverse-merger shell with no bearing on TeraWulf's commercial book.

Concentration has narrowed and tilted toward a single partner. The Nautilus economics moved materially against TeraWulf across filings. The March 2022 framing of 150 of 300 MW attributable to the company implied roughly a 50% share (14112). The August 2022 restatement set a 33.33% interest against Talen's 66.67% (13714), and by March 2023 that had become 25% versus 75%, with one of five board seats (12872, 12938, 12874, 12840). Operationally, revenue depends on one pool operator, Foundry, and one custody and execution provider, NYDIG. That is a real single-vendor exposure even though it is not customer concentration.

The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed in it. The only financial datum is a $19,072 (thousands) Q1 2022 joint-venture investment (13918). The 58,000-miner delivery commitment (14247, 14123) set against the 20,000 miners contributed to Nautilus (13713) is internally plausible, but hash-rate-to-revenue conversion and the biweekly Nautilus bitcoin distributions net of power costs (12939) cannot be tested without production and price data. The claim list also truncates at claim 12838, so later developments, including any hosting or HPC customer contracts, are outside this read.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. As a self-mining bitcoin operator, its revenue comes from block rewards earned through a pool, not from contracted offtakers. Every named counterparty is a supplier, service provider, joint-venture partner or financing agent, and all of them are named. The filings never lean on unnamed 'strategic partners' or 'anchor customers,' and there is no MOU, LOI or 'discussions' language on record. The relationships are executed contracts with dates attached. Bitmain and Minerva supply equipment under four non-fixed-price agreements and one purchase agreement (claims 14051, 14052, 13977), later expanded to five further Bitmain future sales agreements (claim 12847). NYDIG provides custody under an agreement dated March 10, 2022 (claim 13412), and NYDIG Execution handles sales under an agreement dated September 16, 2022 (claim 13414). Foundry operates the mining pool under a service agreement (claim 13346). The Cantor, B. Riley and D.A. Davidson sales agreement is an equity-issuance facility, not a commercial relationship (claim 13973). Claim 21075 refers to IKONICS's 2006 auditor ratification. It predates TeraWulf's reverse merger into that shell and has no diligence relevance here. Claim 12838 arrived truncated and cannot be assessed.

Concentration is heavy and has tightened rather than diversified. One pool operator, one custodian and its affiliated execution desk, and one hardware vendor dominate the operating chain. The single joint venture, Nautilus with Talen's Cumulus Coin subsidiary (claims 14120, 12841), shrank as a share of TeraWulf's footprint. TeraWulf's equity interest fell from 33.33% (claim 13714) to 25% (claim 12872), with board representation of one manager against four for Cumulus (claim 12874). The announced 150 MW attributable share (claim 14112) should therefore be read against the later economics. Bitcoin is distributed pro rata to hash contributed, net of power and operating costs (claim 12939). The record also shows related-party links through Beowulf E&D (claims 13718, 12942) and NovaWulf (claim 13849).

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed in it. The only monetary points are the $19,072 thousand Q1 2022 joint-venture investment (claim 13918) and the $339,200 Beowulf transition-services fee (claim 12942). The 58,000-miner delivery commitment (claims 14247, 14123) is a capacity input, not contracted revenue. The 2022 filings already hedge that it was achieved 'despite' the agreements, which suggests delivery or deployment shortfalls that warrant follow-up against reported hash rate.

TeraWulf's record contains no customer agreements in the conventional sense. Through March 2023 its revenue came from self-mined bitcoin (claim 13441), so every named counterparty is a supplier, service provider, joint-venture partner, or affiliate rather than a buyer of TeraWulf's output. The earliest entry, claim 21075, concerns the auditor of IKONICS, the predecessor public shell, and has no bearing on commercial traction. One thing works in TeraWulf's favour: the record has none of the soft language this review screens for. No MOUs, LOIs, 'discussions', or unnamed 'strategic partners' or 'anchor customers' appear. Every counterparty is identified and every arrangement is described as executed. These include equipment purchase agreements with Bitmain and Minerva (claims 14051, 14052, 14028, 13977), the Talen joint venture dated May 13, 2021 (claim 14120), the NYDIG custodial agreement of March 10, 2022 (claim 13412), the NYDIG Execution agreement of September 16, 2022 (claim 13414), and the Foundry pool service agreement (claim 13346). The weaker point is that the Bitmain contracts are 'non-fixed price' (claim 14027), so a signed agreement did not lock in cost. Claim 14123 also frames the 58,000-miner delivery commitment with 'despite securing', which points to a delivery shortfall rather than fulfilment. Counterparty concentration is high and has moved against TeraWulf. Its economics in the Nautilus facility steadily diluted. It began as 150 of 300 MW, roughly half (claim 14112). The August 2022 amendment cut this to 33.33% with a 20,000-miner contribution (claims 13714, 13713). The March 2023 second amended agreement cut it again to 25% (claims 12872, 12920, 12921), with one of five board seats (claim 12874). Power for the facility also depends on Talen's Susquehanna Station, at a fixed $0.02 rate under a five-year term (claims 12838, 14113). Custody and sales both run through NYDIG entities, and hashrate runs through a single pool, Foundry. Related-party links to Beowulf E&D and NovaWulf (claims 13718, 12942, 13849) add a further governance dimension. Revenue cannot be reconciled from this record. The only dollar figures are a $19,072 joint-venture investment line for Q1 2022 (claim 13918, units not stated) and a $339,200 transition-services fee payable by Nautilus (claim 12942). Neither ties to a disclosed revenue total. Because Nautilus distributes bitcoin in proportion to hashrate after costs (claim 12939), TeraWulf's revenue from it depends on miner deployment and bitcoin price rather than on any contracted take. The record ends in March 2023, so any later hosting or compute customer arrangements fall outside this read.

TeraWulf's record contains no conventional customer announcements, and no MOU, LOI, framework or pilot language either. Every counterparty is described as party to an executed agreement, but none is a buyer of a TeraWulf service in the usual sense. As a self-miner, the company earns block rewards through Foundry Digital's pool (claims 13440, 13346, 13335), holds bitcoin with NYDIG under a March 2022 custodial agreement (13412, 13333), and sells through NYDIG Execution under a September 2022 execution agreement (13414, 13334). These are dated, signed service relationships, not revenue counterparties. The only hint of third-party revenue is a passing reference to bitcoin 'earned from hosting arrangements' (12852), and the hosting client is never named. That is the one place where an unidentified counterparty sits behind the revenue line. Claim 21075, an IKONICS auditor ratification from 2006, is a legacy artifact of the reverse merger and has no diligence weight.

On the supply side, the Bitmain and Minerva agreements are specific: four non-fixed-price purchase agreements with Bitmain and one with Minerva (13977, 14051, 14052), for 58,000 miners through December 2022 (14247). The 'Despite securing' framing in 14123 suggests deliveries fell short of that commitment. The contracted figure should not be read as installed capacity.

The most material counterparty is Talen, and here concentration moved against TeraWulf. The May 2021 joint venture (14120) was presented as 150 of 300 MW attributable to TeraWulf (14112). By August 2022 TeraWulf held 33.33% of Nautilus against a 20,000-miner contribution (13713, 13714). By March 2023 it held 25%, with one of five board seats (12872, 12874). Economic dependence on a single partner-controlled nuclear-powered site therefore grew while governance influence shrank. Distributions follow hash-rate contribution net of power and operating costs (12939). Related-party exposure also deserves weight: Beowulf E&D, owned by the CEO, provides infrastructure and administrative services to TeraWulf and Nautilus, including a $339,200 transition arrangement (12845, 13718, 12942), and NovaWulf is a disclosed affiliate (13849).

Revenue arithmetic cannot be reconciled from this record. The only figures are a $19.07 million Q1 2022 joint-venture investment outflow (13918) and the Beowulf transition fee. No revenue, hash-rate or energized-miner totals are on file to test against the 58,000-unit commitment or the attributable megawatts. Reconciliation would require the 10-K segment and equity-method disclosures.

TeraWulf's record through March 2023 contains no signed customer agreement in the conventional sense, because the business model has no named buyers. Revenue is self-mined bitcoin earned through participation in a pool operated by Foundry Digital (claims 13440, 13346, 13335). Foundry is a service provider rather than a customer, and the pool relationship is formed simply by accessing and using the pool. The other executed, dated contracts are vendor and infrastructure arrangements: the NYDIG custodial agreement of March 10, 2022 (claims 13412, 13333), the NYDIG Execution agreement of September 16, 2022 (claims 13414, 13334), and the Bitmain and Minerva equipment purchase agreements (claims 13977, 14027, 14028). The 58,000-miner delivery figure (claims 14247, 14123) describes supply commitments made to TeraWulf, not demand for its output. Claim 21075 is a 2006 auditor ratification for the predecessor shell, IKONICS, and has no commercial relevance. The one unidentified counterparty set appears in claim 12852, which references bitcoin earned from hosting arrangements without naming any hosted client. That is the only place a customer-like relationship is described and left anonymous. The most material named counterparty is Talen Energy, a joint-venture partner rather than a customer. Its position strengthened steadily over the record. TeraWulf's share fell from 150 of 300 MW, roughly half (claim 14112), to a 33.33% equity interest (claim 13714), then to 25% with one of five board seats (claims 12872, 12874). Concentration therefore shifted toward the counterparty, with distributions tied to hash-rate contributions net of costs (claim 12939). Related-party exposure is notable. Beowulf E&D, which is owned and controlled by the CEO, provides services to both TeraWulf and Nautilus (claims 12845, 13718, 12942), and NovaWulf is disclosed as an affiliate (claim 13849). Revenue arithmetic cannot be reconciled from this record because no revenue totals appear. The only figures are cash outflows: $19,072 thousand of JV investment in Q1 2022 (claim 13918) and a $339,200 transition-services fee payable by Nautilus (claim 12942). Any revenue claim should be tested against hashrate, uptime and bitcoin price, not against announced deals.

On this record TeraWulf has no signed customer agreements in the conventional sense. The disclosures show a self-mining business whose named counterparties are suppliers, service providers, a joint-venture partner and related parties, not buyers of capacity or compute. The one exception is a passing reference to 'bitcoin earned from hosting arrangements' (claim 12852). The hosting counterparties behind it are never identified, and nothing on file sizes them. Claim 21075, an IKONICS auditor ratification from 2006, belongs to the pre-merger shell and should be excluded from any customer read.

The named counterparties are concrete and documented as executed agreements rather than MOUs or LOIs. Bitmain and Minerva supplied equipment under four non-fixed-price purchase agreements and one purchase agreement (claims 14246, 14051, 14052, 13977). Talen Energy, acting through its Cumulus affiliate, is the Nautilus joint-venture partner (claims 14120, 14112, 14128). Foundry runs the mining pool (claims 13335, 13440). NYDIG provides custody and NYDIG Execution handles sales (claims 13412, 13414). NYPA supplies power to Lake Mariner (claim 12890). Beowulf E&D, owned by the CEO, and the NovaWulf fund are disclosed as related parties (claims 12845, 13849), which matters for any arm's-length assessment. None of these relationships produces third-party revenue. Foundry, as pool operator, is the closest thing to a revenue conduit.

Counterparty concentration has deepened over time and moved in TeraWulf's disfavour. The Talen joint venture was first framed as up to 300 MW, with 150 MW attributable to TeraWulf (claim 14112). It was then restated as a 33.33% equity interest (claim 13714) and later as 25% against Cumulus's 75%, with one of five board seats (claims 12872, 12874). Cumulus alone held the expansion option (claim 12922). Equipment supply narrowed to two vendors, and operations depended on Foundry and NYDIG.

The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed. The only dollar amounts are a $19,072 thousand joint-venture investment outflow in Q1 2022 (claim 13918) and a $339,200 transition-services fee paid by Nautilus to the related party Beowulf E&D (claim 12942). There is also a gap on equipment. The company announced 58,000 miners for 2022 delivery (claims 14247, 14123) but committed only 20,000 to Nautilus (claim 13713), so the deployment of the remaining fleet is not explained in these disclosures. The record ends in March 2023 and contains no later compute or hosting customer announcements.

TeraWulf's record through March 2023 contains no signed revenue customer in the conventional sense. As a self-miner, its revenue comes from the Bitcoin network through Foundry Digital's pool, under a Service Agreement formed simply "by accessing and using the Pool" (claims 13335, 13346, 13440). Every other named counterparty is a supplier, financing agent, service provider or joint-venture partner, and each is identified by name. These are Bitmain and Minerva for equipment (14246, 13977), NYDIG for custody and execution under agreements dated March 10 and September 16, 2022 (13412, 13414), NYPA for 90 MW at Lake Mariner (12890), and Cantor, B. Riley and D.A. Davidson under an at-the-market Sales Agreement (13973). Claim 21075 is an IKONICS auditor ratification inherited from the reverse-merger shell and is not commercial evidence. No "anchor customer" or unnamed "strategic partner" language appears. The closest thing to an unidentified counterparty is a passing reference to "bitcoin earned from hosting arrangements" (12852), which names no hosting client.

The language is contractual rather than MOU-style, but its firmness varies. The Bitmain agreements are explicitly "non-fixed price" (14051, 14027), so the 58,000-miner delivery figure (14247, 14123) describes a supply framework rather than a fixed purchase commitment. The record does not show how many units actually arrived.

The Talen relationship has moved against TeraWulf. The joint venture was first framed as 150 MW of 300 MW attributable to TeraWulf (14112), then reduced to a 33.33% equity interest (13714), then to 25% with one of five board seats (12872, 12874). Cumulus also holds an option to expand to 300 MW funded solely by itself (12922). Related-party exposure is persistent. Beowulf E&D, controlled by the CEO, provides services to both TeraWulf and Nautilus (12845, 13718, 12942), and NovaWulf is disclosed as an affiliate (13849).

The revenue arithmetic cannot be reconciled from this record. No claim discloses mining revenue, realised hash rate or Nautilus distributions. The only figures are a $19,072 thousand Q1 2022 joint-venture investment (13918) and a $339,200 transition-services fee (12942). The 20,000 miners committed to Nautilus (13713), set against 58,000 contracted, leave the Lake Mariner fleet size unstated. The record also cuts off mid-entry at claim 12890 and ends in March 2023, so any later hosting or compute contracts fall outside this read.

TeraWulf's record contains no customer contracts in the conventional sense, and the dataset should be read that way. The company mines bitcoin for its own account. The claims on file describe suppliers, a joint-venture partner, service providers and financing agents, not buyers of a product. The one item that predates the company, claim 21075, is an IKONICS auditor ratification from 2006 inherited through the reverse-merger shell. It has no commercial diligence value.

The agreements that were actually executed are all named. Bitmain and Minerva signed equipment purchase agreements (14051, 14052, 13977, 12847). Foundry Digital provides pool services under a Service Agreement (13335, 13346). NYDIG has a Custodial Agreement dated March 10, 2022 (13412, 13333) and an Execution Agreement dated September 16, 2022 (13414, 13345). Cantor, B. Riley and D.A. Davidson are party to an at-the-market Sales Agreement (13973). The record contains no unnamed 'anchor customer' language. The closest thing to opacity is related-party exposure: NovaWulf is described as an affiliate (13849), and Beowulf E&D is paid $339,200 for transition services (12942, 13718). These relationships deserve scrutiny for governance reasons rather than as revenue claims.

Counterparty concentration has tightened around Talen. The 2022 framing described a partnership for up to 300 MW with 150 MW attributable to TeraWulf (14112), which implies parity. By August 2022 that had become a 33.33% equity interest (13714). By March 2023 it was 25%, with one of five board seats (12872, 12874, 12938). Nautilus's power also comes from Talen's Susquehanna Station at a fixed rate (12838, 14113), so a single counterparty is landlord, power vendor and controlling partner. On the input side, the 58,000-miner delivery commitment (14247, 14123) appears in a sentence beginning 'Despite securing,' which suggests the deliveries were not realized on schedule. The 20,000 miners contributed to Nautilus (13713) are a related but separate figure.

The revenue arithmetic cannot be reconciled from this record. Claim 13441 states that the majority of revenue comes from self-mined bitcoin, but no revenue figure is disclosed. The only dollar amounts are the $19,072 thousand joint-venture investment for Q1 2022 (13918) and the Beowulf fee. Nautilus distributions are tied to hash-rate contribution net of costs (12939), which makes them dependent on deployed miners rather than any contract value. The record also ends mid-quote at claim 12848, so later developments, including any hosting or HPC tenant agreements, are absent and should not be inferred.

TeraWulf's record through March 2023 contains no signed customer agreement in the conventional sense, because the company sells self-mined bitcoin rather than contracted services. Every named counterparty is a supplier, a joint-venture partner, a service provider, or a financing agent. Bitmain and Minerva are identified as equipment vendors under executed purchase agreements (claims 13977, 14028, 14051), although the Bitmain contracts are described as 'non-fixed price', which weakens the 58,000-miner delivery commitment (claims 14123, 14247). Talen Energy is named as the Nautilus joint-venture counterparty under a dated agreement (claim 14120). The NYPA power arrangement (claim 12842), the NYDIG custodial and execution agreements (claims 13333, 13414), and the Foundry pool service agreement (claim 13335) are all executed and dated. None of these relationships uses MOU, LOI, or 'discussions' language, and none generates revenue from a third party. The one reference to 'hosting arrangements' (claim 12852) names no hosting customer, so it is the only place where an unidentified revenue counterparty appears. Claim 21075 is an IKONICS auditor ratification from 2006. It belongs to the reverse-merger shell's history and should not be read as a TeraWulf relationship. Counterparty concentration has narrowed over time rather than diversified. Nautilus began as a nominally 300 MW venture with 150 MW attributable to TeraWulf (claim 14112), with a 33.33% equity interest against Talen affiliate Cumulus's 66.67% (claim 13714). It was then restructured to 25% versus 75% (claims 12872, 12921), with TeraWulf appointing one of five managers (claim 12874). Bitcoin flows run entirely through Foundry and NYDIG (claim 13441), which creates single-provider dependence on the realization side. The disclosed figures cannot reconcile revenue to deals, because no deal carries contracted revenue. The figures that are disclosed are outflows: $19.072 million of Q1 2022 joint-venture investment (claim 13918) and a $339,200 transition-services payment from Nautilus to the affiliated Beowulf E&D (claim 12942). Related-party exposure, including NovaWulf (claim 13849) and Beowulf, is the more material diligence issue. Any later contracted-revenue relationships fall outside this record and would need separate verification.

TeraWulf's record through early 2023 contains no signed revenue customers in the conventional sense. Every executed agreement is a supply, power, service, financing, or joint-venture contract, and revenue comes from self-mined bitcoin (claim 13441). The firmest documents are the Bitmain and Minerva equipment purchase agreements (claims 14051, 14052, 13977). These cover 58,000 miners for delivery during 2022 (claims 14247, 14123). Other executed contracts are the Foundry pool service agreement (claims 13335, 13346), the NYDIG custody agreement dated March 10, 2022 (claims 13412, 13344), and the NYDIG Execution agreement dated September 16, 2022 (claims 13414, 13345). Power arrangements are described as secured: Susquehanna nuclear supply at a fixed $0.02 rate (claims 14113, 12838) and an initial 90 MW from NYPA with only 'potential' to expand (claims 12842, 12890). The record contains no MOU or LOI language. The softer wording sits in the Talen arrangement, framed as 'up to 300 MW' with 150 MW attributable once built (claim 14112), and in the NYPA expansion option. Claim 21075, the 2006 auditor ratification for IKONICS, is a legacy shell-company item and should not count as a commercial relationship.

All counterparties are named, and none is described anonymously as an 'anchor customer'. Several, however, are related parties. NovaWulf is an affiliate (claim 13849), and Beowulf E&D provides operating services to Nautilus, including a $339,200 transition arrangement (claims 13718, 12942). Diligence should treat those relationships separately from arm's-length agreements.

Concentration has tightened rather than diversified. TeraWulf's Nautilus equity fell from 33.33% (claim 13714) to 25% (claim 12872). Its board representation is one of five managers (claim 12874), and its miner contribution is 20,000 units (claim 13713). As a result, the flagship nuclear-powered asset is now controlled by Talen's Cumulus Member. Execution and custody are both concentrated in the NYDIG group (claims 13441, 12853). Claim 12852 mentions bitcoin 'earned from hosting arrangements', but no hosting customer is ever named. That gap should be examined.

The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed, only a $19,072 thousand joint-venture investment outflow (claim 13918). The 58,000 contracted miners also cannot be matched to deployed capacity at Lake Mariner and Nautilus. Delivery slippage is implied by the phrase 'despite securing' (claim 14123), but it is not quantified.

TeraWulf's record contains no conventional customer contracts, and that is the main diligence finding. The company says the majority of its revenue comes from self-mined bitcoin (13441), routed through the Foundry pool (13440). Every named agreement on file is therefore a supply, power, joint-venture, financing or service relationship rather than a sale to an identified buyer. Claim 21075, the 2006 IKONICS auditor ratification, is a legacy artifact of the predecessor shell and should be disregarded.

Within that frame, the executed paper is reasonably well identified. Signed instruments include a Minerva equipment purchase agreement (14028) and four Bitmain sales and purchase agreements (14027). The Bitmain agreements are explicitly non-fixed price, so the 58,000-miner delivery figure (14247, 14123) is a volume commitment without locked economics, and the filing frames it with 'Despite securing', which signals a shortfall. Other dated, executed agreements are the Talen joint venture of May 13, 2021 (14120), the Susquehanna-linked power supply with a five-year term (14113), the NYDIG custody agreement of March 10, 2022 (13412), the NYDIG Execution agreement of September 16, 2022 (13414), and the at-the-market Sales Agreement with Cantor, B. Riley and D.A. Davidson (13973). The Foundry 'Service Agreement' is formed merely by accessing the pool (13335), so it is closer to click-through terms than negotiated paper. Softer language sits around Bitmain's 'strategic relationship' and 'preferred access' (12891), the NYPA arrangement's 'potential to expand' beyond an initial 90 MW (12890), and the forward-tense 'expected to be the first' nuclear-powered site (12923).

The only partially anonymised counterparty is 'Cumulus Member', described only as a Talen affiliate (14120). Related-party links are disclosed but warrant attention: NovaWulf is an affiliate (13849), and Beowulf E&D supplies services to Nautilus, including a $339,200 transition fee (12942).

Concentration has drifted toward Talen's control. TeraWulf's attributable share started at 150 of 300 MW (14112), fell to a 33.33% equity interest (13714), and then to 25% with one of five board seats (12872, 12874). Over the same period, operating dependence consolidated onto Foundry and NYDIG.

The revenue arithmetic cannot be reconciled from this record. Apart from a $19,072 joint-venture investment line (13918) and the Beowulf fee, no revenue figures are disclosed. The 20,000 miners pledged to Nautilus (13713) cannot be tied to realised bitcoin distributions (12939) or to delivered fleet counts.

TeraWulf's record through March 2023 contains almost no customer agreements in the conventional sense. What it shows are supply, joint-venture, power and service contracts, most of them executed and with named counterparties. Readers should not mistake these for revenue-generating customer relationships. The executed agreements are concrete. Bitmain (four non-fixed-price purchase agreements) and Minerva (an equipment purchase agreement) are named suppliers (claims 14051, 14028, 13977). NYDIG holds the custody agreement dated March 10, 2022, and NYDIG Execution holds the execution agreement dated September 16, 2022 (claims 13412, 13414). Foundry Digital provides the pool service agreement (claim 13335). The New York Power Authority supplies an initial 90 MW at Lake Mariner (claim 12842), with expansion described only as 'potential'. The 58,000-miner delivery commitment (claims 14247, 14123) was a contracted supply schedule, and the March 2022 filing already frames it with 'despite securing', which signals delivery shortfall rather than strength. No MOU, LOI or 'discussions' language appears, and no unnamed 'strategic partner' or 'anchor customer' is invoked. The one opacity is claim 12852, which refers to bitcoin 'earned from hosting arrangements' without naming any hosting client. That makes it the only unidentified commercial counterparty on the record. Claim 21075, an IKONICS auditor ratification from 2006, predates TeraWulf's reverse merger into the IKONICS shell and is not relevant to customer diligence. Concentration has shifted mainly through the Talen relationship. TeraWulf originally held 50% of a 300 MW Nautilus venture (claims 14112, 14120). That fell to 33.33% with a 20,000-miner contribution in August 2022 (claims 13713, 13714) and then to 25% by March 2023 (claims 12938, 12860). Cumulus/Talen also controls four of five board seats (claim 12874). Economic exposure to the flagship nuclear-powered site has therefore narrowed while dependence on Talen's governance has deepened. NYDIG, meanwhile, concentrates both custody and liquidation of all mined bitcoin. The revenue arithmetic cannot be reconciled from this record. The only disclosed figure is a $19,072 thousand joint-venture investment outflow in Q1 2022 (claim 13918). Revenue derives from self-mined bitcoin paid through Foundry (claim 13441) and biweekly Nautilus distributions net of power and operating costs (claim 12939), and none of these is a contracted customer revenue stream. Any diligence model therefore rests on hashrate, bitcoin price and power cost, not on announced deals. The record supplied here ends in March 2023 and predates any later hosting or HPC customer announcements, which would need separate verification.

TeraWulf's record contains no customer agreements in the conventional sense. The company is a self-miner whose revenue comes mainly from bitcoin it mines itself (claim 13441). The named counterparties are therefore suppliers, service providers, a power authority and a joint-venture partner, not buyers of a product. Every counterparty on record is identified by name, and none is described only as a 'strategic partner' or 'anchor customer'. Claim 21075 is unrelated. It concerns the 2006 auditor ratification at IKONICS, the predecessor shell, and should be excluded from any commercial read.

Among the signed instruments, the strongest are the four non-fixed-price Bitmain purchase agreements and the Minerva purchase agreement (claims 14051, 14052, 13977). The NYDIG custody and execution agreements are dated March 10 and September 16, 2022 (claims 13412, 13414). The Foundry pool service agreement is formed simply by accessing the pool (claim 13346), which is a thin contractual basis. Several items carry softer or conditional language. The 58,000-miner delivery schedule (claim 14247) is a supply commitment, not delivered capacity. The NYPA arrangement secures 'an initial 90 MW' with only 'the potential to expand' (claim 12842). The Susquehanna power supply runs a five-year term with renewal options (claim 14113).

Concentration has narrowed rather than diversified. The Talen joint venture started as up to 300 MW, of which 150 MW was attributable to TeraWulf (claim 14112). The August 2022 amendment cut TeraWulf's stake to 33.33% against a 20,000-miner contribution (claims 13714, 13713). The March 2023 second amended agreement cut it again to 25%, with one of five board seats (claims 12938, 12874). Related-party dependence is material: NovaWulf is an affiliate (claim 13849), and Beowulf E&D draws an $8.46 million annual base fee plus $339,200 in transition services (claims 12867, 12942).

Revenue cannot be reconciled from this record. No revenue figure appears. The only cash datapoint is a $19,072 thousand joint-venture investment in Q1 2022 (claim 13918). There is also an unresolved gap: 58,000 contracted miners (claim 14123) against 20,000 committed to Nautilus, with no disclosed placement of the remaining units or hashrate yield. The record is truncated mid-quote at claim 12920, so later amendments may change this picture.

TeraWulf's record through early 2023 contains no signed customer agreements in the conventional sense, and that absence is itself the main finding. As a self-mining bitcoin operator, its revenue comes from block rewards paid through a mining pool rather than from contracted buyers. Every named counterparty on file is a supplier, financing agent, joint-venture partner or service provider. Claim 21075 is not a TeraWulf counterparty at all. It is a 2006 IKONICS auditor ratification inherited through the reverse-merger shell, and it should be disregarded for diligence purposes.

The counterparties are consistently named, and the language is executed rather than aspirational. Bitmain and Minerva are identified as equipment suppliers under four non-fixed-price purchase agreements and one executed purchase agreement (claims 14246, 14051, 14028, 13977). NYDIG holds custody under a dated March 10, 2022 agreement (claims 13412, 13333). NYDIG Execution handles bitcoin sales under a September 16, 2022 agreement (claims 13414, 13334). Foundry Digital operates the pool under a service agreement (claims 13335, 13440). NYPA supplies an initial 90 MW to Lake Mariner (claims 12842, 12890). The record contains no MOU, LOI, 'discussions' or unnamed 'anchor customer' language. The only soft phrasing is NYPA's 'potential to expand' (claim 12842), and the 58,000-miner delivery schedule (claims 14247, 14123), which the 2022 filings themselves preface with 'Despite securing', signalling delivery shortfalls.

Concentration has narrowed rather than diversified. The Talen joint venture began as a 300 MW project with 150 MW attributable to TeraWulf (claim 14112). The August 2022 amendment cut TeraWulf's stake to 33.33% with a 20,000-miner contribution (claims 13713, 13714). By March 2023 the stake was 25%, alongside one of five board seats (claims 12938, 12860, 12874). Operational dependence sits on a small set of parties: NYDIG for both custody and execution, and Foundry as the sole disclosed pool. Related-party ties to Beowulf E&D and NovaWulf (claims 13718, 12942, 13849) add affiliate concentration.

The disclosed figures cannot be reconciled with any revenue. The only numbers on record are cash outflows: a $19,072 thousand joint-venture investment in Q1 2022 (claim 13918) and a $339,200 transition-services fee paid by Nautilus to Beowulf E&D (claim 12942). Neither ties to a contracted revenue stream. Revenue for this business has to be modeled from hashrate, power cost and bitcoin price, with JV economics flowing through biweekly hashrate-weighted distributions (claim 12939). It cannot be derived from the counterparty disclosures in this set, which ends mid-record at claim 12853.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense. The company is a self-mining bitcoin operator: the 10-K language says the majority of revenue comes from self-mined bitcoin earned through a pool operated by Foundry Digital (claims 13440, 13346), held by NYDIG under a March 10, 2022 custodial agreement (13412, 13333), and sold through NYDIG Execution under a September 16, 2022 execution agreement (13414, 13345). These are executed, named service contracts, but they are vendor relationships, not revenue-bearing offtake. The equipment side is similarly concrete: Minerva and Bitmain are named counterparties to signed purchase agreements (14028, 14027, 13977), expanded by March 2023 to include five additional Bitmain future sales agreements (12847). The 58,000-miner delivery figure (14247, 14123), however, is a delivery schedule under non-fixed-price contracts, not a guarantee of installed capacity. Claim 21075, an IKONICS auditor ratification from 2006, is legacy shell-company material and should be disregarded.

No counterparty in this record is described anonymously as a strategic partner or anchor customer. Every relationship is identified by name, which is to the company's credit. The weightier disclosures involve insiders. NovaWulf is acknowledged as an affiliate (13849), and Beowulf E&D, which provides operating services to Nautilus, receives $339,200 under a transition services agreement (12942, 13718). Both relationships warrant related-party scrutiny rather than treatment as arm's-length commercial validation.

Concentration has moved through the single Talen joint venture. The original 2021 framing gave TeraWulf 150 MW of a 300 MW build, roughly half (14112, 14120). The August 2022 amendment cut that to 33.33% against a 20,000-miner contribution (13714, 13713), and by March 2023 it stood at 25% with one of five board seats (12938, 12874, 12848). TeraWulf's exposure to the only other site beyond wholly owned Lake Mariner (12840) therefore shrank by half within about eighteen months, while Talen's Cumulus affiliate gained governance control.

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate, or bitcoin-production figures appear, and the only cash datapoint is a $19,072 thousand joint venture investment in Q1 2022 (13918). The fixed nuclear power price at Nautilus (12838, 13 and 14113) supports cost modeling, but not top-line verification. The source list is also truncated mid-entry at claim 12867 in 2023, so any later contracts fall outside this assessment.

On the record supplied, TeraWulf has no signed customer agreements at all. Every named counterparty is a supplier, a joint-venture partner, a service provider or a capital-markets agent. That is what you would expect from a self-mining bitcoin operator whose revenue, by its own description, comes predominantly from self-mined bitcoin (claim 13441). The earliest item, the McGladrey & Pullen auditor ratification (claim 21075), belongs to the IKONICS shell that TeraWulf later merged into, and it says nothing about commercial traction. The Cantor Fitzgerald, B. Riley and D.A. Davidson sales agreement (claim 13973) is an at-the-market equity facility, and the January 2023 voting and support agreement (claim 13391) is a governance instrument. Neither should be read as a partnership. The useful finding is that every counterparty is identified by name; none is described as an unidentified strategic partner or anchor customer. Bitmain and Minerva are executed purchase agreements, though the Bitmain contracts are explicitly non-fixed price (claims 14027, 14028, 12847). The 58,000-miner delivery figure is framed as delivery schedules, not received units (claims 14123, 14247). Foundry is the pool operator (claims 13346, 13440), and NYDIG serves as custodian and sales executor under dated agreements (claims 13333, 13334). Talen, through Cumulus Coin, is the joint-venture counterparty (claims 12841, 12848). Concentration has moved in the wrong direction for TeraWulf on its most important arrangement. The March 2022 framing implied 150 MW of a 300 MW build, or roughly half (claim 14112). The August 2022 amendment cut TeraWulf's equity to 33.33% (claim 13714), and by March 2023 it stood at 25% against Talen's 75%, with one of five board seats (claims 12872, 12874). Related-party exposure runs alongside this. Beowulf E&D, an affiliate, is paid $339,200 by Nautilus for transition services (claim 12942), and NovaWulf is disclosed as an affiliate (claim 13849). Neither relationship is arm's-length validation. Revenue arithmetic cannot be reconciled from this record because no revenue, hashrate or bitcoin-production figures are disclosed. The only quantified flows are a $19,072 thousand joint-venture investment in Q1 2022 (claim 13918), and Nautilus distributions that are pro rata to hash contribution, net of power and operating costs (claim 12939). Those distributions depend on the fixed-price nuclear power contract (claim 12838). The record is truncated after March 2023, so later announcements, including any hosting or compute customer contracts, are not evaluated here.

TeraWulf's record through early 2023 contains no customer agreements in the conventional sense, so the signed-versus-aspirational test applies to suppliers, joint-venture partners, service providers and related parties rather than to buyers of a product. Revenue is described as self-mined bitcoin earned through a pool operated by Foundry Digital (claims 13440, 13346), with bitcoin sales handled by NYDIG Execution (13414, 13345) and custody by NYDIG under a dated March 10, 2022 agreement (13412, 13333). These counterparties are named and documented with execution dates, but they are service vendors, not revenue sources. The earliest entry, claim 21075, is an IKONICS auditor ratification from 2006 that has no bearing on TeraWulf's commercial book and should be disregarded.

Nearly every relationship on file is identified by name and described as executed, not framed as an MOU, LOI or 'discussion.' The Bitmain and Minerva equipment contracts are named (14246, 13977), though the Bitmain contracts are 'non-fixed price' (14051), which leaves pricing open. The 58,000-miner delivery figure appears in a sentence beginning 'Despite securing' (14123), which reads as a prelude to a shortfall rather than a confirmation of supply. The Nautilus expansion to 300 MW is an option held by the partner, not a commitment (12922). The anonymous labels in this record come from TeraWulf's own legal structure, such as 'Cumulus Member' and 'TeraWulf Member,' rather than from any unidentified anchor customer.

Concentration has moved through the single Talen-affiliated joint venture, and TeraWulf's share has shrunk. That share went from 150 of 300 MW (14112) to a 33.33% equity interest (13714) and then to 25% with one of five board seats (12938, 12874). Several dependencies run through insiders. Beowulf E&D, which is controlled by the CEO, provides services to both TeraWulf and Nautilus (12845, 12942), and NovaWulf is an affiliate (13849). The power contract at the core of the site's economics is a fixed-rate supply from Susquehanna (12838).

The revenue figures cannot be reconciled with the deals on this record. The only disclosed number is a Q1 2022 joint-venture investment outflow of $19,072 thousand (13918). No pool revenue, hosting income or per-counterparty revenue is cited that could be tied back to the announced agreements. The claim list is also truncated at claim 12891, so later announcements, including any hosting or HPC customer contracts, are outside the scope of this read.

TeraWulf's record through early 2023 contains essentially no signed revenue customers. That is expected for a self-mining bitcoin operator, but it matters for anyone reading these filings as commercial traction. The documented counterparties are executed agreements, but they sit on the cost and infrastructure side. Bitmain and Minerva are equipment suppliers under four non-fixed-price purchase agreements and one purchase agreement (claims 14051, 14028, 13977). Foundry Digital operates the mining pool under a click-through Service Agreement (claims 13346, 13335). NYDIG provides custody and NYDIG Execution sells bitcoin, under agreements dated March 10 and September 16, 2022 (claims 13412, 13345). Cantor, B. Riley and D.A. Davidson are sales agents on an equity program, not customers (claim 13973). Claim 21075, a 2006 auditor ratification for IKONICS, belongs to the pre-merger shell and should be excluded.

The most consequential relationship is the Nautilus joint venture with an affiliate of Talen Energy (claims 14120, 14112). It is a signed agreement, but it is an equity partnership, not a customer contract. Its terms have moved against TeraWulf. The March 2022 framing attributed 150 MW of up to 300 MW to TeraWulf (claim 14112). By August 2022 TeraWulf's stake was 33.33% against Cumulus's 66.67% (claim 13714), and by March 2023 it was 25% versus 75% (claims 12938, 12872). TeraWulf also holds only one of five board seats (claim 12874). The headline language therefore overstates TeraWulf's economic share relative to the later operative terms.

Two related-party flows deserve flagging. Beowulf E&D, which is controlled by the CEO, supplies administrative, infrastructure and transition services, including $339,200 payable by Nautilus (claims 12845, 12942). NovaWulf is disclosed as an affiliate (claim 13849). Neither is an arm's-length commercial validation.

Concentration has stayed narrow, centred on Talen and Cumulus for power and site, Bitmain for hardware, and Foundry and NYDIG for monetization. The "58,000 miners" figure (claims 14247, 14123) is a supply commitment, and the filing introduces it with "despite securing," which implies delivery shortfall rather than deployed capacity.

Revenue arithmetic cannot be reconciled from this record. No customer-level revenue is disclosed, because revenue is pool payouts converted through NYDIG (claim 13441). The only hard figures are cash outflows: the $19,072 joint-venture investment (claim 13918), the fixed $0.02 nuclear power price (claim 12838), and the transition-services fee. The input was truncated at claim 12891, so later claims, including any post-2023 hosting or HPC tenant announcements, were not assessed.

TeraWulf's record through early 2023 contains no signed revenue customer in the conventional sense, and that is the main finding. As a self-mining bitcoin operator, its revenue comes from block rewards paid through a pool, so the named counterparties are suppliers, service providers, joint-venture partners and affiliates rather than buyers. Investors should not read the volume of agreements as commercial traction. The supply side is concretely documented. Executed agreements with Bitmain (four non-fixed-price purchase agreements) and Minerva are named repeatedly (claims 14246, 14051, 14028, 13977). The headline commitment of 58,000 miners for delivery during 2022 (claims 14247, 14123) arrived alongside an admission that delivery was not assured: the operative clause begins "Despite securing." The operating stack is also signed and dated. It includes the Foundry pool service agreement (claims 13346, 13335), the NYDIG custodial agreement of March 10, 2022 (claims 13412, 13161) and the NYDIG Execution agreement of September 16, 2022 (claims 13414, 13345). Foundry is effectively the sole revenue conduit (claim 13440), which is the real counterparty concentration here. The Talen joint venture is a named, executed agreement dated May 13, 2021 (claims 14120, 14112). However, its terms moved against TeraWulf. Its equity interest fell from 33.33% (claim 13714) to 25% (claims 12938, 12872), its board representation is one of five managers (claim 12874), and Talen's Cumulus side held a unilateral 100 MW expansion option (claim 12922). The fixed $0.02 nuclear power price (claim 12838) is a cost input, not revenue. Related-party exposure is material. Beowulf E&D, which the CEO controls, supplies administrative, infrastructure and transition services (claims 12845, 13718, 12942), and NovaWulf is disclosed as an affiliate (claim 13849). Two items are not commercial relationships at all: the Cantor/B. Riley/D.A. Davidson sales agreement is an equity distribution facility (claim 13973), and claim 21075 is a legacy IKONICS auditor ratification. Revenue reconciliation cannot be performed from this record. Apart from the $19.1 million joint-venture investment line (claim 13918) and the $339,200 transition fee (claim 12942), no dollar figures attach to the announced deals. Converting the 20,000 contributed miners (claim 13713) or the 58,000-unit commitment into hashrate and revenue would require delivery, energization and bitcoin price data not present here. The claim list is truncated at claim 12891, so later developments, including any hosting or HPC customer contracts, are outside this read.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, because the company sells no product or service to identified buyers. Its revenue comes from self-mined bitcoin earned through a Foundry Digital mining pool and held by NYDIG (13440, 13441, 13346). Every named counterparty is a supplier, joint-venture partner, service provider, financing agent, or related party. Claim 21075 concerns the predecessor shell IKONICS ratifying an auditor in 2006 and has no commercial bearing. Claims 13973 (an at-the-market sales agreement with Cantor, B. Riley and D.A. Davidson) and 13391 (a shareholder voting agreement) are capital-markets and governance documents, not commercial ones.

The executed, dated agreements are the NYDIG custodial agreement of March 10, 2022 (13412, 13344), the NYDIG Execution agreement of September 16, 2022 (13414, 13345), the Foundry pool service agreement (13335), the Minerva purchase agreement, and four Bitmain purchase agreements (13977). The Bitmain contracts are expressly "non-fixed price" (14051, 14027), so price and volume are softer than "secured" (14246) implies. The 58,000-miner delivery figure (14247) is introduced with "Despite securing" (14123), which signals a delivery shortfall rather than performance. The Talen relationship rests on a signed JV agreement dated May 13, 2021 (14120). The early framing of "up to 300 MW" with 150 MW attributable to TeraWulf (14112) was aspirational capacity language.

On naming, Talen, Bitmain, Minerva, Foundry and NYDIG are all identified. The counterparty that actually controls Nautilus appears only as "Cumulus Member" or "Cumulus Coin" (13714, 12921). Beowulf E&D (13718, 12942) and NovaWulf (13849) are affiliates, so those arrangements are related-party rather than arm's-length.

TeraWulf's position in its one material partnership has diluted steadily. The implied 50% share (14112) fell to 33.33% in August 2022 (13714, 12920) and then to 25% (12872, 12938, 12921). Governance followed, with one of five board seats (12874), and the 25% remains "subject to adjustment based on relative capital contributions." The miner contribution was cut to 20,000 (13713).

The revenue arithmetic cannot be reconciled against deals, because no announced agreement carries a revenue figure. The disclosed numbers are cost-side: a $19,072 thousand JV investment in Q1 2022 (13918), a $339,200 transition-services fee paid to an affiliate (12942), and fixed-price Susquehanna nuclear power (12838). Economics flow back through bitcoin distributions net of power and operating costs (12939). This record ends in March 2023 and does not cover any later announcements.

TeraWulf's record through March 2023 contains no customer agreements at all. The company self-mines bitcoin, so every named counterparty is a supplier, a joint-venture partner, a service provider or a related party. The nearest thing to a revenue counterparty is the Foundry Digital mining pool (claims 13440, 13335). Its payouts are the source of the "majority of our revenue" (13441). The earliest entry, claim 21075, is an IKONICS auditor ratification carried over from the predecessor shell and has no commercial weight. The good news for diligence is that nothing on record hides behind 'strategic partner' or 'anchor customer' language. Bitmain, Minerva, Talen, Susquehanna, NYPA, Foundry and NYDIG are all identified, and the relationships are described as executed agreements rather than MOUs or discussions (13977, 14120, 13412, 13414, 12842). The caveats sit inside those signed documents. The four Bitmain contracts are "non-fixed price" (14051). The 58,000-miner delivery commitment is introduced with "Despite securing" (14123), which signals a shortfall against the contracted schedule. Concentration has tightened rather than diversified. The Talen venture began at 150 MW of a 300 MW build, roughly half (14112). An August 2022 filing cites 33.33% (13714), and the amended structure then settled at 25% (12920, 12921, 12938). TeraWulf now appoints one of five managers (12874), so its most visible growth asset is governed by its counterparty. Monetization runs through a single chain: Foundry for pool rewards, NYDIG for custody (13412) and NYDIG Execution for sales (13414). Related-party exposure is also material. Beowulf E&D provides infrastructure and operating services to Nautilus (13718, 12942), NovaWulf is an affiliate (13849), and insiders signed a voting and support agreement (13391). Revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-to-revenue or per-agreement economics are disclosed. The only figures are a Q1 2022 joint-venture outlay of 19,072 (in thousands; 13918) and a $339,200 transition-services fee (12942). Both are costs, not revenue. The 20,000 miners contributed to Nautilus (13713) against 58,000 contracted units could anchor a capacity model, but only if delivery actuals were disclosed, and they are not. The record also stops in March 2023, so it says nothing about any later diversification of counterparties.

TeraWulf's record contains no signed customer agreements in the conventional sense, and that is the most important point for diligence. Every named, executed contract is a supplier, service or joint-venture arrangement. Revenue comes from self-mined bitcoin paid out through a pool, not from contracted offtake (claims 13441, 13440). The language throughout is that of executed agreements rather than MOUs, LOIs or 'discussions'. Bitmain and Minerva equipment agreements are described as executed (14028, 13977). NYDIG custody and NYDIG Execution agreements carry specific dates, March 10, 2022 and September 16, 2022 (13412, 13414). Foundry's pool Service Agreement is entered into by accessing the pool (13346), and the NYPA power arrangement is described as secured (12842). Nothing is attributed to unnamed 'strategic partners' or 'anchor customers'. Every counterparty is identified: Talen, Bitmain, Minerva, NYDIG, Foundry, NYPA and Beowulf E&D. NovaWulf is disclosed as an affiliate (13849). One entry, claim 21075, is an IKONICS auditor ratification from 2006. It is legacy shell-company material, not a TeraWulf relationship, and should be excluded. Concentration has shifted rather than diversified. The Talen joint venture began as up to 300 MW, of which 150 MW was attributable to TeraWulf (14112). The August 2022 amendment then cut TeraWulf to 33.33% (13714), and it was 25% by early 2023 (12872, 12860). Cumulus appoints four of five Nautilus managers (12874), so TeraWulf's economic exposure to that site now sits under a counterparty-controlled board. On the revenue side, Foundry is the single disclosed pool operator (13162, 12854), and NYDIG handles both custody and sale execution. That puts pool payout and liquidity on two counterparties. The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate or bitcoin-production figures are disclosed. The 58,000 miners contracted for 2022 delivery (14247, 14123) sit against a 20,000-miner contribution to Nautilus (13713), and no delivered count, installed count or Lake Mariner count closes the gap. The quantified flows that are disclosed are all outbound. They are a $19,072 thousand joint-venture investment (13918), an $8.46 million annual Beowulf E&D base fee (12867) and a $339,200 transition fee (12942). The supplied record also appears truncated at claim 12890, so later hosting or customer announcements, if any, are not assessed here.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense, signed or otherwise, and no MOU, LOI or 'discussions' language either. As a self-miner, the company earns bitcoin through a pool rather than from paying customers. Every named counterparty is a supplier, service provider, capital-markets agent or joint-venture partner, and each is described as an executed agreement. Equipment came from Bitmain under four non-fixed-price purchase agreements and from Minerva under one (claims 14051, 14052, 13977), with 58,000 miners scheduled for 2022 delivery (14247, 14123). Foundry Digital operates the mining pool (13346, 13335). NYDIG holds the bitcoin in custody under a March 2022 agreement and executes sales under a September 2022 agreement (13344, 13345). Cantor, B. Riley and D.A. Davidson are sales agents (13973). NovaWulf is disclosed as an affiliate (13849). Every counterparty is identified by name, and none is described only as an unnamed 'strategic partner' or 'anchor customer.' One record is misattributed: claim 21075 concerns IKONICS Corporation's 2006 auditor ratification and should be excluded from this file. Concentration shows up in the Talen relationship rather than in customers. The Nautilus joint venture at Susquehanna was first presented as 300 MW with 150 MW attributable to TeraWulf (14112). An August 2022 amendment cut TeraWulf's interest to 33.33% with a 20,000-miner contribution (13714, 13713, 12920). The March 2023 restatement cut it again to 25%, with TeraWulf appointing one of five managers (12872, 12874, 12921). The company's economic exposure to its flagship nuclear-powered site therefore halved in a year, and its governance stake is a minority one. Bitcoin distributions follow each member's contributed hash rate after costs (12939), so Nautilus output depends on hash rate rather than on any offtake contract. The revenue arithmetic cannot be reconciled from this record. The only figure disclosed is a $19,072 thousand joint-venture investment in Q1 2022 (13918), which is capital deployed, not revenue. No revenue totals, hash-rate yields or pool-payout figures are on file to test against the 58,000-miner commitment or the reduced Nautilus share. The record also ends in March 2023, so any later hosting or compute arrangements would need separate verification.

On this record TeraWulf has no signed customer agreements, and none are claimed. Through March 2023 the company presents itself as a self-miner. Revenue comes from bitcoin it mines itself through a pool operated by Foundry Digital (claims 13440, 13335, 12903). Every named counterparty is therefore a supplier, service provider, joint-venture partner, related party or financing agent, not a buyer. The record contains no MOU, LOI or 'discussions' language, and no unidentified 'anchor customers' or 'strategic partners'. Each counterparty is named, and each agreement is described as executed. Claim 21075, an IKONICS auditor ratification from 2006, comes from the predecessor shell and is not a commercial relationship.

The firmness of these agreements varies more than the wording suggests. The Bitmain contracts are described as 'non-fixed price' (claims 14051, 14027), which leaves pricing and likely volume open. The 58,000-miner delivery commitment for 2022 is introduced with 'Despite securing' (claim 14123), which points to a shortfall against plan rather than delivery. The March 2023 filing adds five 'future sales and purchase agreements' with Bitmain (claim 12899), extending forward commitments without showing that earlier ones were fulfilled.

Concentration moved against TeraWulf in its main partnership. The Talen venture was first framed as up to 300 MW, with 150 MW attributable to TeraWulf, roughly an even split (claim 14112). By August 2022 TeraWulf's stake was 33.33% (claim 13714), and by March 2023 it was 25%, with one of five board seats (claims 12872, 12874, 12889). Economics and control both shifted to Cumulus Coin, a Talen subsidiary. Operating dependence also narrowed to a few providers: Foundry for pool revenue, and NYDIG for custody, execution and a deposit-control arrangement (claims 13412, 13414, 12905). Related parties appear as well. Beowulf E&D is paid $339,200 under a transition services agreement (claim 12942), and NovaWulf is disclosed as an affiliate (claim 13849).

Revenue cannot be reconciled from this record. The only dollar figures are the $19,072 thousand joint-venture investment outflow (claim 13918) and the Beowulf fee, and both are costs. The record gives no hashrate, bitcoin-production or revenue figures to test against the miner counts or the equity share. The claim list is also cut off after claim 12889, so later agreements, including any move toward third-party hosting customers, are not covered here.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. The company is a self-miner whose revenue comes from its own mined bitcoin (claim 13441). What the filings describe as agreements are supply, joint-venture, custody and service contracts, and should be read that way. Claim 21075, an auditor ratification for IKONICS, the predecessor shell, is not a commercial relationship at all.

The named counterparties are specific and backed by executed documents. The equipment suppliers are Bitmain, under four non-fixed-price purchase agreements, and Minerva, under an executed purchase agreement (claims 14051, 14028, 13977). Together these covered 58,000 miners for 2022 delivery (claims 14247, 14123). Because the Bitmain contracts are non-fixed-price, the unit count is firmer than the dollar commitment. Foundry Digital operates the mining pool under a Service Agreement (claims 13335, 13440). NYDIG provides custody under a dated March 10, 2022 agreement (claim 13412), and NYDIG Execution handles sales under a September 16, 2022 agreement (claim 13414). None of these is framed as an MOU or 'discussions,' and no unnamed 'anchor customer' or 'strategic partner' appears anywhere in the record.

The relationship that has drifted is the Talen joint venture. Early disclosure described up to 300 MW, with 150 MW attributable to TeraWulf (claim 14112), which implies parity. By August 2022, TeraWulf's equity in Nautilus stood at 33.33% (claim 13714). By March 2023 it was 25%, with one of five board seats (claims 12872, 12874, 12840). Talen's Cumulus subsidiary alone holds the 100 MW expansion option (claim 12922). Concentration therefore tightened toward a single, increasingly dominant partner. The related-party layer adds to this. Beowulf E&D is owned by the CEO and provides administrative, infrastructure and transition services, the latter for $339,200 (claims 12845, 12942). NovaWulf is also an affiliate (claim 13849).

The revenue arithmetic cannot be reconciled from this record. The only quantified figure is a $19,072 (thousand) joint-venture investment outflow in Q1 2022 (claim 13918). The record gives no revenue totals, hash-rate output, or Nautilus distribution amounts to test against the 58,000-miner commitment or the dilution in the stake. Nautilus distributions follow hash-rate contribution net of power and operating costs (claim 12939). That means TeraWulf's economic share may diverge from its nominal equity percentage. A diligence reader should ask for delivered-miner counts and for Nautilus bitcoin distributions by period before crediting the announced capacity. The record ends in early 2023, so any later hosting or compute counterparties are outside this analysis.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. As a self-mining bitcoin producer, its revenue comes from block rewards routed through a pool, not from an identified buyer. What the record holds instead are supply, joint-venture, service and financing contracts. The first step is therefore reclassification rather than a signed-versus-MOU split. Claim 21075 predates TeraWulf entirely: it is an IKONICS auditor ratification carried over from the reverse-merger shell and should be excluded. Claim 13973 is an at-the-market sales agreement with Cantor, B. Riley and D.A. Davidson, and claim 13391 is a shareholder voting agreement. Both are capital-markets items, not commercial relationships.

On naming, the file is unusually clean. Every operative counterparty is identified: Bitmain and Minerva as miner suppliers (14051, 14028), Talen as JV partner and Susquehanna as power source (14120, 14113), Foundry as pool operator (13335), and NYDIG as custodian and execution agent (13412, 13414). No unnamed 'anchor customer' or 'strategic partner' appears. The softer language sits in the terms, not the names. The Bitmain agreements are 'non-fixed price' (14051), so the headline 58,000-miner delivery commitment (14247, 14123) did not lock in pricing. Later filings introduce that figure with 'despite securing' (14123), which signals delivery shortfall rather than fulfilment. Beowulf E&D (13718, 12942) and NovaWulf (13849) are disclosed affiliates, and their services should be read as related-party arrangements, not arm's-length validation.

Concentration has narrowed rather than diversified. TeraWulf's attributable share of Nautilus fell from 150 of 300 MW, roughly half (14112), to 33.33% in August 2022 (13714), then to 25% in March 2023 (12860, 12872). It now appoints one of five board managers (12874). Bitcoin distributions track hash-rate contribution net of power and operating costs (12939), which leaves economics at Nautilus increasingly set by Talen's Cumulus entity. Downstream, pool, custody and execution all run through two firms, Foundry and NYDIG.

The revenue arithmetic cannot be reconciled from this record. The only disclosed figure is a $19,072 thousand joint-venture investment outflow in Q1 2022 (13918), which is a cost, not revenue. No hashrate-to-revenue bridge, miner energization count against the 58,000 commitment, or Nautilus distribution totals are disclosed here. The record also stops at March 2023 (the 12906 entry is truncated), so any later hosting or compute agreements would need separate verification before being characterised as signed customer revenue.

TeraWulf's record through March 2023 contains no signed revenue customer at all, and that is the central finding. Every named counterparty is a supplier, a power provider, a joint-venture partner, a service vendor, a financing agent or an affiliate. The company earns revenue by mining bitcoin for itself and selling it, so the documents that look like commercial agreements are cost-side contracts, not demand-side ones.

The cost-side agreements are genuinely executed and fully named. They include the Bitmain and Minerva equipment purchase agreements (claims 13977, 14027, 14028), the Talen joint venture dated May 13, 2021 (claim 14120), and the Susquehanna nuclear power supply at a fixed rate (claims 14129, 12838). They also include the NYPA agreement for an initial 90 MW (claim 12842), the dated NYDIG custody and execution agreements (claims 13412, 13414), the Foundry pool service agreement (claim 13335) and the Cantor, B. Riley and D.A. Davidson sales agreement (claim 13973). The softer language sits around these contracts rather than inside them. The four Bitmain contracts are non-fixed price, and the NYPA capacity carries only a potential to expand. The later description of a strong, collaborative relationship with Bitmain (claim 12891) is relationship language, not a contract. The one unnamed demand-side reference is bitcoin earned from hosting arrangements (claim 12852), which appears without any counterparty identified. Two related parties also sit inside the structure: NovaWulf is an affiliate (claim 13849), and Beowulf E&D provides services to Nautilus for $339,200 (claim 12942).

Concentration has shifted away from TeraWulf over time. At the outset, 150 of 300 MW were attributable to TeraWulf (claim 14112). That became a 33.33% equity interest (claim 13714) and then 25% against Cumulus's 75%, with one of five board seats (claims 12938, 12874). The 58,000-miner supply commitment (claims 14247, 14123) is itself framed with the word despite. Only 20,000 of those miners were earmarked for Nautilus (claim 13713).

The revenue arithmetic cannot be reconciled from this record. No revenue figure, hashrate output or hosting income is disclosed that could be tied to the announced agreements. The only quantified flows are costs: joint-venture investments of 19,072 (claim 13918), the Beowulf fee and the fixed power price. Readers should treat the announcements as evidence of a secured supply chain, not of contracted demand.

TeraWulf's record through March 2023 contains no signed customer agreement in the conventional sense. Every named counterparty is a supplier, service provider, power source, financing agent or joint-venture partner, and revenue is described as coming mainly from self-mined bitcoin (claim 13441). The one reference to third-party revenue, bitcoin "earned from hosting arrangements" (claim 12852), names no hosting customer and gives no volume, so it should be read as undisclosed rather than contracted.

The named relationships are mostly documented as executed contracts. Bitmain and Minerva moved from "structured and secured" language (claim 14246) to specific executed instruments: four non-fixed-price purchase agreements with Bitmain and one with Minerva (claims 14027, 14028, 13977). The 58,000-miner delivery figure (claims 14247, 14123) is a supply commitment, not a delivery record, and the "Despite securing" framing in claim 14123 implies the schedule was not being met. NYDIG custody (claim 13412) and NYDIG Execution (claim 13414) are dated agreements. Foundry's pool relationship (claim 13335) is a click-through service agreement. Power supply is named on both sites: Susquehanna for Nautilus at a fixed rate (claims 14113, 12838) and an initial 90 MW from NYPA at Lake Mariner (claim 12842), with expansion described only as potential. Claim 12891, framed as a "strong, collaborative relationship" with Bitmain, is relationship language and adds nothing contractual beyond the earlier purchase agreements.

Concentration has shifted within the Talen joint venture rather than across counterparties. TeraWulf's Nautilus stake fell from 33.33% (claim 13714) to 25% (claims 12938, 12872), and that 25% remains subject to adjustment based on capital contributions. TeraWulf appoints one of five managers (claim 12874), and the 20,000-miner contribution (claim 13713) now sits in a vehicle it does not control. Output from Nautilus arrives as biweekly bitcoin distributions net of power and operating costs (claim 12939), so exposure to Talen is operational and governance-driven, not a customer receivable.

The revenue arithmetic cannot be reconciled from this record because no revenue, hashrate or bitcoin-production figures are disclosed. The only dollar amounts are a $19,072 joint-venture investment outflow for Q1 2022 (claim 13918) and a $339,200 transition-services fee paid by Nautilus to the related party Beowulf E&D (claim 12942). Both are cost-side items. The record also ends in March 2023 and is truncated at claim 12891, so it captures no later customer contracting.

TeraWulf's record in this window contains no identified revenue customers. Nearly every named counterparty is a supplier, financier, custodian, or joint-venture partner. The company says most revenue comes from self-mined bitcoin paid out through Foundry's pool (13441, 13440), so any customer-concentration read here is really a supplier and partner concentration read.

Several agreements with named counterparties are clearly executed. These include the Bitmain purchase agreements, described as four non-fixed price sales and purchase agreements (14051, 14027, 13977), and the Minerva purchase agreement (14052, 14028). The Talen joint venture dates to May 13, 2021 (14120). The Susquehanna nuclear power arrangement runs five years with renewal options (14113, 14129) and was later described at a fixed rate (12838). TeraWulf also has a NYDIG custody agreement dated March 10, 2022 (13412) and a NYDIG Execution agreement dated September 16, 2022 (13414), plus a Foundry service agreement (13335). The Cantor, B. Riley and D.A. Davidson sales agreement (13973) is capital-raising, not commercial. The Beowulf E&D services and the NovaWulf fund are related-party arrangements (13718, 12942, 13849).

Softer language sits beside these contracts. The NYPA deal secures an initial 90 MW with only the "potential to expand" (12842, 12890). Talen was framed as a partnership for "up to 300 MW" (14112). The record cuts off at a "strategic relationship" with Bitmain (12891), which should not be read as a separate contract. Hosting arrangements appear in passing (12852), but no hosting client is named anywhere in the record.

Concentration has shifted toward Talen. TeraWulf's attributable Nautilus capacity was first presented as 150 of 300 MW (14112). Its stake then fell to 33.33% (13714) and later to 25%, with TeraWulf appointing one of five managers (12938, 12872, 12874). TeraWulf still committed 20,000 miners to Nautilus (13713), and distributions depend on hash rate net of costs (12939). The company's economic exposure has narrowed while its dependence on Talen's governance has grown.

The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed. The only financial datapoint is a $19,072 joint-venture investment for Q1 2022 (13918). The 58,000-miner delivery schedule for 2022 (14247, 14123) is qualified by the phrase "despite securing," which hints at a shortfall. Subtracting the 20,000 Nautilus miners leaves roughly 38,000 units against an initial 90 MW at Lake Mariner (12842). Whether those units could all be energized cannot be confirmed without efficiency data.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense, because a self-mining bitcoin operator sells hashrate into a pool rather than to contracted buyers. Every counterparty on file is a supplier, joint-venture partner, service provider or financial intermediary. That record is also unusually free of soft language. There is no MOU, LOI, 'discussions' or unnamed 'anchor customer' phrasing, and each relationship is described as executed and attributed to a named entity.

The named counterparties fall into four groups. Equipment suppliers are Bitmain, under four non-fixed-price purchase agreements, and Minerva, under one executed purchase agreement (claims 14246, 14051, 14028, 13977). The joint-venture partner is Talen Energy, through its Cumulus affiliate, at the Nautilus facility, with nuclear power supplied from Susquehanna Station (claims 14120, 14113, 12919). Operating intermediaries are Foundry as pool operator and NYDIG as custodian and execution agent (claims 13440, 13412, 13414). Related parties are Beowulf E&D, which is owned by the CEO, and NovaWulf, an affiliate (claims 12845, 13849).

The firmest qualifier concerns the equipment agreements, not their counterparties. The 58,000-miner delivery schedule is introduced with the word 'despite' (claim 14123), which signals under-delivery against the headline commitment. The Bitmain agreements are also explicitly non-fixed-price (claim 14027), so they are not firm on price.

Concentration moved away from TeraWulf's own economics at Nautilus. The original structure attributed 150 MW of 300 MW to TeraWulf (claim 14112). The August 2022 amendment set TeraWulf's interest at 33.33% with a 20,000-miner contribution (claims 13714, 13713), and the March 2023 restatement cut it to 25% with one of five board seats (claims 12872, 12874, 12920). Talen's affiliate also holds a unilateral option, which it alone would fund, to expand capacity by 100 MW (claim 12922). Revenue and custody concentration sit with Foundry and NYDIG, which handle substantially all mined bitcoin flow (claim 13441).

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed. The only dollar amounts are a $19.1 million joint-venture investment outflow in Q1 2022 (claim 13918) and a $339,200 transition-services fee payable to the CEO-controlled Beowulf E&D (claim 12942). Nautilus distributions are netted of power and operating costs before bitcoin reaches TeraWulf (claim 12939), so gross-to-net attribution needs audited segment data. The claim record breaks off mid-entry at claim 12906, which also limits coverage after March 2023.

TeraWulf's record through March 2023 contains no revenue customer in the conventional sense, so the signed-versus-soft distinction applies to its supply chain, joint venture, and service providers. The company states that the majority of its revenue comes from self-mined bitcoin (claim 13441). Every named counterparty is a vendor, partner, or intermediary rather than a buyer of its output. The language is also consistently executed rather than exploratory. The Minerva agreement and four Bitmain purchase agreements are described as executed (claims 13977, 14028), and the NYDIG custody and execution agreements carry specific signing dates (claims 13412, 13414). The Talen joint venture is dated May 13, 2021 and was amended and restated on August 27, 2022 (claims 14120, 12901). The record contains no MOU, LOI, or 'discussions' language. The softer phrasing sits inside signed deals: 'up to 300 MW' at Nautilus (claim 14112), NYPA's 'potential to expand' beyond an initial 90 MW (claims 12842, 12890), and equity stakes 'subject to adjustment' (claim 12938). Every counterparty is named: Bitmain, Minerva, Talen, NYPA, NYDIG, Foundry, and the Cantor, B. Riley, and D.A. Davidson sales agents (claim 13973). The exception is hosting. Bitcoin earned from 'hosting arrangements' is referenced, but no hosting client is identified (claim 12852). Related parties also appear, including NovaWulf (claim 13849) and Beowulf E&D, which receives a $339,200 transition-services payment from Nautilus (claim 12942). Concentration has tightened. TeraWulf initially expected 150 of 300 MW at Nautilus (claim 14112), implying parity. Its stake then fell to 33.33% (claim 13714) and later to 25%, with one of five board seats (claims 12872, 12874). Operationally, the company relies on a single pool, Foundry (claim 13440), and a single custodian and execution venue, NYDIG (claims 13441, 13442). Miners come from two suppliers. The disclosed arithmetic cannot be reconciled from this record. The 58,000-miner delivery schedule for calendar 2022 (claim 14247) is introduced with 'Despite securing' (claim 14123), which signals a shortfall that is never quantified. The 20,000-miner contribution to Nautilus (claim 13713) and the $19,072 joint-venture investment (claim 13918, units unstated) are not tied to any disclosed revenue or hashrate figure. The claims supplied end in March 2023 and appear truncated, so later commercial agreements are not assessed here.

TeraWulf's disclosure record through March 2023 contains no signed customer agreements in the conventional sense. Every named counterparty is a supplier, power provider, joint-venture partner, service vendor or financing agent. Revenue comes from self-mined bitcoin paid out through the Foundry pool (claims 13440, 13335), so the relevant diligence question is not who buys from TeraWulf. It is whether the input and venture contracts behind that output are firm. On that measure the language is mostly executed rather than aspirational. Equipment supply agreements are described as executed with Bitmain (four non-fixed-price agreements) and Minerva (claims 13977, 14028). The NYDIG custody and execution agreements carry specific dates (claims 13412, 13414). The Nautilus venture rests on a dated May 2021 agreement and an August 2022 amended and restated agreement (claims 14120, 12901). The softer items are the NYPA supply, framed as an initial 90 MW with only the 'potential to expand' (claim 12842), and the 58,000-miner delivery commitment, which the company introduces with 'despite securing' (claim 14123). That phrasing signals the contracted volumes did not translate into delivered units. Counterparties are generally named, with one partial exception. The Nautilus partner shifts from 'Talen Energy Corporation' (claim 14112) to 'an affiliate of Talen' (claim 14120) and then to 'Cumulus Member' (claims 13714, 12872), so readers should track which Talen entity actually bears the obligations. Several relationships are related-party: NovaWulf is disclosed as an affiliate (claim 13849), and Beowulf E&D provides services to Nautilus for $339,200 plus expenses (claims 13718, 12942). Concentration has moved against TeraWulf within the flagship venture. Its stake went from 150 of 300 MW, roughly half (claim 14112), to 33.33% (claim 13714), then to 25% with one of five board seats (claims 12872, 12874). Pool, custody and sale execution are each concentrated in a single provider, Foundry or NYDIG. The revenue arithmetic cannot be reconciled from this record. No revenue totals appear, only a $19.072 million joint-venture investment outflow (claim 13918), and the record ends in March 2023.

TeraWulf's record through March 2023 contains no signed customer agreements at all. Every named counterparty is a supplier, service provider, joint-venture partner or financing agent. The equipment agreements are executed purchase contracts rather than letters of intent: four non-fixed price Bitmain agreements and one Minerva agreement (claims 14051, 14052, 13977). The same holds for the Foundry pool arrangement, although that Service Agreement was formed simply "by accessing and using the Pool" (claim 13346) and reads more like click-through terms than a negotiated contract. NYDIG's custody and execution agreements are dated March 10 and September 16, 2022 (claims 13412, 13414), and the Cantor/B. Riley/D.A. Davidson Sales Agreement (claim 13973) is an equity-issuance facility, not commercial demand. Unlike many peers, TeraWulf does not lean on unnamed "strategic partners" or "anchor customers." The softer identifications are affiliate-level entities. These include "an affiliate of Talen" and "Cumulus Member" (claims 14120, 12938), NovaWulf, which is described only as "an affiliate of the Company" (claim 13849), and Beowulf E&D, which is paid $339,200 for transition services (claim 12942) but whose relationship to insiders the record does not characterize. That last point warrants a related-party check.

The Talen concentration has shrunk steadily. TeraWulf was first attributed 150 MW of up to 300 MW, roughly half (claim 14112). The August 2022 filing then gave it 33.33% (claim 13714), and the March 2023 filings gave it 25% with one of five board seats (claims 12938, 12874). Claim 12920 dates the reduction to 25% to that same August 2022 amendment, so the contemporaneous 33.33% figure is inconsistent with the later description. The 58,000-miner delivery commitment (claims 14247, 14123) is prefaced by "Despite securing," which implies a shortfall, and only 20,000 miners were slated for Nautilus (claim 13713).

The revenue arithmetic cannot be reconciled from this record, and there is little to reconcile. Revenue is described as predominantly self-mined bitcoin (claim 13441). Nautilus pays out bitcoin in proportion to each member's hash rate, net of costs (claim 12939), which makes it an equity-method asset, not a customer. The only dollar figure is a $19,072 joint-venture investment outflow (claim 13918). Its units are not stated in the quote, and it relates to capital deployed, not revenue. Note that the supplied claim list is truncated mid-entry at claim 12899, so any later customer or lease announcements fall outside this read.

TeraWulf's disclosed record through March 2023 contains no signed customer agreements. Every named counterparty is a supplier, service provider, joint-venture partner, or financing agent. The company says the majority of its revenue comes from self-mined bitcoin (claim 13441), so the counterparty that pays it is effectively the Foundry pool (claims 13440, 13346, 13335), not a contracted offtaker. The agreements that are signed and named are procurement and infrastructure contracts. These are four non-fixed-price purchase agreements with Bitmain and one with Minerva (claims 14051, 14052, 13977), NYDIG custody and execution agreements dated March 10 and September 16, 2022 (claims 13412, 13414, 12904, 12906), and a deposit account control agreement with NYDIG and Wilmington Trust (claim 12905). The Cantor, B. Riley and D.A. Davidson sales agreement (claim 13973) is an equity issuance program, and the voting and support agreement (claim 13391) is a governance matter. Neither is commercial demand. Unlike many issuers, TeraWulf does not rely on unnamed 'strategic partners' or 'anchor customers'; every counterparty in this record is identified. The softer language sits elsewhere. The 58,000-miner commitment is introduced with 'despite securing' (claims 14123, 14247), which points to a delivery shortfall rather than a firm supply position. The Bitmain terms are explicitly non-fixed price. Two service counterparties are related parties: NovaWulf, which the company states is an affiliate (claim 13849), and Beowulf E&D, which provides Nautilus operating services, including a $339,200 transition arrangement (claims 13718, 12942). Counterparty concentration has narrowed over time. At the March 2022 filing, TeraWulf described up to 300 MW with Talen, 150 MW of it attributable to TeraWulf (claim 14112). The August 2022 restatement cut its stake to 33.33% with a 20,000-miner contribution (claims 13713, 13714). Filings from March 2023 then show 25%, one of five board seats, and bitcoin distributions net of power and operating costs (claims 12938, 12872, 12874, 12939, 12920). The company's economic exposure to a single partner has therefore fallen, while its control has weakened. On revenue arithmetic, there is nothing to reconcile, because no deal carries a contract value that could be tested against reported revenue. The only quantified flow is $19,072 of joint-venture investment in Q1 2022 (claim 13918), which is capital going out, not revenue. The record provided ends partway through claim 12899, so any customer contracts disclosed after March 2023 are outside this read.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, because the business sells no product to a named buyer. Revenue is described as 'self-mined bitcoin' (claim 13441), earned through pool participation with Foundry Digital under a click-through Service Agreement 'by accessing and using the Pool' (claims 13346, 13335) and liquidated through NYDIG Execution (claims 13414, 13334). Every counterparty on record is a supplier, service provider, financing agent, joint-venture partner or related party, not a revenue customer. The agreements are signed and dated, however. Bitmain and Minerva equipment contracts are described as executed (claims 14028, 13977, 12899), NYDIG custody dates to March 10, 2022 (claim 13412), and the Talen joint venture dates to May 13, 2021, amended and restated August 27, 2022 (claims 14120, 12901). None of the language is MOU, LOI or 'discussions'. The softer notes are that the Bitmain contracts are 'non-fixed price' and include 'future sales and purchase agreements' (claims 14051, 12899), so pricing and delivery were not locked. The 58,000-miner delivery figure is also introduced with 'Despite securing' (claim 14123), which implies a shortfall against the commitment. Every counterparty is named. There are no unidentified 'strategic partners'. The substantive concentration story is the Talen relationship, which diluted steadily. It began as up to 300 MW with 150 MW 'attributable to TeraWulf' (claim 14112). It fell to a 33.33% stake with a 20,000-miner contribution (claims 13713, 13714). It then fell to 25% with one of five board seats (claims 12872, 12874, 12889). Operating dependence also sits with Beowulf E&D, which is controlled by the CEO. TeraWulf pays it an $8.46 million annual base fee (claims 12845, 12867), and it also provides transition services to Nautilus for $339,200 (claim 12942). NovaWulf is disclosed as an affiliate (claim 13849). The Cantor/B. Riley/D.A. Davidson Sales Agreement (claim 13973) is an at-the-market equity facility and should not be read as commercial traction. The record does not permit revenue reconciliation. No revenue, hashrate or bitcoin-production figures appear, and the only number is a $19,072 thousand Q1 2022 joint-venture investment outflow (claim 13918). The distribution mechanics are defined, with biweekly bitcoin distributions pro rata to hash contributed, net of power and operating costs (claim 12939). The record also ends mid-quote at claim 12903, so later-period announcements are not assessed here.

TeraWulf's record through March 2023 contains no signed customer agreements in the usual sense. The company reports that most of its revenue comes from self-mined bitcoin (claim 13441). Its named counterparties are therefore suppliers, service providers, a joint-venture partner and related parties, not buyers of a contracted product. None of this language is MOU, LOI, framework or pilot language. Every agreement is described as executed or entered into, with dates attached. The equipment agreements are the Minerva purchase agreement and the four non-fixed-price Bitmain agreements (claims 14051, 14028, 13977), later joined by five Bitmain future sales agreements (claim 12899). The rest are the NYDIG custodial and execution agreements (claims 13412, 13414), the Foundry pool service agreement (claim 13346), and the Talen joint venture dated May 13, 2021 (claim 14120). No unnamed 'strategic partner' or 'anchor customer' appears. The disclosures that deserve more scrutiny are the related-party ones. Beowulf E&D, owned and controlled by the CEO, received a reduced annual base fee of $8.46 million from 2023 (claims 12845, 12867). NovaWulf is also described as an affiliate (claim 13849).

Concentration has changed mainly through the Talen relationship, and TeraWulf's position in it has shrunk. The 2022 framing was up to 300 MW, 150 MW of which would be attributable to TeraWulf (claim 14112). The August 2022 restatement moved that to a 33.33% equity interest against Cumulus's 66.67% (claims 13714, 12901). By March 2023 the figures were 25% versus 75%, with one of five board seats (claims 12872, 12874). Economics flow as biweekly bitcoin distributions pro rata to hash rate, net of power and operating costs (claim 12939). This makes Talen a co-owner and cost-sharer rather than a revenue counterparty. Operational dependence is concentrated in Foundry for pool payouts and NYDIG for both custody and sales, since Foundry sends mined bitcoin daily to the NYDIG-custodied wallet (claim 13441).

The revenue arithmetic cannot be reconciled from this record. The 58,000-miner delivery commitment for calendar 2022 (claims 14247, 14123) is stated alongside the word 'despite', which implies a delivery shortfall that is never quantified. Only 20,000 miners are tied to Nautilus (claim 13713). The Bitmain contracts are non-fixed-price, so their capital cost is open. The $19.1 million Q1 2022 joint-venture investment (claim 13918) is not linked to any disclosed output or distribution figure. Readers should also note that the record ends at claim 12903 in March 2023, so any later hosting or compute contracts fall outside this evidence.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense, and none of the MOU, LOI or pilot language this read is meant to flag. The company describes itself as a self-miner whose revenue comes from bitcoin it mines itself and holds in custody (13441). So every named counterparty is a supplier, service provider, financing agent or joint-venture partner, not a buyer of TeraWulf's output.

The counterparties are specifically named and described as executed contracts. They are Bitmain and Minerva for equipment, under four non-fixed-price purchase agreements and one equipment purchase agreement (14246, 13977); Talen for the Nautilus joint venture (14120, 12901); Foundry as pool operator (13346); NYDIG for custody and execution, with dated March and September 2022 agreements (13412, 13414); and Wilmington Trust under a deposit account control agreement (12905). Two relationships warrant related-party attention. NovaWulf is disclosed as an affiliate (13849), and Beowulf E&D is paid $339,200 by Nautilus for transition services (12942). No counterparty on the record is described only as an unidentified "strategic partner" or "anchor customer."

The main change over the period is TeraWulf's shrinking share of its one flagship asset. The original Talen arrangement attributed 150 MW of up to 300 MW to TeraWulf, roughly half (14112). The August 2022 draft cut its equity to 33.33% against a 20,000-miner contribution (13713, 13714). The executed amendment took it to 25%, with one of five board seats (12920, 12938, 12874). Economic dependence on Talen, which supplies both the power (14129) and the controlling equity, therefore rose even as TeraWulf's claim on the facility fell. Downstream, Foundry and NYDIG sit on every mined coin.

The disclosed figures cannot yet be reconciled into revenue. The record shows supply agreements for 58,000 miners slated for 2022 delivery, and the language around that figure is already defensive ("despite securing," 14123). It also shows 20,000 miners committed to Nautilus and a Q1 2022 joint-venture outlay of 19,072 (13918), apparently in thousands. No revenue, hashrate or deployed-fleet figure appears among these claims, so the gap between contracted and installed capacity cannot be tested. The claim list provided also cuts off mid-entry at 12899, in March 2023. Any later shift toward hosting or compute-leasing customers falls outside this read and should be assessed on its own claims.

TeraWulf's record contains almost no customer agreements in the conventional sense. The named contracts are overwhelmingly supply, power, joint-venture, financing and service arrangements. The company's own description is that the majority of revenue comes from self-mined bitcoin, routed through a Foundry-operated pool (13441, 13440). The pool is therefore the closest thing to a paying counterparty, and it is named and papered through a Service Agreement (13346, 13335). The one reference to third-party revenue, 'any bitcoin earned from hosting arrangements' (12852), names no hosting client, prices no hosting capacity and gives no term, so that line of business cannot be diligenced from this record.

On the signed side, the counterparties are identifiable and executed. These are Bitmain and Minerva for equipment (14051, 14052, 13977), though the four Bitmain contracts are expressly non-fixed-price. They also include NYDIG for custody from March 10, 2022 (13412), NYDIG Execution for sales from September 16, 2022 (13414), NYPA for an initial 90 MW (12842), and the Talen joint venture dated May 13, 2021 (14120), whose Nautilus site draws Susquehanna nuclear power on a five-year term with renewals (14113). Softer language surrounds some of this. The 58,000-miner delivery commitment is introduced with 'despite securing' (14123), which signals a shortfall against contract. The later 'strategic relationship' with Bitmain (12891) is relationship framing rather than a new agreement. Two further names, NovaWulf and Beowulf E&D, are affiliates (13849, 13718), so they should be read as related parties, not arm's-length partners.

Concentration has shifted less by adding counterparties than by diluting TeraWulf's position within the one large partnership. Nautilus began as 150 of 300 MW attributable to TeraWulf (14112). That fell to a 33.33% interest in August 2022 (13714), then to 25% with one of five board seats by March 2023 (12872, 12874). Economic control migrated to Talen's Cumulus Member.

The revenue arithmetic cannot be reconciled from these disclosures because the record contains no revenue figures at all. The dollar amounts that do appear are outflows or related-party flows. One is a $19,072 thousand joint-venture investment in Q1 2022 (13918). The other is a $339,200 transition-services fee payable by Nautilus to affiliate Beowulf E&D (12942). Neither supports a customer-revenue base. A reader should treat TeraWulf through this period as a hashrate producer with a named vendor and service stack, not as a business with contracted customers.

The TeraWulf record through March 2023 contains no signed customer agreements. Every named counterparty is a supplier, capital partner, service provider or power source, not a buyer of TeraWulf output. That fits a self-mining model where, as the company states, the majority of revenue comes from self-mined bitcoin (claim 13441). The only reference to third-party revenue is a passing mention of bitcoin earned from hosting arrangements (claim 12852), and it names no counterparty, discloses no contract terms and gives no volume.

The named relationships are well identified and appear to be executed rather than aspirational. They include four non-fixed-price purchase agreements with Bitmain and one with Minerva (claims 14027, 14028, 13977), the Talen joint venture dated May 13, 2021 (claim 14120), the NYDIG custodial agreement of March 10, 2022 (claim 13412), the NYDIG Execution agreement of September 16, 2022 (claim 13414), the Foundry pool service agreement (claim 13346) and an initial 90 MW NYPA supply arrangement (claim 12842). The softer language sits at the edges. The 58,000-miner figure is a delivery provision in supply agreements, not a completed delivery (claims 14123, 14247). The Bitmain strategic relationship is characterized rather than contracted (claim 12891). NYPA capacity carries only a potential to expand beyond 90 MW (claim 12842). NovaWulf is disclosed as an affiliate (claim 13849), so it is a related party rather than an arm's-length partner.

Concentration has moved against TeraWulf within the one material joint venture. The 2022 Nautilus terms gave TeraWulf a 33.33% equity interest against 66.67% for the Talen-affiliated Cumulus member (claim 13714). By March 2023 this had fallen to 25% versus 75% (claims 12872, 12938), with one of five board seats (claim 12874) and distributions tied to relative hash contribution (claim 12939). Execution, custody and sales all run through NYDIG entities (claims 13412, 13414), and pool payouts run through Foundry (claim 13440). That leaves meaningful operational dependence on two service counterparties even though there is no customer concentration in the conventional sense.

The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed alongside the deals. The quantified items are cash outflows or pass-throughs: a Q1 2022 joint venture investment of 19,072 (in thousands, per claim 13918) and a $339,200 transition services fee payable by Nautilus to Beowulf E&D (claim 12942). The supplied record is truncated at claim 12891, so later announcements, including any hosting or compute customers, are outside the scope of this read.

TeraWulf's record through early 2023 contains no signed customer agreements. Every named counterparty is a supplier, joint-venture partner, service provider or related party, which fits a self-mining model in which, as the company states, "the majority of our revenue comes from our self-mined bitcoin" (claim 13441). The record also contains no MOU, LOI, pilot or "discussions" language. What it does contain is repetition: the same executed agreements are restated across successive filings, and that repetition should not be read as deal flow.

The named counterparties are specific. Bitmain and Minerva supply mining equipment, under four non-fixed-price purchase agreements and one purchase agreement respectively (claims 14051, 14052, 13977). The Nautilus joint venture is with Talen, first described generically and later identified as Cumulus Coin, LLC, a Talen subsidiary (claims 14120, 12841). Foundry Digital provides mining-pool services (claim 13346). NYDIG provides custody and NYDIG Execution handles bitcoin sales (claims 13412, 13414). Cantor Fitzgerald, B. Riley and D.A. Davidson are at-the-market sales agents, not commercial partners (claim 13973). The related parties deserve scrutiny: Beowulf E&D is owned and controlled by the CEO and provides services to both TeraWulf and Nautilus (claims 12845, 12942), and NovaWulf is disclosed as an affiliate (claim 13849). No unnamed "strategic partner" or "anchor customer" appears in the record.

The most material shift is in TeraWulf's economic share of Nautilus rather than in counterparty count. The 2022 framing attributed 150 MW of 300 MW, roughly half, to TeraWulf (claim 14112). That fell to a 33.33% equity interest (claim 13714) and then to 25%, with one of five board seats and further adjustment possible based on capital contributions (claims 12938, 12874, 12840). Dependence on Talen deepened while TeraWulf's control diminished.

The disclosed figures do not reconcile cleanly, and no revenue figure is tied to any agreement. The company claimed to have secured supply of 58,000 miners for calendar 2022 (claims 14247, 14123), but only 20,000 were committed to Nautilus (claim 13713), and the non-fixed-price Bitmain terms leave the cost unanchored. The $19,072 joint-venture investment (claim 13918) and the $339,200 transition-services fee (claim 12942) are cash flows running out of TeraWulf and to a related party, not revenue. Note that the claim record is truncated at claim 12847.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. The company's revenue is self-mined bitcoin paid out through a Foundry pool (13440, 13346) and custodied or sold through NYDIG under dated agreements from March 10 and September 16, 2022 (13412, 13414). Every named counterparty is a supplier, service provider, power source, financing agent or joint-venture partner rather than a buyer of TeraWulf's output. The one reference to 'hosting arrangements' (12852) implies third-party customers, but none is named, sized or dated, and it should be treated as unidentified.

The named relationships are generally described in executed-contract language. The Bitmain and Minerva purchase agreements are described as executed (14028, 13977). The Talen joint venture carries a specific signing date of May 13, 2021 (14120). The Susquehanna power arrangement has a stated five-year term (14113), and the NYPA allocation is a specific 90 MW (12842). Softer language is limited to characterizations layered on top of these contracts. The 'strategic relationship' with Bitmain promising 'preferred access and advantaged cost structure' (12891) is not evidenced by a separate agreement, and the Nautilus 'first behind-meter nuclear' positioning (12923) is expressly forward-looking. The 'despite securing' phrasing around the 58,000-miner commitment (14123, 14247) signals delivery shortfalls against the signed supply terms.

Concentration has moved against TeraWulf within its single largest partnership. The initial Talen framing gave TeraWulf 150 of 300 MW, or half (14112). By August 2022 that had become a 33.33% equity stake against a 20,000-miner contribution (13714, 13713). By March 2023 it was 25%, subject to further adjustment, with one of five board seats (12938, 12874). Related-party exposure also runs through Beowulf E&D, which provides operations to Nautilus under a $339,200 transition agreement (12942, 13718), and through the affiliate NovaWulf (13849).

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-to-revenue or per-counterparty dollar figures are disclosed. The only cash figures are a $19,072 joint-venture investment outflow, likely in thousands (13918), and the Beowulf fee. Readers should also note that many claims are duplicates of the same NYDIG and Foundry agreements across filings, which overstates the apparent breadth of the relationship base. The supplied record also ends mid-claim in March 2023, so any later counterparty developments fall outside this read.

TeraWulf's record through March 2023 contains no revenue-paying customer in the conventional sense, so separating signed customer contracts from MOU or pilot language mostly doesn't apply. As a self-mining bitcoin producer, its disclosed counterparties are suppliers, a joint-venture partner, a power source, service providers and financing agents. Nearly all of them are named and described as executed. The equipment position rests on an executed purchase agreement with Minerva and four non-fixed price agreements with Bitmain (claims 14051, 14028, 13977), later joined by five future sales agreements with Bitmain (claim 12847). The Nautilus facility is a documented joint venture with Talen, amended and restated on August 27, 2022 (claims 14120, 12848), with Cumulus Coin identified as the Talen subsidiary (claim 12841). Its nuclear power comes from the Susquehanna Station at a contractual fixed rate over a five-year term (claims 14113, 12838). The revenue-adjacent plumbing is also dated and named: Foundry runs the mining pool (claim 13335), NYDIG provides custody from March 10, 2022 (claim 13412), and NYDIG Execution handles sales from September 16, 2022 (claim 13414). The record has no unnamed 'strategic partner' or 'anchor customer' language. The softer items are the affiliate relationships, NovaWulf (claim 13849) and Beowulf E&D (claims 13718, 12942), which deserve related-party scrutiny rather than customer treatment.

Concentration has tightened, not diversified. The money that turns into revenue runs through one pool operator and one custodian and execution group. At the same time, TeraWulf's economic share of Nautilus has shrunk. The 150 of 300 MW attributable in early 2022 (claim 14112) became a 33.33% interest in August 2022 (claim 13714) and then 25% by March 2023 (claims 12872, 12840). TeraWulf also holds one of five board seats (claim 12874), which leaves Talen in control.

The disclosed arithmetic does not reconcile with the announced deals, mainly because the record has little arithmetic in it. The 58,000-miner delivery commitment (claims 14247, 14123) is introduced with 'despite securing,' which suggests a shortfall that is never quantified. Only 20,000 miners are committed to Nautilus (claim 13713). The only disclosed figures are a $19,072 joint-venture investment for Q1 2022 (claim 13918) and a $339,200 transition-services fee (claim 12942). No mined-bitcoin revenue appears that could be tied to fleet size, hashrate or the stepped-down Nautilus stake. Several quotes are also truncated mid-sentence.

The first thing to say about this record is that it contains no customer agreements at all. Every named counterparty here sits on the supply, infrastructure, or monetization side of the business: Bitmain and Minerva as equipment vendors (claims 14246, 14051, 14052, 14027, 14028, 13977), Talen Energy and its Cumulus Coin subsidiary as joint venture partner (claims 14112, 14120, 12889), Foundry Digital as mining pool operator (claims 13440, 13413, 13346, 13162), NYDIG as custodian and execution venue (claims 13412, 13414, 13344, 12905, 12906), Beowulf E&D as a related-party services provider (claims 13718, 12942), and Cantor Fitzgerald, B. Riley and D.A. Davidson as ATM sales agents (claim 13973). A mining company selling a commodity into a pool has no named buyer to disclose, so the absence is structural rather than evasive, but any framing of these as customer wins would be a category error.

On identification, the record is unusually clean. There is no pattern of unnamed strategic partners or anchor customers; the counterparties are named, with executed instruments and dates. The hedging that does exist is in scope rather than identity. The Bitmain agreements are explicitly non-fixed price (claims 14051, 14027), so the 58,000-miner delivery commitment (claims 14247, 14123) carries no determinable dollar value. The Pennsylvania capacity is described as up to 300 MW with 150 MW attributable to TeraWulf only once conditions are met (claim 14112), and the JV equity split is subject to adjustment based on relative capital contributions (claims 12938, 12921).

Concentration moved in one direction, and it was dilutive. TeraWulf held 33.33 percent of Nautilus in August 2022 (claim 13714) against a 20,000-miner contribution (claim 13713); by March 2023 that was 25 percent (claims 12872, 12938) following the August 2022 amendment (claim 12920), with one of five board managers (claim 12874). Monetization simultaneously concentrated into exactly two vendors, Foundry and NYDIG (claim 13441).

The revenue arithmetic cannot be reconciled because there is almost none to reconcile. The only figures on record are a 19,072 thousand dollar JV investment outflow (claim 13918) and a 339,200 dollar transition services payment from Nautilus to Beowulf E&D (claim 12942), which is a related-party cost, not TeraWulf revenue. Revenue is stated to come from self-mined bitcoin (claim 13441), meaning it tracks hashrate and price, not any announced deal.

The first thing the record establishes is that TeraWulf, across the entire disclosed window, has announced no customer agreements at all. Every named counterparty is either a supplier, a joint-venture partner, a service vendor, a capital-markets agent, or a related party. Bitmain and Minerva are equipment vendors (14246, 14051, 14052, 14027, 14028, 13977); Talen Energy and its subsidiary Cumulus Coin are a JV counterparty (14112, 14120, 12889); Foundry Digital is a mining-pool operator and NYDIG a custodian and execution agent (13440, 13441, 13412, 13414, 13346, 12905, 12906); Cantor Fitzgerald, B. Riley and D.A. Davidson are ATM sales agents (13973); Beowulf E&D and NovaWulf are affiliates (13718, 13849, 12942). For a self-mining business that is structurally correct, but it means no announcement in this set constitutes demand validation, and readers should resist reading supply-side names as commercial traction.

On identification, the disclosure discipline is actually good: there are no unnamed strategic partners or anchor customers in the record. The softness is in the contract language rather than the identity. The Bitmain agreements are described as four non-fixed price sales and purchase agreements (14051, 14027), which locks neither price nor delivery, and the 58,000-miner figure is a delivery commitment, not a purchase executed (14247, 14123). The word despite in 14123 is itself an admission that the commitment underdelivered. The Talen language is capacity-ceiling framing, up to 300 MW with 150 MW attributable (14112).

Concentration has moved in one direction, and it is dilution rather than diversification. TeraWulf's Nautilus interest falls from 33.33% (13714) to 25% following the August 2022 amendment (12920, 12938, 12872, 12921), with governance at one of five managers (12874). Revenue-side concentration is effectively single-channel: Foundry for the pool, NYDIG for custody and sale.

The arithmetic does not reconcile because there is nothing to reconcile it against. No deal in this set carries a disclosed revenue figure. The only named dollar amounts run outward — $19,072 thousand of JV investment (13918) and a $339,200 transition-services payment to affiliate Beowulf E&D (12942). Separately, 20,000 miners were committed to Nautilus (13713) against 58,000 contracted, leaving the majority unaccounted for in the disclosed structures. Note that this record terminates in March 2023 and says nothing about later HPC hosting arrangements.

The first thing worth saying about TeraWulf's disclosure record is that it contains almost no customer agreements at all. What is catalogued here is a procurement and services stack, not a demand book: equipment supply from Bitmain and Minerva (14246, 14051, 14052, 14027, 14028, 13977), a joint venture with Talen Energy (14120, 14112, 14128), power offtake from the Susquehanna substation (14113, 14129) and the New York Power Authority (12842, 12890), pool services from Foundry Digital (13346, 13335, 13413), custody and execution from NYDIG (13344, 13345, 13412, 13414), and an ATM sales agreement with Cantor Fitzgerald, B. Riley and D.A. Davidson (13973). Revenue in this period is self-mined bitcoin sold through an execution agent, so the counterparties are vendors and infrastructure providers rather than buyers.

On identification, the record is unusually clean. Every material counterparty is named; there is no instance in this set of an unnamed 'strategic partner' or an undisclosed 'anchor customer' standing in for a contract. The related-party threads are also labeled rather than obscured — Beowulf E&D as services provider to Nautilus (13718, 12942) and NovaWulf as an affiliate (13849) — which is the correct treatment even if it narrows the independence of the counterparty set.

Concentration has moved in one direction, and not a favorable one. TeraWulf's Nautilus economics fell from a 33.33% member interest with a 20,000-miner contribution (13714, 13713) to 25% under the August 2022 amendment and the March 2023 Second A&R agreement (12920, 12938, 12872, 12921), with governance at one of five managers (12874). Dependence on a single partner deepened as the claim on the asset shrank.

The arithmetic does not close on the equipment side. Agreements for 58,000 miners across calendar 2022 (14247, 14123) are never reconciled to deployed hashrate in this record, and the 'despite securing' framing in 14123 concedes delivery slippage without quantifying it. The only hard dollar figures disclosed are outflows — $19,072 thousand of JV investment (13918) and a $339,200 transition services payment (12942). Nothing here supports a revenue bridge; the record simply stops at 2023-03-30, before any later hosting or compute contracting.

The first thing the record establishes is a definitional problem: across roughly sixty months of named-counterparty disclosure, TeraWulf has announced essentially no customer agreements. Every named counterparty sits on the supply or intermediary side of the business. Bitmain and MinerVA are equipment vendors (claims 13977, 12847, 14051, 14052, 14027, 14028); Talen Energy is a joint venture partner and, through the Susquehanna substation, a power supplier (claims 14112, 14120, 14128, 14113, 14129); the New York Power Authority supplies Lake Mariner (claims 12842, 12890); Foundry Digital operates the mining pool the company hashes into (claims 13440, 13413, 13335, 13346, 13162); NYDIG custodies and executes (claims 13412, 13414, 13333, 13334, 13344, 13345); Cantor Fitzgerald, B. Riley and D.A. Davidson are ATM agents, not customers (claim 13973); and Beowulf E&D and NovaWulf are related parties (claims 13718, 12942, 13849). That is a clean set of named, executed agreements with specific instruments cited — Digital Asset Custodial Agreement, Execution Agreement, Service Agreement — so the signed-versus-soft distinction is easy here, but it resolves in an unusual direction. There is no anchor customer to identify because, on this record, the company sells bitcoin, not services.

Concentration has therefore evolved on the input side rather than the demand side. The 2022 posture was vendor-concentrated: 58,000 miners contracted for delivery within a single calendar year from two suppliers (claims 14246, 14247, 14123), with the March 2023 disclosure adding five further Bitmain agreements (claim 12847). The 2023 posture is monetization-concentrated: a single pool operator and a single custodian/execution venue intermediate substantially all revenue (claim 13441). Equity concentration moved against TeraWulf within Nautilus, from a stated 33.33% interest in August 2022 (claim 13714) to 25% by March 2023 (claims 12872, 12938), with board control one of five managers (claim 12874).

Revenue arithmetic cannot be reconciled to announced deals because none of these agreements carries contracted revenue. Nautilus pays in bitcoin distributions scaled to hash rate net of power and operating costs (claim 12939), which is a price-and-uptime variable, not a bookable commitment. The 58,000-miner figure is a purchase obligation, and claim 14123's own "despite securing" framing signals that delivery lagged the contract.

The first thing to say about TeraWulf's disclosure record is that almost nothing in it is a customer agreement. The named counterparties are suppliers, a joint-venture partner, a power authority, a mining pool and a custodian — inputs, not demand. Bitmain and MinerVA appear as executed equipment purchase agreements (14246, 14051, 14028, 13977, 12847) covering 58,000 miners for calendar 2022 (14247, 14123); Talen Energy appears as a joint venture partner for up to 300 MW at Nautilus (14112, 14120, 14128); the New York Power Authority appears as a 90 MW supply agreement at Lake Mariner (12842, 12890); and Susquehanna nuclear output is secured on a five-year term with two three-year renewals (14113, 14129). These are firm, named and executed, but they are procurement.

On the revenue side the concentration is stark and it is intermediary concentration rather than customer concentration. Essentially all mining revenue is earned through a single named pool operator, Foundry Digital, under a Service Agreement (13440, 13413, 13335, 13346, 13162), with proceeds custodied at NYDIG under the March 10, 2022 Digital Asset Custodial Agreement (13412, 13333, 13344) and liquidated through NYDIG Execution under the September 16, 2022 Execution Agreement (13414, 13334, 13345). One pool, one custodian, one execution venue. There is no named buyer of anything because self-mined bitcoin has no counterparty; the language of 'anchor customer' or 'strategic partner' never has to be tested here because no such claims are made in the record.

Counterparty economics have moved against TeraWulf where it matters. Its Nautilus interest is disclosed at 33.33% in August 2022 (13714) and at 25% by March 2023 (12872, 12938, 12921), with board representation of one manager out of five (12874) — a minority position on both cash flow and control, with distributions in kind every two weeks net of allocated power and operating cost (12939).

Revenue arithmetic cannot be reconciled against announced deals because no announced deal carries a contract value. The only dollar figures attached to named counterparties are outflows: $19.072 million of JV investment in Q1 2022 (13918) and a $339,200 transition services payment to related party Beowulf E&D (12942, 13718). Related-party density is high (13849, 13391). The record as supplied stops in March 2023; any later contracted-revenue claims are outside it.

The striking feature of TeraWulf's disclosure record through March 2023 is that it contains almost no customer agreements at all. Nearly every named counterparty sits on the supply side or the service side of the ledger: Bitmain and MinerVA as equipment vendors (claims 13977, 12847, 14028, 14051), Talen Energy as joint venture partner (14112, 14120, 14128), the New York Power Authority and Susquehanna Station as power suppliers (12890, 12842, 14113), Foundry Digital as mining pool operator (13440, 13335), NYDIG and NYDIG Execution as custodian and sales agent (13412, 13334), Beowulf E&D as an affiliated services provider (13718, 12942), and Cantor Fitzgerald, B. Riley and D.A. Davidson as ATM agents (13973). TeraWulf is the buyer or the counterparty's client in essentially all of these.

On naming discipline, the record is unusually clean. There is no MOU, LOI, term sheet, or "in discussions" language anywhere in the set; the operative verbs are "executed," "entered into," and "secured." Nor are there unidentified "strategic partners" or "anchor customers." The two soft edges are the "five future sales and purchase agreements with Bitmain" in claim 12847, which are forward-dated and option-like rather than delivered volume, and the reference to "bitcoin earned from hosting arrangements" in 12852, where no hosting counterparty is ever named.

Concentration has moved against TeraWulf rather than toward diversification. Its Nautilus equity fell from 33.33% (13714) to 25% (12872, 12938), with board representation of one manager out of five (12874), while revenue capture remains effectively single-channel: one pool operator, one custodian, one execution venue.

The arithmetic does not reconcile because there is nothing to reconcile it against. The 58,000-miner commitment (14247, 14123) carries no aggregate price, since the Bitmain contracts are explicitly non-fixed price (14051). Nautilus distributions are paid in bitcoin pro rata to hash rate net of costs (12939), so no contracted dollar revenue exists. The only hard figures on record are $339,200 of transition services (12942), $19.072 million of JV investment (13918), and 20,000 contributed miners (13713) — all outflows or in-kind contributions, none of them bookable backlog.

The first thing to say about TeraWulf's disclosure record is that it contains almost no customer agreements at all. Every named counterparty on record is either a supplier, a power provider, a joint-venture partner, a related party, or a downstream service vendor. Bitmain and MinerVA are equipment vendors under executed purchase agreements (claims 13977, 12847); the New York Power Authority supplies 90 MW at Lake Mariner (claims 12842, 12890); Talen Energy is a JV partner, not a buyer (claims 14112, 14120, 14128). The only counterparty that actually remits value to TeraWulf is Foundry Digital, the mining pool operator, under a Service Agreement (claims 13335, 13346), with NYDIG holding the resulting bitcoin in cold storage (claims 13333, 13344) and executing sales (claims 13334, 13345). In a self-mining model that is the revenue stack, and it is single-threaded.

On the signed-versus-soft question, the record is unusually clean in one respect and unhelpful in another. There is no MOU, LOI, or 'in discussions' language here, and no unidentified 'strategic partner' or 'anchor customer' standing in for a name — every agreement cited is described as executed or entered into on a specific date. But that cleanliness is a function of what these agreements are. Procurement and power contracts are easy to sign and name; they obligate the company to pay, not the counterparty. The absence of soft language should not be read as evidence of demand, because no demand-side contract is being disclosed in the first place.

Concentration has narrowed rather than diversified. Revenue routes through one pool operator and one custodian, and TeraWulf's economic interest in Nautilus fell from a stated 33.33% (claim 13714) to 25% (claims 12872, 12938, 12921) while board representation sat at one of five managers (claim 12874) — a diluted minority in the asset most often presented as a partnership win.

The arithmetic does not reconcile, mainly because there is nothing to reconcile it against. The 58,000-miner delivery commitment for calendar 2022 (claims 14247, 14123) is a capacity input, never bridged to output in this record, and by March 2023 the commitment had grown to include five further Bitmain agreements (claim 12847). The only dollar figures disclosed are a $19,072 thousand JV outflow (claim 13918) and a $339,200 related-party transition fee paid to Beowulf E&D (claim 12942) — both costs, neither revenue.

The first thing worth saying about TeraWulf's record through March 2023 is that it contains almost no customer agreements. What it contains is a dense set of executed supplier, power, custody and joint-venture contracts, and the editorial instinct to read named counterparties as demand-side relationships would be wrong here. Bitmain and Minerva Semiconductor are equipment vendors under four non-fixed-price purchase agreements and one purchase agreement respectively (claims 14051, 14052, 14027, 14028, 13977); Talen Energy and its Cumulus affiliate are a joint-venture co-developer (14112, 14120, 14128, 14141); the New York Power Authority and the Susquehanna Station substation are power suppliers (12842, 12890, 14113, 14129); NYDIG is a custodian and NYDIG Execution a sales agent (13412, 13414, 13333, 13345); Cantor Fitzgerald, B. Riley and D.A. Davidson are ATM underwriters, not commercial counterparties (13973); Beowulf E&D and NovaWulf are related parties (13718, 13849, 12942).

To the credit of the filings, there is no MOU, LOI or discussions language in this set, and no unnamed strategic partner or anchor customer standing in for a real name. Every relationship is identified and every agreement is described as executed. The soft edge is capacity framing rather than pipeline framing: Talen is a partnership to develop up to 300 MW, only 150 MW attributable to TeraWulf (14112), and the 58,000 miners contracted for January through December 2022 delivery (14247, 14123) appear in a sentence that begins despite securing, which is itself the concession that delivery lagged the contract.

Concentration moved in one direction and it was adverse. Revenue arrives through a single channel, the Foundry Digital mining pool (13440, 13413, 13335, 12854), so the effective customer count is one. Meanwhile the Nautilus economics diluted from a 33.33 percent equity interest in August 2022 (13714) to 25 percent by March 2023 (12872, 12938), against one of five board managers (12874), after contributing 20,000 miners (13713).

The arithmetic does not reconcile because there is nothing to reconcile it against. No deal in this record carries a disclosed revenue figure. The only dollar amounts are outflows: a 19,072 thousand dollar JV investment (13918) and a 339,200 dollar transition payment from Nautilus to affiliate Beowulf E&D (12942). Nautilus economics arrive as in-kind biweekly bitcoin distributions pro rata to hash rate, net of power and operating costs (12939), which is a variable claim on production, not contracted revenue.

The first thing worth establishing is that almost nothing in TeraWulf's counterparty record describes a customer. The named parties are suppliers, a joint-venture partner, service vendors and affiliates, and the agreements are genuinely executed rather than aspirational. Bitmain and MinerVA appear as signed equipment purchase agreements, with the 2023 disclosure specifying an executed purchase agreement with MinerVA, four non-fixed price sales and purchase agreements with Bitmain and five future sales and purchase agreements (12899, 14051, 14052, 13977). Talen, through Cumulus Coin, is a signed joint venture, first in May 2021 and amended and restated in August 2022 (14120, 12901). Foundry supplies the mining pool under a Service Agreement and NYDIG supplies custody and trade execution under agreements dated March 10 and September 16, 2022 (13346, 13344, 13345). Cantor Fitzgerald, B. Riley and D.A. Davidson are underwriting agents under a Sales Agreement, not commercial counterparties (13973). Beowulf E&D and NovaWulf are related parties (13718, 12942, 12867, 13849).

On identification, the record is unusually clean: every counterparty is named. The one instance of softer characterization is Bitmain, described as a strong, collaborative relationship and a strategic relationship, language that sits alongside rather than in place of the underlying signed contracts (12891). There are no unnamed anchor customers or undisclosed strategic partners to discount.

Concentration has tightened rather than diversified. The 2022 set spanned two equipment vendors plus a partner; by early 2023 the revenue-bearing path narrows to a single pool operator, Foundry, with NYDIG as the sole custodian and execution venue (13440, 13441, 13442, 13162, 13161). Economic exposure to Nautilus moved the wrong way, from a 33.33% interest in August 2022 to 25% by March 2023, with one of five board managers (13714, 12938, 12872, 12874).

The arithmetic does not reconcile in the conventional sense because there is nothing to reconcile. Revenue is self-mined bitcoin, not contracted counterparty revenue; Nautilus returns arrive as biweekly in-kind distributions net of power and operational costs (12939). The 58,000 miners contracted for 2022 delivery are framed with a concessive despite, 20,000 were earmarked for contribution to Nautilus, and the associated cash flows disclosed are outbound: $19.1 million of JV investment in Q1 2022 and an $8.46 million annual Beowulf fee (14247, 13713, 13918, 12867).

The first thing the record establishes is that TeraWulf, across this window, announced almost no customers at all. Every named counterparty sits on the input side or the plumbing side of the business: equipment suppliers (Bitmain and Minerva Semiconductor, claims 14246, 14051, 14052, 14027, 14028, 13977), a joint venture partner (Talen Energy, claims 14120, 14112, 14128, 12919), a pool operator (Foundry Digital, claims 13440, 13413, 13346, 13335, 13162), a custodian and execution agent (NYDIG and NYDIG Execution, claims 13412, 13414, 13333, 13344, 13345, 12904, 12906), a control-account bank (Wilmington Trust, claim 12905), three ATM underwriting agents (claim 13973), and two related parties, Beowulf E&D and NovaWulf (claims 13718, 12942, 13849). What is conspicuously absent is soft language. There are no MOUs, LOIs, frameworks, or 'discussions' here; the verbatim language is consistently 'entered into,' 'executed,' 'secured.' The disclosure risk in this record is not vagueness but identification of the wrong kind of counterparty.

Two qualifications matter. The Bitmain arrangements are described as 'four non-fixed price sales and purchase agreements' (claims 14051, 14027) — contracted volume without contracted price, which is weaker than the aggregate 58,000-miner delivery commitment headline suggests (claims 14247, 14123). And claim 14123's own framing, 'Despite securing Equipment Supply Agreements,' concedes that the commitment did not convert on schedule. Susquehanna is named as a generating station, not as a contracting party; the power is received through the JV via a substation interconnection (claims 14113, 14129).

Concentration moved in the wrong direction on the one deal that carried economics. TeraWulf's Nautilus interest is disclosed at 33.33% in August 2022 (claim 13714), then at 25% from March 2023 (claims 12872, 12938, 12921) following the August 2022 amendment (claim 12920), with one of five board managers (claim 12874).

The arithmetic does not reconcile because it cannot. Revenue is pool payout from a single counterparty, Foundry (claims 13440, 13441); Nautilus contributes bitcoin in kind every two weeks net of power and operating cost (claim 12939); and the JV shows up as investment, $19,072 thousand in Q1 2022 (claim 13918), not as revenue. The announced deals are procurement, not demand, and the record here ends in March 2023.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. Revenue comes from self-mined bitcoin (13441), so nearly every named counterparty is a supplier, service provider, power source, or joint-venture partner rather than a buyer. The one reference to third-party demand is a passing mention of "bitcoin earned from hosting arrangements" (12852). It names no hosting client, gives no contract term or pricing, and is the only hint of a revenue counterparty in the file.

The named relationships are specific and mostly documented as executed. They include equipment purchase agreements with Bitmain (four non-fixed-price contracts) and Minerva (14051, 14052, 13977), the Talen joint venture dated May 13, 2021 (14120), a NYDIG custodial agreement dated March 10, 2022 (13412), a NYDIG Execution agreement dated September 16, 2022 (13414), the Foundry pool service agreement (13346), and a 90 MW NYPA power agreement at Lake Mariner (12842). The softer language sits around Bitmain. By 2023 the relationship is described as "strong, collaborative" and "strategic" with "preferred access" (12891), but no new contract terms are disclosed. Several counterparties are related parties rather than arm's-length partners: NovaWulf is an affiliate (13849), and Beowulf E&D provides services to Nautilus, including a $339,200 transition arrangement (12942).

Concentration has shifted toward Talen, and TeraWulf's position there has weakened. The 2022 framing was an up-to-300 MW venture with 150 MW attributable to TeraWulf (14112) and a 33.33% equity stake (13714). The March 2023 amended agreement cut this to 25% (12938, 12872), gave TeraWulf one of five board seats (12874), and made distributions depend on relative hash-rate contribution (12939). Power for Nautilus depends on Susquehanna Station under a five-year term with renewal options (14113, 14129).

The revenue arithmetic cannot be reconciled from this record, because no revenue line is tied to any named counterparty. The operational figures that are disclosed show gaps. Supply agreements covered 58,000 miners for delivery during 2022 (14247), yet the filing introduces this with "Despite securing" (14123), which implies a shortfall. The planned contribution to Nautilus was 20,000 miners (13713). The $19,072 thousand of Q1 2022 joint-venture investment (13918) is capital outflow, not revenue. Readers should treat the Cantor/B. Riley/D.A. Davidson sales agreement (13973) and the January 2023 voting agreement (13391) as financing and governance documents, not commercial wins.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense. The company is a self-miner, and it says so directly: "the majority of our revenue comes from our self-mined bitcoin" (13441). The named counterparties are suppliers, service providers, power sellers and a joint-venture partner, not buyers of a TeraWulf product. On the positive side, the file contains no MOU, LOI or "in discussions" language at all. Nearly every counterparty is identified by legal name and tied to a dated, executed instrument. These include the NYDIG Custodial Agreement of March 10, 2022 (13412, 13344) and the NYDIG Execution Agreement of September 16, 2022 (13414, 13345), as well as the Foundry pool Service Agreement (13346), the Bitmain and Minerva purchase agreements (13977, 14028), the Talen JV (14120) and the NYPA power agreement (12842). The softer language is narrower. The Bitmain "strategic relationship" offering "preferred access" (12891) is characterization rather than a disclosed term. The "hosting arrangements" referenced in 12852 are the only customer-like revenue source in the record, and no hosting client is named. Two named counterparties are related parties, not arm's-length ones. NovaWulf is described as "an affiliate of the Company" (13849), and Beowulf E&D receives $339,200 in transition fees from Nautilus (12942). The most material evolution is in the Talen/Nautilus economics rather than counterparty breadth. TeraWulf's position shrank from 150 of 300 MW, roughly half (14112), to a 33.33% equity interest (13714), and then to 25% under the March 2023 amendment (12938, 12872, 12921). Over the same period, governance moved to one of five board seats (12874). Exposure to the single largest partner therefore rose even as TeraWulf's claim on the asset fell. The equipment arithmetic does not reconcile cleanly. The company cited 58,000 miners under supply agreements for delivery during 2022 (14247), but framed that figure with "despite securing" (14123), signalling a shortfall. Only 20,000 miners are committed to Nautilus (13713). The Bitmain contracts are also explicitly "non-fixed price" (14051), so the dollar commitment is indeterminate. Apart from a $19,072 (thousand) Q1 2022 JV investment (13918), the record discloses no revenue figures, so announced relationships cannot be tied to reported revenue. The claim list is also truncated at 12921, and later entries may alter this read.

TeraWulf's record contains no customer agreements in the conventional sense, and that absence is the central diligence point. As a self-mining bitcoin producer, its revenue comes from block rewards paid through a pool, and the named counterparties are suppliers, service providers, a joint-venture partner and financing agents rather than buyers of a product. The contracts that are described as executed are specific and named: four non-fixed price purchase agreements with Bitmain and one with Minerva (14051, 14028, 13977), the Talen joint venture dated May 13, 2021 (14120), a five-year Susquehanna nuclear power supply (14113, 14129), an initial 90 MW with NYPA (12842), NYDIG custody dated March 10, 2022 (13412), NYDIG Execution dated September 16, 2022 (13414), and Foundry pool services (13346). Softer language appears where the relationship is characterized rather than papered. The 'strategic relationship' with Bitmain promising 'preferred access and advantaged cost structure' (12891) is an assertion, not a contract term, and the 'non-fixed price' qualifier (14051) means the supply agreements did not lock in cost. The one reference that points toward genuine third-party customers, 'bitcoin earned from hosting arrangements' (12852), names no hosting client at all. Several named parties are related, including Beowulf E&D (13718, 12942) and NovaWulf (13849), and should not be read as arm's-length validation. The Sales Agreement with Cantor, B. Riley and D.A. Davidson (13973) is an equity-issuance facility, not a commercial relationship. Concentration has shifted rather than diversified. The Talen venture was first framed as 150 MW of 300 MW attributable to TeraWulf (14112), then set at 33.33% equity against a 20,000-miner contribution (13714, 13713), then cut to 25% with Talen's Cumulus appointing four of five managers (12938, 12874, 12921). Economic exposure to the flagship nuclear site therefore fell while dependence on a single partner's governance rose. Supply is similarly concentrated in Bitmain, and the 58,000-miner delivery commitment is introduced with 'despite securing' (14123), which signals shortfall rather than fulfillment. The claims provide no revenue figures to reconcile against these deals. The only amounts disclosed are $19,072 in joint-venture investment (13918) and a $339,200 related-party transition fee (12942), both of which are cost or capital lines. Revenue arithmetic cannot be tested from this record, and the supplied claim list is truncated after 12921.

TeraWulf's disclosed record through March 2023 contains almost no customer agreements in the conventional sense, and that is the central diligence point. The company's revenue is self-mined bitcoin, and the named counterparties are suppliers, a joint-venture partner, a power provider, and service vendors rather than buyers of a product (13441, 13440). Against that, the paper is unusually firm. There is no MOU, LOI, or 'in discussions' language anywhere in the set. The Bitmain and Minerva purchase agreements are described as executed (14028, 13977). The Talen joint venture carries a signing date (14120) and a dated amendment (12848). The NYDIG custody and execution agreements are dated (13412, 13414), and the NYPA power arrangement is described as secured (12890). The softer language sits in capacity figures rather than in the counterparties. The 58,000-miner delivery commitment (14247, 14123) runs through non-fixed-price agreements (14051), and the Nautilus power term includes optional renewals (14113).

Every material counterparty is named, with one exception. Bitcoin 'earned from hosting arrangements' is referenced (12852), but no hosting customer is identified anywhere in the record, so that revenue stream cannot be attributed. Several named parties are affiliates or governance counterparties rather than arm's-length commercial partners. NovaWulf is disclosed as an affiliate (13849), and Beowulf E&D is paid to provide services to Nautilus (13718, 12942). Readers should treat these separately from third-party validation.

Concentration has evolved against TeraWulf within its most prominent partnership. The initial Talen framing gave TeraWulf 150 MW of up to 300 MW, roughly half (14112). The August 2022 restatement cut its equity to 33.33% (13714). The March 2023 amendment cut it again to 25%, with one of five board seats (12938, 12874). Over the same period, operational dependence consolidated on a small set of vendors: Foundry for pool revenue (13335) and NYDIG for custody and sales (13344, 13345).

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate-output, or hosting-fee figures are disclosed. The only dollar amounts are a $19.1 million Q1 2022 joint-venture investment outflow (13918) and a $339,200 transition-services fee payable by Nautilus (12942). The 20,000 miners contributed to Nautilus (13713) cannot be tied to the 58,000 contracted (14247) without delivery data. Note that the supplied record ends mid-claim at 12921, so later disclosures are not reflected here.

TeraWulf's record through March 2023 contains no revenue-paying customer in the conventional sense, so the useful distinction is between executed operating contracts and the softer language around them. The company describes its revenue as self-mined bitcoin earned through Foundry Digital's pool, custodied at NYDIG and sold via NYDIG Execution (13441, 13335). That is a pool-participation model, not an offtake. Each of those counterparties is named, and the custodial and execution agreements carry specific execution dates of March 10, 2022 and September 16, 2022 (13344, 13345). The Bitmain and Minerva supply arrangements are also named and described as executed (13977), but the Bitmain contracts are explicitly "non-fixed price" (14051), so they fix supply more than cost. Unusually for this category, nothing on the record hides behind unidentified 'strategic partners' or 'anchor customers.' The softness lies in commitments rather than identities. The 58,000-miner delivery schedule for calendar 2022 (14247) reappears two weeks later prefaced by "Despite securing" (14123), which signals a shortfall against the contractual figure. Concentration is where the record moves most. The Talen joint venture began as 150 MW attributable out of 300 MW (14112). An August 2022 filing then shows a 33.33% interest backed by 20,000 contributed miners (13714, 13713). March 2023 disclosures give 25%, with one of five board seats (12938, 12874). Claim 12860 dates the cut to 25% to the same August 2022 amendment, which conflicts with 13714 and should be reconciled against the amendment itself. Economic dependence has therefore shifted from a co-equal partnership toward a minority position in a Talen-controlled vehicle, and toward NYPA-sourced power at Lake Mariner (12842). Alongside this sits a related-party layer: Beowulf E&D's $8.46 million annual base fee (12867), a $339,200 transition services agreement (12942), and NovaWulf's affiliate status (13849). Revenue arithmetic cannot be reconciled from these claims, because no revenue, hashrate or bitcoin-production figure is on record. The only quantified cash item is a $19,072 thousand joint-venture investment in Q1 2022 (13918). The 58,000-miner and 20,000-contribution figures are capacity commitments, not revenue. Readers should treat the deal list as a map of named suppliers and a shrinking JV stake, not as a contracted revenue base.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, and the analysis should be read with that in mind. As a self-mining bitcoin producer, the company earns revenue from block rewards routed through a mining pool, so every named counterparty is a supplier, service provider, joint-venture partner or financing agent rather than a buyer of contracted output. None of the language is MOU, LOI or 'discussions' wording. The agreements are described as executed, which is unusual and to the company's credit, though some carry softer terms. The Bitmain agreements are specifically 'non-fixed price' (claims 14051, 14027, 13977), and the 58,000-miner delivery schedule is introduced with 'despite securing' (claim 14123), which signals that delivery fell short of the contracted volume. Every counterparty is identified by name, with no anonymous 'strategic partner' or 'anchor customer' framing. Equipment comes from Bitmain and Minerva (claims 14246, 14052, 14028). The Nautilus joint venture is with Talen Energy through Cumulus Coin (claims 14120, 12889). Power comes from Susquehanna Station and NYPA (claims 14113, 12842). Pool services come from Foundry Digital (claims 13335, 12854), and custody and execution from NYDIG (claims 13344, 13345). The Cantor, B. Riley and D.A. Davidson sales agreement is an equity distribution arrangement, not commercial revenue (claim 13973). Beowulf E&D and NovaWulf are affiliates, so the $8.46 million annual services fee and the $339,200 transition payment are related-party flows that deserve separate scrutiny (claims 12867, 12942, 13849). Counterparty concentration has increased over time rather than diversified. The Talen venture was first framed as 150 of 300 MW attributable to TeraWulf (claim 14112), was cut to a 33.33% interest with 20,000 contributed miners in August 2022 (claims 13714, 13713), and then fell to 25% with one of five board seats (claims 12938, 12874). TeraWulf therefore became a minority partner in a facility that Talen controls. Revenue itself runs almost entirely through a single pool operator, Foundry, and a single custodian and execution venue, NYDIG. The disclosed revenue arithmetic cannot be reconciled from this record, because the claims contain no revenue or hashrate-output figures to test against the announced deals. The only quantified flows are the $19,072 thousand joint-venture investment in Q1 2022 (claim 13918) and the related-party fees. The 58,000-miner commitment cannot be tied to realized deployment either, since only 20,000 units are traced to Nautilus. The claim set also appears truncated at claim 12854, so later disclosures may change this picture.

TeraWulf's record through March 2023 contains no revenue-generating customer agreements in the conventional sense. The company states that the majority of its revenue comes from self-mined bitcoin (claim 13441), so most of the named counterparties are suppliers, service providers or joint-venture partners rather than buyers. Read that way, the disclosure is unusually well identified. Every material relationship is named. Bitmain and Minerva supply equipment (claims 14246, 14051, 14028). Talen, through an affiliate, is the Nautilus joint-venture partner (claims 14128, 14120, 12901), and Susquehanna Station supplies power (claims 14129, 14113). Foundry runs the mining pool (claims 13346, 12903), NYDIG and NYDIG Execution provide custody and sales (claims 13412, 13414), and Wilmington Trust is party to the account-control agreement (claim 12905). The one unidentified reference, an unnamed 'joint venture' in claim 14141, is resolved elsewhere in the same filing. The apparent volume of relationships is inflated by repetition: the NYDIG, Foundry and Talen agreements each appear in four to six near-identical claims.

The contract language is signed rather than aspirational. It uses 'entered into' and 'executed' throughout, with no MOU, LOI or pilot framing. The softer signal lies in the terms. The Bitmain contracts are 'non-fixed price' (claim 14027). The 58,000-miner delivery schedule for 2022 (claim 14247) is introduced with 'Despite securing' (claim 14123), which implies the deliveries fell short of the contract.

Concentration around the one operating asset has shifted against TeraWulf. The original framing was 150 MW of a 300 MW facility attributable to the company, roughly half (claim 14112). The August 2022 restatement set its stake at 33.33% (claim 13714), and the March 2023 terms cut it to 25% (claims 12872, 12938), subject to adjustment by capital contributions. TeraWulf also holds one of five board seats (claim 12874). Operationally, TeraWulf depends on single named providers for pool access, custody and execution. It also relies on related parties: Beowulf E&D provides services under a $339,200 transition agreement (claims 13718, 12942), and NovaWulf is disclosed as an affiliate (claim 13849).

The revenue arithmetic cannot be reconciled from this record. No agreement carries a disclosed contract value or revenue commitment. The only figures on record are a $19,072 outflow for joint-venture investment in Q1 2022, units unstated (claim 13918), and the transition-services fee. Revenue depends on mined output under Nautilus's biweekly hash-rate distribution formula net of power and operating costs (claim 12939), not on counterparty contracts. The claim set also appears truncated at claim 12899, so later disclosures fall outside this read.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense, and nothing in it uses MOU, LOI, framework or pilot language. Every named relationship is described as executed, but none of these counterparties pays TeraWulf. Talen is a joint-venture partner (claims 14120, 14112, 12841). Bitmain and Minerva are equipment vendors (claims 14051, 14052, 13977). NYDIG and NYDIG Execution are custodian and broker (claims 13412, 13414). Foundry operates the mining pool (claim 13335). Susquehanna Station and NYPA supply power (claims 14113, 12842). Beowulf E&D is a related-party service provider (claims 13718, 12942). The underwriters Cantor, B. Riley and D.A. Davidson (claim 13973) and the Voting and Support Agreement shareholders (claim 13391) are capital-markets and governance counterparties, not commercial ones. The only reference to paying third parties is "bitcoin earned from hosting arrangements" (claim 12852), and no hosting customer is named or sized anywhere in the record. Revenue therefore comes from the Foundry pool, not from contracted buyers.

Concentration runs through infrastructure dependence rather than customer exposure, and the central relationship has narrowed over time. Talen was initially framed as a partner on up to 300 MW, with 150 MW attributable to TeraWulf (claim 14112). By August 2022 that had become a 33.33% equity interest (claim 13714) and a 20,000-miner contribution (claim 13713). The March 2023 second amended agreement brought it to 25%, with one of five board seats (claims 12938, 12874, 12921). The dating is inconsistent, however. Claim 12920 attributes the reduction to 25% to the August 2022 amendment, while the contemporaneous filing puts that amendment at 33.33% (claim 13714). Diligence should resolve which document governs. Separately, the 58,000 miners contracted from Bitmain and Minerva (claim 14123) came with language conceding that the delivery schedule had not produced the expected deployment.

The revenue arithmetic cannot be reconciled from these disclosures because no deal carries a revenue figure. The only quantified amounts are cash outflows: $19,072 thousand invested in the joint venture in Q1 2022 (claim 13918) and a $339,200 transition-services payment from Nautilus to the related party Beowulf E&D (claim 12942). Nautilus distributes bitcoin based on hash rate after deducting power and operating costs (claim 12939), so TeraWulf's economics rest on network difficulty and bitcoin price rather than counterparty commitments. Any later presentation of contracted or hosted revenue should be checked against this baseline, since the historical record names no revenue-paying counterparty at all.

TeraWulf's record through March 2023 contains no customer agreements in the conventional sense. The company is a self-miner, and its revenue comes from block rewards paid through Foundry Digital's mining pool (claims 13440, 13346, 12854). It does not come from contracted offtakers. Every named counterparty is a supplier, service provider, financing agent, related party or joint-venture partner, and every one is described in executed-agreement language. There are no MOUs, LOIs, 'discussions' or pilots anywhere in the record. There are also no unnamed 'strategic partners' or 'anchor customers.' Bitmain and MinerVA are named as equipment sellers (13977, 12899), and 12899 adds five Bitmain future sales agreements to the original four. NYDIG is named as custodian (12904) and NYDIG Execution as broker (13414). Wilmington Trust is party to a deposit control agreement (12905). Cantor, B. Riley and D.A. Davidson are sales agents (13973). Talen's subsidiary Cumulus Coin is the Nautilus partner (12889).

The apparent breadth of the record is overstated by repetition. The NYDIG custody agreement alone appears in at least six claims (13412, 13333, 13344, 13161, 12904, 13441), and the Bitmain/MinerVA supply terms appear in five. Diligence should count agreements, not claims.

Concentration runs toward related parties and a single partner. Beowulf E&D, which is tied to management, earns an $8.46 million annual base fee (12867) and a $339,200 Nautilus transition fee (12942). NovaWulf is disclosed as an affiliate (13849). TeraWulf's position in Nautilus has diluted over time. It was 33.33% under the August 2022 agreement (13714) and 25% under the March 2023 second amendment (12938, 12921), with board control of four seats to one in Talen's favor (12874). Claim 12920 attributes the 25% figure to the August 2022 amendment, which conflicts with 13714 and should be reconciled against the underlying exhibits.

The revenue arithmetic cannot be reconciled from this record, because no announced deal carries a revenue value. The only quantified items are cost or investment lines: $19,072 thousand of joint-venture investment in Q1 2022 (13918), roughly 2.0 cents/kWh Nautilus power (12859), and the Beowulf fees. Any revenue model has to be built from hashrate, pool payouts and the biweekly Nautilus distributions net of power and operating costs (12939), not from counterparty contracts. The claim list as supplied also ends mid-entry at claim 12848 and contains nothing after March 2023. Any later hosting or customer announcements are therefore outside this read and would need separate classification.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense, and nothing in it is framed as an MOU, LOI or pilot. The company sells self-mined bitcoin. Every named counterparty is a supplier, service provider, joint-venture partner or related party rather than a buyer of capacity. The executed documents are specific and dated. They include the NYDIG Digital Asset Custodial Agreement of March 10, 2022 (13412, 13333), the NYDIG Execution agreement of September 16, 2022 (13414, 13334), the amended Talen joint venture of August 27, 2022 (12901, 12848), and equipment agreements with MinerVA and Bitmain (12899, 12847). The Foundry pool arrangement is weaker than its repeated citation suggests. It is a Service Agreement formed "by accessing and using the Pool" (13335), which reads as standard terms rather than a negotiated contract. Several Bitmain agreements are also "non-fixed price" or "future" purchase agreements (12899), and the "strategic relationship" with Bitmain promising "preferred access" (12891) is descriptive language without disclosed terms.

Naming is not the problem here, because every counterparty is identified. Related-party exposure deserves more attention. Beowulf E&D is owned and controlled by the CEO (12845) and receives an $8.46 million annual base fee (12867) plus a $339,200 transition-services payment from Nautilus (12942). NovaWulf is disclosed as an affiliate (13849). The claim count also overstates breadth, since roughly a dozen ids repeat the same NYDIG and Foundry agreements across filings.

Concentration runs through operations rather than customers. Revenue depends on a single pool operator, Foundry, and a single custodian and execution venue, NYDIG (13441). Power comes from NYPA at an initial 90 MW, with expansion described only as "potential" (12842), and from the Susquehanna nuclear supply at about 2.0 cents/kWh (12859, 12838). Additional Nautilus capacity is an option held by Cumulus Coin rather than TeraWulf, and it expires May 13, 2024 (12922).

Revenue cannot be reconciled with these deals because the record discloses no revenue, hashrate or contract values. The only dollar figures are costs. One internal inconsistency needs resolving. A claim dated 2022-08-28 gives TeraWulf 33.33% of Nautilus (13714), while later filings put the stake at 25% after the August 2022 amendment (12872, 12860, 12840), with both figures still "subject to adjustment based on relative capital contributions" (12938). The record ends in March 2023, so any later hosting or compute customer contracts fall outside this assessment.

TeraWulf's record through March 2023 contains no signed customer agreements in the conventional sense. The company self-mines bitcoin, so the revenue counterparty is the network rather than a named buyer, and the claims describe service providers, suppliers, a power seller, a joint-venture partner and a related party. What the record does show is signed documentation, dated and identified. The NYDIG Custodial Agreement of March 10, 2022 (13412, 13333, 13161) and the NYDIG Execution Agreement of September 16, 2022 (13414, 13334, 13345) are each disclosed with an execution date. The Foundry pool relationship is a service agreement formed by accessing the pool (13346, 13335), which is weaker than a negotiated contract. Equipment supply is backed by executed purchase agreements with MinerVA and Bitmain (12899, 12847), so the softer 'strategic relationship' language about Bitmain (12891) has contractual support behind it. Many entries restate the same few agreements across successive filings, and the real counterparty count is small. Every counterparty is named, and nothing is described as an unidentified 'anchor customer' or 'strategic partner'. The conditional language sits elsewhere. The NYPA supply covers an initial 90 MW with only a 'potential to expand' (12890). Cumulus holds an option, not an obligation, to add up to 100 MW before May 13, 2024 (12922). The first-of-its-kind nuclear framing is explicitly an expectation (12923). Concentration moved in the direction of dependence on Talen. The Nautilus stake fell from 33.33% (13714) to 25% (12872, 12938, 12860), and the board is set at one TeraWulf manager against four for Cumulus (12874). Bitcoin distributions follow hash-rate contribution net of costs (12939), so TeraWulf's economics at the site depend on a venture it does not control. The related-party thread also deserves attention. Beowulf E&D, which is owned by the CEO, receives an $8.46 million annual base fee (12867, 12845) and a $339,200 transition-services payment from Nautilus (12942). The disclosed amounts on record are costs and rates, such as 2.0 cents/kWh power (12859, 12838), rather than revenue, so the revenue arithmetic cannot be reconciled against these claims. The record as supplied also ends mid-entry at claim 12852, and any later hosting or compute contracts fall outside this read.

TeraWulf's record through March 2023 contains almost no customer agreements in the conventional sense. The company describes its revenue as coming mainly from self-mined bitcoin (claim 13441). Most of the named, executed contracts are therefore supplier, service, power or joint-venture arrangements rather than sales to customers. The signed documents are specific and dated. They include the NYDIG Digital Asset Custodial Agreement of March 10, 2022 (claims 13412, 13344, 12904), the NYDIG Execution Agreement of September 16, 2022 (claims 13414, 13334), the Foundry pool Service Agreement (claims 13335, 13346), and the amended and restated Talen joint venture agreement of August 27, 2022 (claims 12901, 12848). The equipment contracts with MinerVA and Bitmain are also counted, as one purchase agreement and nine sales and purchase agreements (claims 12899, 12847). Softer language sits alongside these. The Bitmain 'strategic relationship' promising 'preferred access and advantaged cost structure' is an unquantified characterization, not a separate contract (claim 12891). The NYPA supply is firm only for an initial 90 MW, with expansion described as potential (claims 12890, 12842). The Nautilus 100 MW expansion is an option held and funded solely by Cumulus Coin before May 13, 2024, not a TeraWulf commitment (claim 12922).

All counterparties that matter here are named, with one gap. Claim 12852 refers to bitcoin 'earned from hosting arrangements' without identifying any hosting client, which is the closest this record comes to an unnamed customer. The related-party thread also deserves attention. Beowulf E&D is owned and controlled by the CEO (claim 12845). It receives an $8.46 million annual base fee (claim 12867) and separately $339,200 for transition services to Nautilus (claim 12942). These are cost obligations to an insider, not arm's-length commercial validation.

Counterparty concentration is high and structural. Foundry is the single pool operator. NYDIG is both sole custodian and sole sale executor. Bitmain dominates hardware supply. The Nautilus economics depend on Talen affiliate Cumulus Coin, which holds 75% of the equity and four of five board seats (claims 12872, 12874). In the window covered, that concentration was restructured rather than diversified: the August 2022 amendment cut TeraWulf to 25% of Nautilus (claim 12860).

The revenue arithmetic cannot be reconciled from this record. None of the claims discloses bitcoin revenue, hash rate, hosting income or Nautilus distribution amounts. Biweekly distributions are described only by formula, net of power and operating costs (claim 12939). The one hard pricing input is Nautilus power at about 2.0 cents per kWh (claims 12859, 12838). The claim list also ends abruptly at claim 12853, so any later customer or contract disclosures fall outside this read.

TeraWulf's record contains almost no customer claims in the conventional sense. What it holds is a set of executed operating, supply and financing agreements. Mining revenue comes from a pool, not from contracted buyers. The company says the majority of its revenue is self-mined bitcoin held with NYDIG Trust Company (12170), earned through a pool run by Foundry Digital LLC (12162, 12903, 12153). Sales of that bitcoin go through a Digital Asset Execution Agreement with NYDIG Execution (12906, 12853). None of this is MOU, LOI or pilot language. The NYDIG custody agreement is dated March 10, 2022 (13161, 12904), and the related deposit account control agreement with Wilmington Trust is dated March 11, 2022 (12905). The Talen joint venture was amended and restated on August 27, 2022 (12901) and restated again in March 2023 (12921). The NYPA power agreement (12890, 12842) and the Bitmain and MinerVA purchase agreements (12899, 12847) are also described as executed. The softer wording sits around these contracts rather than replacing them. Bitmain is called a strategic relationship offering preferred access (12891). NYPA capacity carries the potential to expand (12890). Nautilus is expected to be the first fully behind-the-meter nuclear mine (12923).

On naming, every counterparty is identified, with one exception. Claim 12852 refers to bitcoin earned from hosting arrangements without naming any hosting client, and no hosting contract appears elsewhere in the record. Readers should treat hosting as an unquantified revenue source with no named counterparty.

Concentration has narrowed rather than broadened. The August 2022 amendment cut TeraWulf's Nautilus stake to 25% (12860, 12920). The March 2023 restatement gave Talen's Cumulus subsidiary 75% of the equity and four of five board seats (12872, 12938, 12874). The same agreement gave Cumulus alone a 100 MW expansion option, which it funds itself, running to May 13, 2024 (12922). By the September 2023 filings, the recurring counterparties are Foundry, NYDIG and Beowulf E&D (12153, 12152, 12171). Beowulf E&D is a related party owned by the CEO (12845) and is paid a reduced base fee of $8.46 million a year (12867). It also received $339,200 under the Nautilus transition services agreement (12942). Beowulf is therefore a material cost counterparty that is not at arm's length.

The revenue arithmetic cannot be reconciled from this record, because no revenue, hashrate or hosting figures are disclosed. The dollar and rate terms on file are all on the cost side: the roughly 2.0 cents/kWh Nautilus power price (12859, 12838), the Beowulf fee, and the transition services payment. Nautilus bitcoin is distributed every two weeks by hash-rate share after costs (12939), so TeraWulf's economics there depend on its contributed hashrate, not its 25% equity. Several entries are also duplicate restatements of the same filing text, which overstates the apparent number of relationships.

The record TeraWulf has put on file is mostly not about customers. Across 2023 and early 2024, nearly every named counterparty is a supplier, vendor or co-investor rather than a buyer of TeraWulf's output. Foundry Digital is the mining pool operator (12153, 12088, 12064), NYDIG Trust Company is the bitcoin custodian (12152, 12089, 12022), Bitmain supplies miners (12110, 12065), and the Power Authority of the State of New York supplies 90 MW of power (12098, 12027, 12016). Talen Energy is a 75% joint-venture partner in Nautilus and supplies nuclear power at a fixed 2.0 cents per kWh (12117, 12085, 12102, 12046). Beowulf E&D provides administrative and infrastructure services under a signed, dated agreement (12171, 12154). Beowulf is overseen by TeraWulf's CEO (12066) and is an affiliate of the landlord Somerset, which holds board observer rights (5587, 5588), so it should be read as a related party rather than a commercial partner. Many of these entries repeat the same filing language across several documents, which inflates the apparent volume of relationships. In this period, revenue came mainly from self-mined bitcoin (12170, 12089). The only buyer-side disclosure is hosting services to unnamed 'third-party entities' (12067), a counterparty that is never identified.

The first named customer appears in December 2024, when G42 is cited for over 70 MW of data center infrastructure (1051), followed by Core42, a G42 company (5225). Here the language is mixed. 'Partnership' (5225) is promotional framing. The May 2025 reference to 'lease agreements' (4994) points to executed contracts, but it also describes a 135 MW expansion option that had already lapsed or been exercised by March 31, 2025. The record on file does not say which. The companion claim 4993 on long-term HPC hosting is truncated in the material provided, so its terms cannot be assessed.

The resulting concentration is stark. Customer exposure moved from effectively none, with the bitcoin network as the revenue source, to a single sovereign-linked group, G42/Core42, with no second named hosting tenant on the record. On revenue arithmetic, nothing in these claims reconciles because nothing is quantified. No contract values, lease rates, terms in years or recognized hosting revenue are disclosed alongside the megawatt figures. The 70 MW and optional 135 MW numbers therefore cannot be tied to reported revenue from this record alone. Diligence should obtain the executed Core42 lease terms, confirm whether the expansion option was exercised, and identify the unnamed hosting clients.

TeraWulf's record through mid-2025 contains very little MOU, LOI or 'discussions' language, but most named agreements are not customer contracts. The 2023–2024 disclosures describe a self-mining business whose counterparties are suppliers, service providers and a joint-venture partner. These are the NYPA 90 MW power purchase agreement (claims 12098, 12027, 12044), the Foundry mining pool (12088, 12064, 12021), NYDIG custody (12089, 12022), Bitmain miner purchases (12110, 12065), and administrative services from Beowulf E&D, a related party overseen by TeraWulf's CEO (12103, 12066, 5588). Talen is an equity partner holding 75% of Nautilus, not a customer (12101, 12085, 12063). The 2.0-cent nuclear power contract is an input cost (12102, 12028). The only revenue-generating third-party activity in that period is 'occasional' miner hosting for unnamed 'third-party entities' (12067). No counterparty for it is identified.

The first signed customer contract with a named counterparty is the December 23, 2024 long-term lease with Core42 Holding US LLC, a G42 company (4993). It was earlier framed as a 'partnership' (5225) and as delivery of 'over 70 MW' for G42's US operations (1051). The 135 MW expansion option was exercisable through March 31, 2025 (4994), but the record contains no disclosure that it was exercised. It should be treated as lapsed or unconfirmed rather than contracted capacity. In August 2025 a second named tenant appears: Fluidstack USA I Inc., under an executed datacenter lease for CB-5 at Lake Mariner (4508, 4457). Google backstops certain of Fluidstack's obligations (4509, 4458). The headline '200+ MW, 10-year' figure (1054) exceeds the 160 MW CB-5 lease actually described (1053). The remaining capacity is not tied to a specific filed agreement in the record.

Customer concentration has therefore moved from none, in a self-mining model, to one named tenant (Core42) and then two (Core42 and Fluidstack). Credit support for the larger tenant rests heavily on Google's backstop. Fluidstack's own balance sheet is not the main support.

The record does not allow the revenue arithmetic to be reconciled. None of these claims disclose contracted lease revenue, per-MW pricing, or recognized hosting revenue. The statement that most revenue comes from self-mined bitcoin (12089) suggests that, as of early 2024, customer revenue was immaterial. Readers should seek the filed lease exhibits and subsequent segment revenue before crediting the MW headlines. Claim 4458 is truncated and should be verified against the source.

TeraWulf's announced customer relationships comprise solely signed lease and service agreements with fully identified counterparties; no MOUs, LOIs, or pilot arrangements appear in disclosed claims. Fluidstack dominates the portfolio across three facility footprints—Lake Mariner with 378 MW under lease (claims 3630, 3640) and an additional 160 MW expansion (claim 1053), plus a 168 MW AI compute joint venture at Abernathy under a 25-year lease (claims 3634, 3642)—totaling approximately 546 MW across 10–25 year lease terms signed between August 2025 and October 2025. Core42, a G42 subsidiary, committed 60 MW via December 2024 signed lease (claim 3639). Google appears not as a direct customer but as a credit backstop guaranteeing Fluidstack lease performance (claim 4509). Beowulf (Somerset affiliate) holds exclusive operator rights under service terms (claim 5588), while Somerset possesses board observation and registration rights (claims 5587, 5586). No unnamed "strategic partners" or "anchor customers" dilute the disclosure; counterparty concentration is unambiguous.

Yet the revenue arithmetic diverges materially. In February 2026, TeraWulf reported "522 critical IT MW totaling over $12.8 billion of revenue" (claim 11750). Breaking down the same filing: Core42 accounts for 60 MW, Fluidstack at Lake Mariner for 380 MW (per claim 11752), and Fluidstack at Abernathy for 168 MW, summing to 608 MW. The 522-to-608 MW gap remains unexplained—whether reflecting capacity under development, timing of operational commencement, or duplication in the Lake Mariner facility's lease structure. Fluidstack represents a minimum 80–90% concentration by contracted MW. This structure—extended long-term leases with a single AI infrastructure customer backed by Google's financial credit—minimizes volume volatility but centralizes refinancing and counterparty credit risk. The undisclosed basis for reconciling announced MW to disclosed revenue suggests either conservative reporting or incomplete transparency regarding capacity readiness or lease timing.

TeraWulf's customer disclosures uniformly employ signed agreement language without MOU, LOI, or pilot qualifications. Core42, a subsidiary of G42, executed long-term data center lease agreements in December 2024 for 60 MW of critical IT load (claims 4993, 11751). Fluidstack became the primary growth driver through a signed three-lease arrangement covering 378 MW at Lake Mariner in August 2025 (claim 3630) and a 50.1%-owned joint venture with 168 MW under a 25-year lease at Abernathy in October 2025 (claim 3634). The New York Power Authority provides 90 MW under a ten-year agreement dated February 2022 (claim 3627). All major counterparties are explicitly named; no capacity is described as "strategic partnership" or "anchor customer" without identification.

Counterparty concentration has consolidated decisively toward Fluidstack following an initial diversification attempt. Early announcements referenced G42 at over 70 MW (claim 1051), but this eventually crystallized into Core42 at 60 MW, representing a retreat from the initially projected scale. Fluidstack now dominates: the Lake Mariner and Abernathy facilities combined deliver 534–546 MW (claims 11727–11728, 3652), representing approximately 95% of total disclosed contracted capacity as of February 2026.

A material discrepancy exists between unit metrics and aggregate revenue claims. TeraWulf reports 522 contracted critical IT MW generating over $12.8 billion in lifetime revenue (claim 11750), yet the enumerated signed agreements total 594–606 MW—Core42 (60 MW), Lake Mariner (366–378 MW per claims 11727 and 3652), and Abernathy (168 MW). This 72–84 MW gap lacks explanation; it remains unclear whether the 522 MW figure represents operational capacity only, excludes a facility, or reflects rescoping of announced agreements. The Lake Mariner capacity itself has undergone multiple restatements: 160 MW (claim 1053), then 200+ MW (claim 1054), then 366–378 MW (claims 11727 and 3652). The $12.8 billion revenue figure is unallocated across counterparties and contract durations, obscuring unit economics and preventing verification that the per-MW revenue implied by this aggregate is consistent across Fluidstack's two largest facilities and the Core42 arrangement.

TeraWulf's disclosed partnership portfolio distinguishes sharply between genuine customer contracts and governance or operational arrangements. The named customers with signed agreements are Core42 and Fluidstack, both substantial players in AI infrastructure. Core42, a G42 subsidiary, committed to 60 MW of critical IT load under long-term lease agreements executed in December 2024 (claims 3628, 3639). Fluidstack, however, dominates the portfolio with three distinct signed lease agreements totaling 546 MW across Lake Mariner (378 MW, claims 3630, 3640) and the Abernathy joint venture (168 MW, claims 3634, 3642), all executed in or by August 2025.

By contrast, Somerset's arrangements—board observer rights, Beowulf as exclusive operator, and Form S-3 registration commitments (claims 5586, 5587, 5588)—constitute governance and financial mechanics, not customer revenue contracts. Heorot's two-year transition services agreement (claim 4834) reflects post-acquisition support, not ongoing capacity sales. Google's role is as a credit backstop for Fluidstack's obligations (claims 4509, 4458), not as a direct tenant. Morgan Stanley and Cantor Fitzgerald appear as financial advisors, not customers. This distinction matters: only Core42 and Fluidstack generate disclosed utilization revenue.

Counterparty concentration has tightened dramatically. Core42's 60 MW represented meaningful diversification through early 2025; Fluidstack's 546 MW emergence has compressed this into near-total customer concentration. Of the 522 critical IT MW claimed as contracted (claim 11750), Fluidstack accounts for approximately 90%, with Core42 representing 11%. The joint venture structure further complicates the customer relationship: TeraWulf's 50.1% ownership in Abernathy (claims 3654, 11740) blurs the line between landlord and equity partner.

Revenue reconciliation presents a puzzle. Claim 11750 cites $12.8 billion across 522 contracted critical IT MW, yet announced capacity totals 606 MW (60 Core42 plus 546 Fluidstack). This 84 MW gap may reflect timing (staggered commencement dates, claim 11752 notes "commencing in 2026"), definitional differences between critical IT load and total build, or unannounced capacity reductions. Google's backstop arrangement (claims 4509, 4458) suggests TeraWulf required credit enhancement on its largest customer relationship, raising questions about Fluidstack's creditworthiness that disclosure does not directly address.

TeraWulf's disclosed customer engagements from October 2024 through April 2026 are predominantly signed, executed agreements rather than letters of intent, framework arrangements, or pilot discussions. The company consistently employs contractual language—"entered into," "agreed to," "will provide"—across all material customer relationships. Core42, a G42 subsidiary, executed binding long-term data center lease agreements in December 2024 (claims 4993, 3628), with documented expansion options expiring March 2025 (claim 4994). Fluidstack similarly appears as a named, definitive counterparty: the Lake Mariner lease was executed in August 2025 (claims 4508, 4457, 3640), followed by joint venture formalization in October 2025 (claims 3641, 3654) allocating 50.1% ownership to TeraWulf. No major announced relationship employs exploratory language—"discussions," "framework," or "strategic partnership without commitment." Somerset and NYPA received named, binding governance or supply rights (claims 5587, 3627). Only advisory arrangements with Cantor Fitzgerald (claim 10322) and Morgan Stanley (claims 10316, 10321) fall outside customer hosting categories, and these are service providers rather than counterparties.

Counterparty concentration has become severely unbalanced toward Fluidstack since mid-2025. Fluidstack commands over 546 megawatts of announced capacity—378 MW at Lake Mariner's Akela facility (claims 3630, 3652) plus 168 MW at Abernathy under a 25-year lease (claims 3642, 3634)—against Core42's 60 MW with a 135 MW expansion option (claims 3628, 4994) and NYPA's 90 MW power supply (claims 3627, 2114). The concentration risk intensified in Q3 2025 when Fluidstack executed three separate lease agreements (claim 3640) and assumed joint venture control at Abernathy, effectively tying TeraWulf's highest-capacity facilities to a single customer. Google's backstop guarantee for Fluidstack (claims 4509, 4458) signals embedded credit risk within this dominant relationship.

Disclosed revenue metrics do not reconcile against announced capacity commitments. TeraWulf has not published unit pricing, take-or-pay minimums, or escalation schedules for any customer cohort. TeraWulf's minority equity stakes in Fluidstack joint ventures (50.1% ownership at Abernathy, claim 3654) defer profit recognition and complicate revenue realization timing relative to the announced megawatt-year commitments. Without disclosed pricing architecture, capacity utilization baselines, or escalation terms, the financial contribution of these named agreements remains opaque.

TeraWulf's customer base comprises exclusively named, signed data center lease agreements rather than preliminary frameworks, MOUs, or undefined partnerships. Core42, identified as a G42-affiliated company providing sovereign cloud and AI infrastructure, signed binding long-term data center lease agreements in December 2024 for 60 megawatts at the La Lupa facility (claims 3651, 3639, 3628). Fluidstack, explicitly named as an AI cloud platform operator, executed definitive lease agreements beginning August 2025 for 378 megawatts at Lake Mariner (Akela) and committed to a 25-year lease for 168 megawatts at Abernathy through an October 2025 joint venture agreement (claims 3642, 3652, 3633). No unnamed anchor customers, pilot programs, or framework agreements surface in customer relationships; counterparties are identified. Counterparty concentration has evolved sharply toward Fluidstack dependency. Core42's 60-megawatt footprint represented the initial major customer in Q4 2024, but Fluidstack has captured 546 megawatts cumulatively—approximately 90 percent of disclosed customer capacity. Notably, the Abernathy expansion operates via a 50.1 percent joint-venture structure (claims 3966, 3970), diluting TeraWulf's direct economic interest in the 168-megawatt lease. Supporting relationships include the New York Power Authority supply agreement (90 MW since February 2022, claim 3627), which is utility infrastructure rather than customer revenue, and transition-services agreements with the Heorot acquired company (claim 4834). A Google backstop guarantee underpins Fluidstack payment risk (claims 4509, 4458). Revenue reconciliation fails: filings disclose megawatt commitments without pricing, lease escalations, or revenue-recognition methodology, precluding validation of whether announced capacity aligns with financial projections. The combination of incomplete pricing disclosure and extreme Fluidstack concentration—offset only modestly by Core42—presents material revenue counterparty risk unresolved by board observation rights granted to Somerset (claim 5587).

TeraWulf's disclosed customer base reflects a clear progression from operational partnerships toward capacity-focused agreements with large-scale AI infrastructure buyers, though disclosure transparency varies materially across counterparties. The earliest material relationship is with Somerset, memorialized through signed lease-related terms including board observer rights and exclusive operational control through Beowulf Electricity & Data Inc. (claims 5586-5588). This appears foundational to facility control but not a primary revenue driver in disclosed filings. The company entered into explicitly 'long-term data center lease agreements' with Core42 on December 23, 2024, committing 60 MW of capacity with a material expansion option for an additional 135 MW exercisable through March 31, 2025 (claims 4993, 4994, 3651). This is unambiguously a signed customer agreement with specified power allocation and term structure. However, Core42 is described in claim 5225 as part of a broader 'partnership,' which obscures whether this constitutes the entire relationship or contains pilot or framework components. Fluidstack emerges as the dominant customer concentration point, with signed capacity commitments substantially exceeding all other counterparties combined. In August 2025, Akela entered three data center lease agreements with Fluidstack USA I Inc. committing 378 MW (claims 3640, 3652, 2105), while separate lease agreements committed an additional 168 MW at Abernathy under Fluidstack USA III with a 25-year term (claims 3641, 3634). These agreements appear unambiguously signed with quantified capacity and duration. TeraWulf additionally structured joint venture ownership (50.1% TeraWulf, 49.9% Fluidstack) via Flash Compute and Abernathy entities (claims 3966, 3654), suggesting customer relationships have deepened beyond simple lease arrangements into shared infrastructure governance. Google's role warrants clarification: the company did not directly sign capacity agreements but instead provided backstop guarantees for Fluidstack lease performance (claims 4509, 4458), suggesting credit enhancement rather than primary customer status. The company disclosed a 10-year, 200+ MW agreement with Fluidstack in August 2025, but subsequent filings detail specific lease components totaling 546 MW without reconciling this aggregate against the headline figure or quantifying revenue impact. Core42's expansion option adds execution uncertainty regarding final capacity utilization through the March 2025 exercise window.

TeraWulf's disclosed customer commitments distinguish sharply between executed lease agreements and advisory relationships. The two primary revenue customers—Fluidstack and Core42—are explicitly named counterparties with signed long-term binding agreements, not MOUs or frameworks. Fluidstack USA I Inc. entered into three executed data center lease agreements at Akela for 378 MW of critical IT load (claims 3630, 3652, 3640), while Fluidstack USA III signed a 25-year pre-lease for the full 168 MW Abernathy build-out (claims 3634, 3642). Core42 Holding US LLC executed long-term data center lease agreements committing 60 MW at La Lupa in December 2024 (claims 3651, 3628, 3639). Both relationships are structured as binding leases with defined terms and capacity, satisfying the legal hallmarks of signed customer contracts rather than pilot or framework arrangements.

Disclosure precision erodes at the periphery. Google's involvement appears as a backstop guarantee for Fluidstack payment obligations (claims 4509, 4458), positioning Google as a credit enhancement mechanism rather than a direct revenue customer. Somerset receives board observer rights (claim 5587) and Beowulf Electricity & Data Inc. continues as exclusive campus operator (claim 5588), functions best characterized as infrastructure partners. The Hawesville seller receives a 6.8% minority equity interest (claim 3878) but remains unnamed. Morgan Stanley and Cantor Fitzgerald (claims 10316, 10321, 10322) appear as financial advisors and underwriting agents, not customers.

Counterparty concentration has crystallized around Fluidstack dominance. The entity controls 546+ MW across Akela and Abernathy leases plus a 49.9% equity interest in Flash Compute (claims 3970, 3956), creating structural revenue dependence on a single colocation operator. Core42 represents the secondary anchor at 60 MW. The New York Power Authority supplies enabling power infrastructure—90 MW under a ten-year agreement executed in February 2022 (claims 3627, 2114)—but is not a revenue customer.

Revenue arithmetic remains unreconciled. Claims specify MW capacity allocations and lease terms spanning ten to twenty-five years, yet disclose no unit pricing, revenue per megawatt, or annual revenue projections from signed agreements. Without pricing transparency, announced megawatt commitments cannot be verified against reported or guidance revenues. The gap between stated operational capacity and financial disclosure prevents readers from assessing whether the 546+ MW Fluidstack relationship and 60 MW Core42 agreements align with the company's revenue expectations or historical unit economics.

TeraWulf's customer and partnership portfolio exhibits acute concentration despite a veneer of diversification across formally named counterparties. All major counterparties are explicitly identified—Somerset (investor, claims 5587–5588), Core42 (lessor, claims 4993, 3639), Fluidstack (lessor and joint venture partner, claims 3630–3634, 3641), NYPA (utility, claim 3627)—rather than branded as anonymous 'strategic anchors.' However, two counterparties dominate signed revenue contracts. Fluidstack occupies 546 MW across Lake Mariner's 378 MW (claim 3630) and Abernathy's 168 MW (claim 3634), each under executed 25-year and long-term data center leases (claims 3640, 3642). Core42 contributes 60 MW at La Lupa under December 2024 lease agreements (claim 3639) plus 70 MW for G42's U.S. operations (claim 1051). Combined, Fluidstack and Core42 represent 676+ MW of the disclosed contract pipeline—a two-customer dependency at roughly 90 percent of named capacity.

The timing and contractual character vary meaningfully. Core42's agreements, signed December 23, 2024 (claim 4993), carried a 135 MW expansion option expiring March 31, 2025 (claim 4994); silence in post-April filings implies non-exercise. Fluidstack's leases are fully executed as of August 2025 (claim 3630). The Abernathy joint venture (October 2025, claim 3641) structures TeraWulf as 50.1% owner alongside Fluidstack at 49.9%, complicating pure customer revenue recognition. Somerset's board observer and Form S-3 registration commitments (claims 5586–5587, October 2024) represent investor protections, not customer contracts. Heorot's two-year transition services agreement (claim 4834, May 2025) is operational support, not capacity revenue.

Revenue arithmetic proves opaque. TeraWulf discloses 676+ MW of named capacity commitments but does not publish lease rates, power densities, or expected annual contract values in these filings. Joint venture structures (Fluidstack partnerships at 50.1% ownership) imply revenue-sharing rather than gross capacity pricing. NYPA's 90 MW, locked in February 2022 (claim 3627), functions as fixed power supply cost, not customer revenue. Without itemized pricing per MW, estimated term lengths, or escalation clauses, investor verification of total contract value against public revenue guidance remains impossible from disclosed information. This opacity compounds the concentration risk: two counterparties controlling strategic capacity, yet financial contribution unmeasurable against audited results.

TeraWulf's disclosed customer relationships consist entirely of signed agreements rather than preliminary frameworks, though the company describes arrangements using varying contractual nomenclature that obscures concentration risk. Somerset received board observer rights and operational control through Beowulf affiliate in October 2024 (claims 5587–5588), representing a capital partnership rather than revenue-generating customer. The material customer base emerged later: Core42, a G42 subsidiary, signed long-term data center leases in December 2024 for stated capacity of seventy megawatts with an expansion option (claims 1051, 4993–4994), though subsequent filings confirm only sixty megawatts of committed critical IT load (claims 3628, 3651), suggesting the expansion option was not exercised or was revised downward. The narrative then consolidates almost entirely around Fluidstack, which evolved from a straightforward hosting lessee into a joint-venture partner. In August 2025, Akela signed three separate data center lease agreements with Fluidstack USA I for three hundred seventy-eight megawatts (claims 3630, 3640, 3652), while a parallel one hundred sixty-megawatt lease for Lake Mariner's CB-5 was executed (claims 1053, 4457–4458). By October 2025, TeraWulf had restructured the relationship into a joint venture, FS CS I LLC, in which TeraWulf holds fifty-point-one percent equity and Fluidstack holds forty-nine-point-nine percent (claims 3966, 3956), governing a one hundred sixty-eight-megawatt HPC campus at Abernathy under a twenty-five-year lease to Fluidstack USA III (claims 3633–3634). Google provided contractual credit support for Fluidstack's lease obligations rather than direct hosting services (claims 4509, 4458). Concentration risk is acute: Fluidstack represents five hundred forty-six megawatts of disclosed capacity commitments, or seventy-eight percent of new announced customer capacity, compared to Core42's sixty megawatts and the pre-existing NYPA agreement's ninety megawatts. The company discloses no pricing schedules, revenue per megawatt, or actual cash received, rendering the revenue reconciliation opaque. Somerset and financial advisors (Morgan Stanley, Cantor Fitzgerald) are named by role rather than contractual substance, whereas customer naming emphasizes capacity metrics divorced from economic performance.

TeraWulf's disclosures identify four named counterparties under binding data center lease agreements: the New York Power Authority (90 MW, signed February 2022), Core42 (60 MW, signed December 23, 2024), and Fluidstack under two structures—378 MW at Akela (signed August 2025) and 168 MW at Abernathy (signed October 2025). No customer arrangements are disclosed as MOUs, letters of intent, or pilot engagements; all major commitments employ binding past-tense agreement language. An October 2024 announcement referenced "over 70 MW of Data Center Infrastructure for G42's US Operations" (claim 1051), but subsequent filings clarify Core42, a G42 subsidiary, as the actual lessee rather than the parent company itself. Google is disclosed as a backstop guarantor of Fluidstack lease obligations (claims 4509, 4458) rather than a primary customer. Somerset and Beowulf appear in governance and operations roles related to TeraWulf's equity capital, not as customers. The company has not publicly identified unnamed "strategic partners" or unnamed "anchor customers."

Concentration has shifted sharply toward Fluidstack since mid-2025. Through 2024, NYPA and Core42 were the only disclosed lessees. Fluidstack's arrival in August 2025 reshaped the portfolio: it now represents 546 MW out of 696 MW of disclosed committed capacity (78.4 percent). The October 2025 joint venture at Abernathy, in which TeraWulf holds 50.1 percent ownership (claims 3966, 3970, 3956), embeds structural alignment beyond a traditional lessor-tenant arrangement.

The claims record discloses no per-MW pricing, rent escalation, ramp schedules, or revenue recognition policies. Without pricing data, TeraWulf's reported financial performance cannot be assessed against the MW-denominated customer commitments, nor can the alignment between disclosed capacity and stated financial guidance be evaluated.

TeraWulf's disclosed customer base consists entirely of named counterparties rather than anonymized 'strategic partners,' with material commitments concentrated in three core relationships. Somerset receives non-voting board observer rights under a 2024 continuation agreement (claim 5587), while its affiliate Beowulf retains exclusive operator status—a structural arrangement that differs from capacity-based customer deals. Core42, identified as a G42 subsidiary, entered into binding long-term data center leases beginning December 2024, committing TeraWulf to deliver 60 MW at La Lupa (claims 3628, 3639) following an initial announcement of 70 MW of infrastructure delivery (claim 1051). These Core42 agreements use explicit "entered into" language establishing binding leases through 2025 onward (claims 4993, 4994), though an expansion option for 135 MW remained exercisable only through March 31, 2025 (claim 4994), distinguishing conditional capacity from signed obligations.

Fluidstack emerged as the dominant customer beginning August 2025, with signed lease agreements covering 378 MW at Lake Mariner (Akela campus) under 10-year terms and 168 MW at Abernathy under 25-year leases (claims 3630, 3652, 3634, 3642). These commitments evolved into a joint venture structure by October 2025, with TeraWulf holding 50.1% of Flash Compute through subsidiary Big Country Wulf LLC and Fluidstack holding 49.9% (claims 3966, 3970). Google's backstop guarantee of Fluidstack obligations (claims 4509, 4458) indicates third-party confidence in Fluidstack's counterparty credit. Counterparty concentration has shifted markedly: Somerset and Beowulf anchored 2024, Core42 drove late 2024 through early 2025 commitments, and Fluidstack's 546 MW aggregate commitment across two campuses plus joint venture participation dominates 2025–2026 disclosed capacity. Secondary relationships include a two-year transition services agreement with Heorot (claim 4834) and the New York Power Authority's longstanding 90 MW power supply arrangement executed in February 2022 (claims 3627, 2114).

No revenue figures appear in these claims, preventing reconciliation of announced megawatt commitments against reported customer revenue. The agreements consistently employ binding contractual language—"entered into," "agreed to lease"—rather than MOU or pilot language, establishing these as signed arrangements rather than framework discussions. All material counterparties are named entities, eliminating ambiguity over customer identity.

TeraWulf's disclosed partnership portfolio is characterized by extreme customer concentration and predominantly signed lease agreements rather than framework arrangements or pilot discussions. All material counterparties are explicitly named: Fluidstack, Core42 (a G42 subsidiary), and Google appear as the only substantive revenue or quasi-revenue counterparties disclosed across 49 claims spanning October 2024 through May 2026. No 'strategic partners' or 'anchor customers' lack identification in the filing record provided.

The concentration is stark. Fluidstack alone commits to 546 MW of capacity across two separate structures: 378 MW under three lease agreements at the Akela/Lake Mariner campus (claims 3640, 3652) executed August 2025, and 168 MW at Abernathy under a 25-year lease (claim 3634) formalized October 2025. Core42 committed to 60 MW at La Lupa in December 2024 (claim 3639), with an option to expand by 135 MW by March 31, 2025 (claim 4994)—though no subsequent claim evidences exercise of that option. These dominate TeraWulf's announced pre-leasing. Google's role is contingent: it backstops Fluidstack payment obligations (claims 4509, 4458) but appears as guarantor rather than direct customer.

The language across claims consistently indicates executed agreements rather than letters of intent or framework discussions. Claims use formulations such as 'entered into' and 'agreed to,' standard for binding leases and joint venture agreements, with no MOU or LOI designations. The Abernathy and Flash Compute structures (claims 3641, 3654, 3956, 3970) establish joint venture ownership at 50.1% TeraWulf / 49.9% Fluidstack, which materially dilutes revenue capture relative to wholly-owned capacity. The Somerset arrangement (claims 5587, 5588, 5586) grants board observation and operational control to an affiliate, suggesting debt or structural investment protection rather than a customer relationship.

Critically, the disclosed claims provide megawatt commitments but no pricing, utilization rates, or revenue figures. Without per-MW lease economics or capacity utilization assumptions, the announced capacity cannot be reconciled against any disclosed revenue or guidance. The New York Power Authority agreement (claim 3627) and Hawesville minority equity interest (claim 3878) provide operational context but no revenue signal. The portfolio's extreme two-customer dependency—Fluidstack and Core42 representing potentially 85%+ of committed capacity—presents material concentration risk that single counterparty claims do not adequately convey.

TeraWulf's customer and partnership announcements consist of signed, named-counterparty agreements with no disclosed letters of intent, framework agreements, or pilot arrangements. The portfolio exhibits pronounced concentration in a single customer.

Fluidstack accounts for the dominant capacity commitment: 378 megawatts under data center lease agreements signed in August 2025 at Akela's Lake Mariner campus (claims 3630, 3640, 3652) and 168 megawatts under a 25-year lease at Abernathy HPC Campus with a 50.1% TeraWulf / 49.9% Fluidstack joint venture structure (claims 3642, 3641). Core42, a G42 subsidiary, is the second-named customer with 60 megawatts of capacity under a lease signed in December 2024 (claims 3651, 3628), plus an expansion option for 135 megawatts through March 31, 2025 (claim 4994). These commitments place Fluidstack at roughly 75 to 90 percent of TeraWulf's announced operational and joint-venture capacity, depending on whether Core42 exercises its expansion right.

Supporting arrangements include the New York Power Authority's 90 megawatt power supply agreement, signed February 2022 for a ten-year term (claim 3627), and a two-year transition services agreement with Heorot covering tax, accounting, and human resources functions (claim 4834). Somerset and its affiliate Beowulf occupy infrastructure and governance roles rather than customer positions (claims 5587, 5588). Google provides a credit backstop for Fluidstack's lease obligations, indicating lender concern about counterparty risk (claims 4509, 4458).

The disclosed record contains no revenue figures, per-megawatt pricing, or escalator percentages. All announced commitments are expressed as megawatt capacity over multi-decade lease terms, not annual revenue. Joint-venture language references annual escalators (claim 9306), but no escalator rates or base rents are disclosed. Without specific contract values, pricing schedules, or revenue rates per megawatt, one cannot reconcile the company's reported revenue or revenue guidance with the capacity it has committed to deliver. The concentration in Fluidstack, combined with the absence of pricing transparency, prevents meaningful assessment of revenue durability or the accuracy of the company's revenue recognition.

TeraWulf's publicly disclosed customer portfolio consists entirely of specifically named counterparties, with no vague references to 'strategic partners' or unnamed 'anchor customers,' though the arrangement with Hawesville seller (claim 3878) blurs the line by structuring compensation as minority equity rather than lease revenue. The company has signed binding long-term data center lease agreements with Core42 (a G42 subsidiary) for 60 MW effective December 2024 (claims 3639, 3628) and with Fluidstack across two facilities totaling 546 MW—378 MW at Lake Mariner/Akela (claims 3630, 3652) and 168 MW at Abernathy (claims 3633, 3634)—both executed in August and October 2025 respectively (claims 4508, 4457, 3654). Supporting infrastructure agreements with the New York Power Authority (90 MW, ten-year term from February 2022, claims 3627, 2114) and Beowulf Electricity & Data as exclusive operator (claim 5588) predate the major customer wins. Counterparty concentration has shifted dramatically: early filings reference Somerset board observer rights and transition services to Heorot (claims 5587, 4834), entities tied to legacy operations, but by mid-2025 Fluidstack represents approximately 64 percent of identified committed capacity. The Abernathy joint venture (claims 3633, 3641, 3654) further deepens Fluidstack's strategic footprint via a 50.1 percent TeraWulf ownership stake. Notably, Google has agreed to backstop Fluidstack's lease obligations (claims 4509, 4458), suggesting credit-quality concerns about the primary customer or contractual protections against customer default. The disclosed revenue arithmetic presents opacity: TeraWulf has announced approximately 696 MW of committed capacity (Core42 60 MW plus Fluidstack 546 MW plus NYPA 90 MW) yet has disclosed no per-megawatt pricing, contract values, or revenue guidance tied to these leases, making verification against publicly stated financial results impossible. The absence of revenue figures—despite multiple filings and amendments between December 2024 and April 2026—alongside Fluidstack's single-customer dominance and Google's backstop arrangement suggests either heavily negotiated rates or ongoing commercial uncertainty about customer demand realization.

TeraWulf's disclosed customer portfolio consists entirely of named, signed agreements rather than framework discussions or pilot arrangements, substantially reducing speculative risk around capacity deployment. Somerset (via affiliate Beowulf) received board observer rights and exclusive operational control beginning October 2024 (claims 5587, 5588), establishing the company's foundational landlord-tenant structure. Core42, a sovereign cloud subsidiary of UAE-based G42, executed long-term data center lease agreements in December 2024 for 60 MW at the La Lupa facility, marking entry of a hyperscaler-class counterparty (claims 4993, 3628, 3651).

Fluidstack has become the dominant customer, representing an estimated 70+ percent of disclosed capacity commitments. The company executed three binding data center lease agreements in August 2025 for 378 MW at Akela (claims 3640, 3652), followed by an amended joint venture agreement in October 2025 for the Abernathy HPC Campus with 168 MW under a 25-year lease (claims 3633, 3641, 3642). These are binding multi-decade arrangements, not pilots: Fluidstack USA I and USA III executed irrevocable leases with ownership formalized through 50.1 percent TeraWulf / 49.9 percent Fluidstack subsidiaries (claims 3966, 3970, 3956).

Concentration risk is material but structurally mitigated. Fluidstack's lease obligations carry a Google backstop guarantee (claims 4509, 4458), shifting payment default and technology obsolescence risk to an investment-grade counterparty. TeraWulf also signed a ten-year power supply agreement with the New York Power Authority in February 2022 for 90 MW (claims 3627, 2114), securing infrastructure-grade offtake certainty at a foundational utility scale.

Revenue reconciliation proves incomplete from disclosed claims. TeraWulf announced "200+ MW, 10-year AI hosting agreements with Fluidstack" (claim 1054) and multi-year commitments from Core42, yet filings do not disclose lease pricing, annual escalators, or near-term revenue run-rate attribution by counterparty. The Hawesville seller's 6.8 percent equity interest with redemption rights (claim 3878) represents contingent liability rather than customer revenue. Without itemized pricing and financial guidance, the dollar-value alignment between announced MW capacity and reported revenues remains opaque, though the contractual long-tails (25-year Abernathy, 10-year Fluidstack, multi-year NYPA power) imply indexed, predictable revenue pools over locked deployment horizons.

TeraWulf's customer base comprises exclusively named, signed agreements rather than exploratory arrangements. Core counterparties—Fluidstack, Core42, Somerset, and the New York Power Authority—are identified by corporate name and transaction type (lease, equity stake, power supply), not by vague descriptor language like 'strategic partners' or 'anchor customers.' This clarity reflects the infrastructure-lease model: electricity and data-center capacity are contractual commodities with defined terms.

The contractual portfolio rests on two major customer relationships. Core42, a G42 subsidiary specializing in sovereign cloud and AI infrastructure, signed long-term data center leases for 60 MW at La Lupa in December 2024 (claims 4993, 3651, 3628). Fluidstack USA I, positioned as a leading AI cloud platform, signed multiple leases totaling 378 MW at Lake Mariner's Akela campus in August 2025, and further structured a 50.1%-owned joint venture (Flash Compute / Abernathy HPC Campus) pre-leasing 168 MW to Fluidstack USA III under a 25-year term (claims 1054, 3630, 3633, 3642). These represent hard-contract commitments: 'entered into' language (claims 4993, 3628, 3630) and equity structures (claims 3970, 3966) indicate signed, binding arrangements rather than MOUs or pilots.

Counterparty concentration shifted dramatically from operator to customer relationships. Early 2024 positioned Somerset/Beowulf Electricity as the exclusive operational partner (claim 5588), warranting board-observer rights for Somerset (claim 5587). By mid-2025, this arrangement faded from new disclosures in favor of revenue-generating customer leases. Fluidstack's footprint expanded to dominate: across Akela and Abernathy joint-venture capacity, the company commands approximately 546 MW—roughly nine times Core42's committed 60 MW. Google's August 2025 backstop guarantee (claims 4509, 4458) signals confidence in Fluidstack's creditworthiness but also highlights single-customer concentration risk.

Revenue arithmetic remains opaque. TeraWulf discloses MW commitments (546 MW to Fluidstack, 60 MW to Core42, 90 MW to NYPA) and term lengths (25-year Fluidstack leases, long-term Core42 deals, ten-year NYPA arrangement executed February 2022) but withholds per-megawatt rates. Without pricing data, claimed hundreds-of-millions in long-term contracted revenue cannot be independently reconciled. The portfolio's stability depends partly on Fluidstack's solvency: Google's backstop covers payment defaults and insolvency scenarios (claims 4509, 4458), limiting downside but underscoring customer credit concentration.

TeraWulf's disclosed customer and partnership portfolio centers on four primary counterparties, all identified by name, with crystallized commercial arrangements rather than exploratory frameworks. The most substantial and recent commitments involve Fluidstack, which has evolved from simple lease arrangements into a multi-property ecosystem. In August 2025, TeraWulf's subsidiary Akela executed three data center lease agreements with Fluidstack USA I Inc., committing 378 MW of critical IT load at Lake Mariner (claims 3640, 3652), while establishing a 160 MW lease for the CB-5 facility (claim 1053). The October 2025 joint venture formalization—structured as 50.1% TeraWulf subsidiary and 49.9% Fluidstack—governing Abernathy HPC Campus's 168 MW, 25-year lease (claims 3654, 3642) signals deepening partnership integration rather than transactional relationships. Across these three facilities, Fluidstack represents approximately 706 MW of committed capacity, constituting roughly 80 percent of TeraWulf's disclosed customer capacity.

Core42, a G42 company, is the second major counterparty with two named facilities: 60 MW long-term leases at La Lupa executed December 2024 (claims 3651, 3639), plus earlier indications of 70 MW infrastructure delivery for G42's US operations (claim 1051). Unlike Fluidstack's joint venture structures, Core42 arrangements remain traditional lessor-lessee relationships with no equity participation disclosed, though their December 2024 entry marked TeraWulf's pivot toward sovereign AI infrastructure markets.

Somerset and its affiliate Beowulf occupy a distinct category—governance and operational partners rather than capacity customers. The October 2024 grant of board observer rights to Somerset (claim 5587) and confirmation that Beowulf continues as exclusive premises operator (claim 5588) reflect control and oversight rather than consumption arrangements. All major agreements employ signed language—"entered into," "agreed to," and "long-term lease agreements"—rather than exploratory terminology.

Google's role is limited to credit support, backstopping Fluidstack's lease obligations (claims 4509, 4458) rather than serving as direct customer. New York Power Authority supplies 90 MW under a 10-year power agreement predating recent customer announcements (claim 3627).

A material gap exists between MW commitments and revenue disclosure. The claims enumerate roughly 800+ MW across Fluidstack, Core42, and NYPA power, yet no specific pricing, annual revenue run-rates, or contract values appear in the provided filings. The "long-term" and "lease agreement" language implies recurring revenue, but arithmetic reconciliation requires access to pricing schedules not present in these public announcements. The Fluidstack concentration—68 percent of disclosed capacity through one counterparty—absent revenue validation, creates interpretive uncertainty about customer quality and revenue resilience.

TeraWulf has disclosed a portfolio of exclusively signed customer and partnership agreements rather than preliminary frameworks or MOUs. The company names all major counterparties, though governance structures vary significantly by deal type.

The dominant counterparty is Fluidstack, appearing across 17 claim IDs (1054, 4508, 4509, 4457, 4458, 1052, 3966, 3970, 3956, 3634, 3633, 3652, 3642, 3654, 2105, 2086, 2077) representing 906+ MW of committed capacity. These agreements range from traditional ten-year hosting leases at Lake Mariner (claims 1054, 4457) to joint venture interests. The Abernathy and Flash Compute structures (claims 3633, 3654, 3966, 3970) involve TeraWulf holding 50.1% ownership with Fluidstack at 49.9%, signaling material control and revenue participation beyond pure colocation.

Core42, identified as a G42 company (claims 5225, 4993, 4994), represents the second-largest counterparty with 60 MW committed at La Lupa since December 23, 2024 (claims 3639, 3651, 2104), plus an expansion option for 135 MW exercisable through March 31, 2025 (claim 4994). The option's current status remains undisclosed.

Somerset and Beowulf Electricity & Data Inc. occupy a distinct governance position with board observer rights and exclusive operator status (claims 5587, 5588) executed October 9, 2024, rather than traditional customer leases. Google provides backstop guarantees for Fluidstack payment defaults (claims 4509, 4458), creating implicit credit enhancement.

Concentration risk is material: Fluidstack-related agreements represent approximately 82% of disclosed committed capacity, with Core42 contributing 12–18% depending on option exercise. The portfolio lacks significant diversification beyond these two anchor entities.

Critical gap: TeraWulf discloses MW capacity commitments across multiple claims but provides no pricing, revenue recognition methodology, or fiscal impact reconciliation. Without tenant-level revenue per MW and term-weighted discounting, revenue arithmetic cannot be verified against announced deal volume.

TeraWulf's disclosed partnership portfolio consists almost entirely of signed, binding agreements rather than non-binding frameworks. All major counterparties—Beowulf, Core42, Fluidstack, Google, NYPA, and Heorot—are explicitly named in the record, with no vague references to 'strategic partners' or unnamed 'anchor customers.' The specificity extends to ownership structures: Fluidstack joint ventures are documented with precise equity splits (50.1% TeraWulf, 49.9% Fluidstack) and subsidiary names (Flash Compute, FS CS I LLC), indicating formal, legally binding arrangements rather than preliminary discussions.

Counterparty concentration has undergone rapid consolidation toward Fluidstack. Through late 2024, TeraWulf's capacity was fragmented among Beowulf (operator), Somerset (equity holder with board rights), and Core42 (60 MW lease signed December 23, 2024, with an unexercised 135 MW expansion option through March 31, 2025). By August 2025, Fluidstack emerged as the dominant off-taker, signing three data center lease agreements for 378 MW at Lake Mariner's Akela and initially agreeing to 200+ MW on a 10-year term. This concentration deepened through late 2025: Fluidstack secured 168 MW at Abernathy under a 25-year lease (fully pre-leased capacity) and a 50.1% stake in the Abernathy joint venture. Total Fluidstack capacity commitments now exceed 746 MW across three campuses—Lake Mariner, Akela, and Abernathy—with terms extending up to 25 years. Google's signed backstop guarantee for Fluidstack obligations (claims 4509, 4458) adds implicit validation of Fluidstack's credit quality.

The disclosed claims omit explicit revenue figures, preventing direct reconciliation of revenue guidance against announced capacity. TeraWulf has disclosed megawatt commitments (Fluidstack 746+ MW, Core42 60 MW, NYPA 90 MW power) but not per-megawatt lease rates or total contract value. The lack of quantified revenue per agreement makes it impossible to verify whether publicly stated revenues align with the announced deal architecture. This asymmetry—detailed capacity and term disclosure coupled with absent pricing—suggests TeraWulf may be optimizing for investor confidence in utilization rather than near-term revenue visibility.

TeraWulf's customer roster reveals a stark concentration around two named cloud infrastructure partners backed by signed, long-term capacity leases. Core42, a G42 subsidiary, committed to 60 MW of critical IT load at La Lupa in December 2024 (claims 3628, 3639), supplemented by a separate 70 MW provision for G42's US operations announced the prior month (claim 1051). The flagship relationship, however, is with Fluidstack, which has become the Company's dominant customer through a series of executed agreements spanning August 2025 through February 2026. Fluidstack's commitments total approximately 1,074 MW across multiple properties: 200+ MW under a 10-year AI hosting agreement at Lake Mariner (claim 1054); 160 MW at the Lake Mariner CB-5 expansion (claim 1053); 168 MW within the Flash Compute joint venture, which TeraWulf owns 50.1% of (claim 3966); 168 MW under a 25-year lease at the Abernathy HPC Campus (claim 3642); and 378 MW across three leases at Akela (claim 3630). All major customer relationships employ executed lease language ("entered into," "agreed to lease," "executed"), distinguishing them from pilot, framework, or discussion-stage arrangements. Google's August 2025 backstop guarantee for Fluidstack leases (claim 4509) signals third-party confidence but remains a contingent obligation rather than primary revenue generation. Supporting infrastructure agreements with the New York Power Authority (90 MW, executed February 2022, claim 3627) and Kentucky Power's ongoing 345 kV substation construction (claim 1310) enable rather than offset customer concentration. The company has not disclosed revenue per megawatt or reconciling revenue figures against these capacity commitments; the filings detail only MW allocations and term lengths. Counterparty concentration has shifted dramatically from nascent stage to near-total dependence on Fluidstack, which accounts for approximately 95% of disclosed committed capacity. Core42 and NYPA remain named secondary relationships, while several transactional partners—the Hawesville development entity seller (claim 3878), Heorot via transition services (claim 4834), and Beowulf as facilities operator (claim 5588)—appear tangential to core revenue generation. This extreme customer concentration, while indicative of successful pre-leasing strategy, concentrates financial risk materially on a single counterparty.

TeraWulf's customer base reflects a clear transition from infrastructure partnerships to signed data center leases, with substantial concentration in two counterparties. Beowulf (claim 5588) and Somerset (claim 5587) represent binding operational and governance relationships from October 2024, establishing platform foundation. However, material revenue derives from three named tenants, all formalized through signed agreements.

Core42, a G42 subsidiary, signed long-term data center lease agreements in December 2024 (claims 4993, 3628) committing 60 MW at La Lupa with an exercisable expansion option for an additional 135 MW through March 31, 2025 (claim 4994). The lease language is definitive—"entered into long-term data center lease agreements" signals binding commitment, yet TeraWulf initially characterized Core42 as a "partnership" (claim 5225), terminology that obscures the commercial relationship's binding nature.

Fluidstack represents the dominant named customer, with cumulative signed commitments totaling 546 MW across two vehicles. In August 2025, Akela executed three data center lease agreements with Fluidstack USA I Inc., obligating 378 MW of critical IT load (claims 3630, 3652). Critically, Google backstopped "certain obligations of Fluidstack" under these leases in the same period (claims 4509, 4458), suggesting TeraWulf required third-party credit enhancement. In October 2025, parties restructured into FS CS I LLC—a formal joint venture with TeraWulf holding 50.1% and Fluidstack holding 49.9% (claims 3970, 3966)—governing the Abernathy HPC Campus's 168 MW fully pre-leased to Fluidstack USA III under a 25-year lease (claim 3634).

Counterparty concentration risk is material. Fluidstack and Core42 together represent approximately 740+ MW of disclosed capacity, or roughly 92% of named customer load. The G42 infrastructure claim (claim 1051: "over 70 MW") lacks specificity—neither signed lease documentation nor revenue basis is clarified, suggesting possible LOI or framework status. Revenue arithmetic reconciliation is absent; TeraWulf discloses MW commitments and lease terms but not rate assumptions or revenue multiples, limiting ability to validate deal economics against customer concentration and duration risk.

TeraWulf's disclosed customer and partnership announcements demonstrate a high degree of concentration around two named counterparties, Fluidstack and Core42, each of whom signed contractually binding data center lease agreements rather than exploratory frameworks. Across twenty-four claims spanning October 2024 through May 2026, TeraWulf names both primary customers explicitly; no material revenue-generating counterparties are described as "strategic partners" or "anchor customers" without identification. The sole exception is claim 9064, which references unnamed "regional educational and workforce development institutions," but this initiative represents operational outreach rather than capacity commitments.

The company's core commercial relationships rest on binding signed leases, not pilots or memoranda of understanding. Core42, identified as a G42 subsidiary, signed long-term data center lease agreements for 60 MW at La Lupa in December 2024 (claims 3639, 3628) with a contractual expansion option for 135 MW exercisable through March 31, 2025 (claim 4994). Fluidstack engaged via multiple signed agreements totaling 706 MW across three campuses: 378 MW at Lake Mariner (claims 3630, 3640, 3652), 160 MW at CB-5 (claim 1053), and 168 MW at Abernathy under a 25-year lease (claims 3634, 3642). All agreements specify exact megawatt commitments and term lengths, indicating closed commercial transactions.

Counterparty concentration has become increasingly pronounced. By February 2026, Fluidstack represents approximately 92 percent of named capacity commitments (706 MW of 766 MW total), and approximately 81 percent even if Core42 exercises its full 135 MW expansion option. This dependency deepened markedly after August 2025, when Fluidstack agreements dominate company disclosures, whereas prior periods (October 2024 to July 2025) featured Core42 and the Somerset operational arrangement. The company has partially hedged concentration risk by securing a Google backstop guarantee for certain Fluidstack lease obligations (claims 4509, 4458), though Google functions as a contingent guarantor rather than an end customer.

Regarding revenue reconciliation: the disclosed claims contain no financial terms, pricing schedules, or revenue figures. They enumerate megawatt allocations and lease durations but provide no per-MW pricing or capacity utilization assumptions necessary to verify the arithmetic against disclosed revenue. This gap represents incomplete disclosure of commercial terms material to diligence.

TeraWulf's disclosed customer agreements are predominantly signed contracts rather than frameworks or pilot arrangements, though the company exhibits extreme concentration risk centered on Fluidstack. Somerset, Core42, and Heorot agreements from late 2024 through mid-2025 (claims 5586, 5587, 5588, 4993, 4834) are explicitly characterized as executed arrangements with defined term structures and operational obligations. The same precision applies to the New York Power Authority power supply agreement executed in February 2022 (claims 2072, 2114), providing 90 MW under a ten-year commitment. All named counterparties are specifically identified rather than obscured as 'strategic partners' or 'anchor customers.' However, the portfolio composition shifted dramatically beginning August 2025, when TeraWulf entered into three data center lease agreements with Fluidstack USA I Inc., committing 378 MW at the Akela facility (claims 3640, 3630, 2105). Subsequent filings reveal an additional 168 MW joint venture at Abernathy under amended governance (claims 3654, 3642, 3634), with Fluidstack CS I Inc. holding 49.9% equity and TeraWulf's Big Country Wulf LLC subsidiary holding 50.1% (claims 3966, 3970, 3956). This JV structure indicates deeper partnership integration than traditional landlord-tenant arrangements, yet no pilot or expansion-option language appears; agreements reference 25-year hosting terms (claim 3642). Core42, by contrast, represents only 60 MW committed capacity at La Lupa under December 2024 long-term leases (claims 3628, 3639, 3651). Fluidstack thus commands approximately 546 MW—roughly 90 percent of identifiable committed capacity—against 60 MW for Core42 and 90 MW for NYPA. The company does not disclose rental rates, occupancy assumptions, or revenue per MW, rendering the revenue arithmetic unverifiable against announced MW commitments. Google's role as backstop guarantor for Fluidstack obligations (claims 4509, 4458) signals credit concerns regarding the primary tenant and warrants scrutiny of Fluidstack's financial stability. TeraWulf's shift from multi-counterparty diversification to single-customer dependence on Fluidstack, coupled with an undisclosed but operationally material joint venture, represents material concentration that exceeds typical data center customer portfolios and introduces refinancing and ramp risk.

TeraWulf's disclosed customer base shows extreme concentration among two named counterparties under signed, long-term agreements. Core42, a G42-affiliated entity, entered into binding data center lease agreements on December 23, 2024 (claims 4993, 3628), committing to 60 MW of critical IT load at La Lupa with renewal optionality. Fluidstack, self-identified as a leading AI cloud platform, represents the dominant contractual footprint: 378 MW across three signed leases at Akela executed in August 2025 (claims 3630, 3652), plus 168 MW at Abernathy under a 25-year binding lease agreement entered in October 2025 (claims 3634, 3642). Fluidstack's aggregate contracted capacity totals 546 MW, representing approximately 90 percent of TeraWulf's disclosed signed customer commitments. Google, though unnamed as a counterparty to TeraWulf itself, became material through a backstop guarantee of Fluidstack's lease obligations (claims 4509, 4458), indicating underwriting concerns about Fluidstack's creditworthiness sufficient to require parent-company guarantees. Infrastructure counterparties New York Power Authority and Kentucky Power (AEP subsidiary) are named and operational: NYPA supplies 90 MW under a ten-year term executed in February 2022 (claims 3627, 2114), while Kentucky Power is actively constructing transmission infrastructure (claim 1310). Beowulf Electricity and Somerset remain tied to facility operations and governance rather than capacity lease revenue (claims 5588, 5587). The Hawesville development involved an unnamed seller receiving 6.8 percent minority equity with redemption rights (claim 3878). Counterparty concentration risk is acute: TeraWulf's revenue stream depends overwhelmingly on Fluidstack's performance and capital deployment across two JV structures (Flash Compute at 50.1 percent ownership, Abernathy at 50.1 percent). Claims disclose only megawatt commitments and lease structures; no revenue figures, pricing rates per MW, or ramp schedules are provided. Without pricing disclosure, the 606 MW identified capacity cannot be cross-checked against any stated revenue guidance, preventing reconciliation of announced deal volume to financial expectations. This asymmetry between operational disclosures and financial silence creates analytic opacity around true revenue contribution and timing.

TeraWulf has disclosed binding agreements with three material revenue-generating counterparties—Core42, Fluidstack, and the New York Power Authority—each structured as signed leases rather than letters of intent, memoranda of understanding, or pilot arrangements. All counterparties are explicitly named, with no reliance on opaque "strategic partner" designations.

Core42, a G42 subsidiary, entered signed long-term data center leases in December 2024 committing 60 MW at La Lupa (claim 4993). The initial G42 announcement (claim 1051, December 3, 2024) referenced "over 70 MW of Data Center Infrastructure" without facility specification; subsequent disclosures clarify Core42 specifically agreed to 60 MW. Core42 also received an expansion option for 135 MW exercisable through March 31, 2025 (claim 4994), but since this filing post-dates the deadline, the option's exercise status remains undisclosed, leaving Core42's potential full commitment uncertain.

Fluidstack dominates the disclosed customer base. Three signed lease agreements executed in August 2025 commit 378 MW at Lake Mariner (claim 3630). An additional 168 MW capacity at Abernathy is fully pre-leased under a 25-year agreement (claim 3634) structured as a joint venture where TeraWulf retains 50.1% ownership (claim 3641). Combined, Fluidstack represents 546 MW of signed capacity—approximately 90 percent of TeraWulf's disclosed data center customer base. Google has agreed to backstop certain Fluidstack lease obligations (claim 4509), adding institutional credibility but signaling potential counterparty credit concerns.

The New York Power Authority supplies 90 MW of power under an agreement executed in February 2022 with a ten-year term (claim 3627).

Customer concentration is acute: Fluidstack accounts for 546 MW signed, Core42 for 60 MW confirmed. No pricing, contract value, or revenue per megawatt is disclosed across any claim. Without economic terms, the MW announcements cannot be validated against stated financial guidance, and the true revenue quality—whether these represent premium, long-duration streams or discounted arrangements requiring guarantor support—remains opaque. The customer base reflects evolution from sovereign-cloud focus (G42/Core42, December 2024) to AI-compute dominance (Fluidstack, August-October 2025), with concentration risk materially elevated.

TeraWulf's announced customer base presents a concentrated partnership profile that requires careful parsing between signed lease commitments and partially-owned joint ventures. The company has disclosed specific, named counterparties—Fluidstack, Core42, and Google—rather than generic "strategic partners," but the contractual structures and ownership arrangements raise questions about revenue quality and customer independence. Fluidstack dominates the disclosed capacity commitments. TeraWulf signed three data center lease agreements with Fluidstack USA I in August 2025 covering 378 MW at Lake Mariner (claims 3630, 3652, 3640), while the Abernathy HPC Campus delivers 168 MW under a 25-year lease to Fluidstack USA III (claims 3642, 3634). These are presented as signed leases with language like "entered into" (claim 3630). However, by December 2025 and February 2026, the relationship evolved fundamentally: TeraWulf and Fluidstack formed joint ventures, with TeraWulf holding 50.1% ownership and Fluidstack holding 49.9% (claims 3966, 3970, 3956, 3633, 3654). This structure converts what appear to be customer lease relationships into partially-owned operating entities, materially reducing true third-party revenue concentration. Core42, a G42 affiliate, signed long-term lease agreements in December 2024 for 60 MW at La Lupa (claims 3639, 3628, 3651) plus the earlier 70 MW data center infrastructure commitment (claim 1051). These leases include expansion options exercisable through March 31, 2025 (claim 4994), indicating conditional capacity rather than fully committed deployments. Google's role warrants scrutiny. Rather than acting as a direct customer, Google agreed to backstop certain Fluidstack lease obligations (claims 4509, 4458)—a credit enhancement that suggests TeraWulf viewed Fluidstack's standalone creditworthiness as insufficient. This is a guarantee structure, not a customer relationship. The disclosed revenue arithmetic does not reconcile cleanly with capacity announcements. TeraWulf has not disclosed pricing per MW or total contracted revenue. With 546 MW committed to Fluidstack (now half-owned) and 60 MW to Core42, customer concentration is acute. The transition of Fluidstack from counterparty to co-owned venture partner eliminates pure third-party customer diversification. Revenue guidance or customer concentration metrics would clarify whether these signed agreements translate to meaningful incremental profitability or primarily represent co-owned operating entity buildouts.

TeraWulf's disclosed customer agreements are predominantly signed contracts with named counterparties rather than framework arrangements or letters of intent. The customer base concentrates heavily on two operators: Fluidstack and Core42. Fluidstack appears across at least seven distinct signed agreements totaling 546 MW: a 10-year, 200+ MW hosting agreement (claim 1054), three separate lease agreements at Lake Mariner committing 378 MW (claims 3630, 3640, 3652), and an October 2025 amended joint venture agreement for Abernathy's 168 MW campus with 50.1% TeraWulf ownership (claim 3654). Core42, a G42 subsidiary, signed long-term data center lease agreements in December 2024 to receive 60 MW of critical IT load at La Lupa (claim 3628), with an expansion option for 135 MW through March 31, 2025 (claim 4994). All major capacity agreements cite specific execution dates and legal structures indicating completion rather than preliminary intent. Google appears not as a customer but as a backstop guarantor for Fluidstack lease obligations (claim 4509), a credit instrument rather than a demand agreement. Somerset occupies a governance role through board observer rights and operational continuity with affiliate Beowulf (claims 5587, 5588) rather than functioning as a capacity customer.

The concentration around Fluidstack and Core42 has intensified substantially. These two counterparties command approximately 740 MW of committed or optioned capacity—the overwhelming majority of disclosed arrangements. The New York Power Authority supplies 90 MW to Lake Mariner under a February 2022 agreement (claim 3627), operating as an upstream infrastructure provider rather than a tenant.

The revenue implications of these agreements remain opaque. While TeraWulf has disclosed approximately 740 MW under contract, the claims provide no per-MW pricing, utilization curves, escalation clauses, or ramp profiles necessary to reconcile with reported revenues. Lease structures range from ten to twenty-five years, and the Fluidstack joint venture introduces equity economics distinct from capacity leasing, suggesting revenue models that cannot be reverse-engineered from capacity announcements. Without customer concentration metrics, payment schedules, or churn assumptions, the reconciliation from signed megawatts to stated financial results remains incomplete.

TeraWulf's announced customer and partnership agreements show a clear distinction between definitive signed deals and earlier vague framing, with material concentration risk emerging by 2025. All named counterparties—Beowulf, Somerset, Core42, Fluidstack, Google, and the New York Power Authority—are explicitly identified rather than cloaked as 'strategic partners' or 'anchor customers,' but the degree of commitment certainty varies significantly. The October 2024 claims regarding Beowulf (5588) and Somerset (5587) establish operational and governance control rather than customer revenue relationships. Core42 introduced real capacity commitments in late 2024, though claim 5225 initially described the relationship as a 'partnership' before subsequent filings clarified a signed long-term lease agreement for 60 MW at La Lupa (claims 3628, 3639, 3651). A discrepancy surfaces in claim 1051 (December 3, 2024), which references 'over 70 MW of Data Center Infrastructure for G42's US Operations' with no executed agreement language, whereas the Core42 relationship subsequently disclosed is only 60 MW—a 10 MW gap suggesting either early overstatement or scope clarification. Fluidstack emerged as the dominant customer starting August 2025 with three signed data center lease agreements totaling 378 MW at Akela (claims 3640, 3652) and a 50.1% joint venture controlling 168 MW at Abernathy with a 25-year lease to Fluidstack USA III (claims 3641, 3642, 3654). By October 2025, Fluidstack represented roughly 546 MW of committed capacity—nearly ten times the Core42 commitment. Google's backstop guarantee (claims 4509, 4458) creates indirect credit exposure rather than direct customer revenue. NYPA's 90 MW power supply agreement (claims 3627, 2114, 2072) is infrastructure rather than a data center customer. This concentration structure—with Fluidstack representing approximately 75 percent of named customer capacity and both major customers linked to AI infrastructure demand—creates material counterparty risk. The revenue arithmetic cannot be verified from available claims, as no explicit per-megawatt pricing or total revenue reconciliation appears in the disclosed announcements. The Heorot transition services agreement (claim 4834) is operationally significant but time-limited to two years post-acquisition, unlike the long-duration Fluidstack and Core42 commitments.

TeraWulf's disclosed customer base comprises entirely named, binding agreements rather than MOUs or framework partnerships. All major counterparties are specifically identified: Fluidstack USA I Inc. and its subsidiaries, Core42 Holding US LLC (G42), the New York Power Authority, and operational partners Somerset/Beowulf. No generic "strategic partner" or unnamed "anchor customer" language appears in the record.

The company has executed signed data center lease agreements and joint venture structures. Core42 signed long-term agreements in December 2024 for 60 MW at La Lupa (claims 3628, 4993). Fluidstack entered binding lease agreements beginning August 2025: three agreements for 378 MW at Lake Mariner's Akela campus (claim 3630), a 25-year lease for 168 MW at the Abernathy HPC Campus (claim 3634), and an amended LLC agreement establishing a joint venture with Fluidstack CS I Inc. at 50.1% TeraWulf ownership (claim 3633). The New York Power Authority commitment dates to February 2022 for 90 MW under a ten-year agreement (claim 3627). Somerset/Beowulf's role as exclusive operator and board observer represents an operational arrangement rather than a customer revenue agreement (claims 5587, 5588).

Counterparty concentration has evolved dramatically. Core42 was the named customer through early 2025, representing TeraWulf's expansion into sovereign cloud infrastructure. The August 2025 Fluidstack agreements marked a fundamental shift: Fluidstack now anchors 546 MW across committed leases and the Abernathy joint venture, dwarfing Core42's 60 MW allocation. This 9:1 capacity ratio creates significant customer concentration risk. The Google backstop guarantee for Fluidstack lease obligations (claim 4509) acknowledges this dependency—Google has agreed to cover Fluidstack's termination fees and rent obligations, effectively mitigating credit risk but underscoring Fluidstack's outsized importance.

The disclosed record contains no revenue figures to reconcile against the announced MW commitments. While lease agreements specify capacity (378 MW, 168 MW, 60 MW, 90 MW) and terms (25-year for Abernathy, 10-year for NYPA, long-term for Core42), actual dollar revenues or utilization rates are absent from these claims. Without revenue disclosures, it remains impossible to verify whether the company's total addressable capacity translates to announced earnings guidance or top-line growth projections.

TeraWulf's customer base is characterized by explicit naming and signed lease agreements rather than preliminary discussions or frameworks. Core42, a G42 subsidiary, represents the earliest material commitment with 60 MW of long-term data center capacity signed December 23, 2024 (claim 4993), while Fluidstack emerged as the dominant counterparty through a series of escalating lease agreements beginning in August 2025. The Fluidstack relationship includes three data center lease agreements totaling 378 MW at the Lake Mariner campus (Akela facility) with initial execution in August 2025 (claims 4508, 4457), a 168 MW 25-year lease at the Abernathy HPC Campus fully pre-leased to Fluidstack USA III (claim 3642), and a 50.1% joint venture in Flash Compute formalized in October 2025 (claims 3970, 3966). Google provided a backstop guarantee against Fluidstack payment defaults rather than serving as a direct customer (claims 4509, 4458), reflecting Fluidstack's criticality to TeraWulf's capacity deployment strategy.

Concentration risk is material. Fluidstack commitments represent approximately 546 MW of the disclosed 606 MW in signed customer capacity, or roughly 90 percent of TeraWulf's identified pre-leased infrastructure. Core42 comprises the remaining announced customer commitment. Notably absent from the claims is any reference to preliminary negotiations, letters of intent, or pilot programs with major hyperscalers or alternative cloud platforms. All agreements employ definitive language: "entered into long-term data center lease agreements," specific megawatt allocations, and defined term lengths (10-year for Core42, 25-year for Fluidstack's Abernathy lease). Somerset Electricity retained operational control through Beowulf Electricity as exclusive operator, secured by board observer status and registration rights (claims 5587, 5588).

Revenue reconciliation cannot be assessed from these claims. While capacity commitments are fully enumerated across named customers, the data set contains no disclosed pricing, annual revenue expectations, or committed payment terms from any counterparty. This gap makes it impossible to verify whether revenue guidance aligns with announced megawatt commitments or to assess relative profitability of individual customer relationships. The concentration on a single cloud-platform customer and absence of diversification among alternative hyperscalers introduces material operational and revenue risk not apparent from capacity figures alone.

TeraWulf has disclosed exclusively signed customer agreements with fully identified counterparties, avoiding vague partnership language. The company's major revenue-generating commitments center on two named customers: Fluidstack and Core42, with no use of placeholder descriptions like 'strategic partners' or 'anchor customers without identification.' Core42, a G42 subsidiary, signed long-term data center lease agreements in December 2024 committing to 60 MW at La Lupa (claims 3628, 3651), formalized in May 2025 filings (claim 4993). Fluidstack emerged as the dominant counterparty beginning August 2025, securing 378 MW across three data center lease agreements at Lake Mariner (claims 3630, 3640, 3652) under a 10-year term, with an additional 168 MW at Abernathy under a 25-year lease through a 50.1%-owned joint venture (claims 3634, 3642, 3654). Combined, Fluidstack controls approximately 546 MW of TeraWulf's named capacity, representing roughly 90% of all disclosed customer commitments. The concentration has tightened considerably: early claims cited G42 broadly (claim 1051, 70 MW in October 2024), which resolved into Core42's specific 60 MW commitment by year-end. Google's August 2025 backstop guarantee on Fluidstack's lease obligations (claims 4509, 4458) materially mitigates counterparty risk by anchoring payment certainty to a creditworthy third party. No other named customers appear in the disclosure record; NYPA represents critical power infrastructure rather than revenue customers (claims 3627, 2114), and Somerset/Beowulf are landlord-operator relationships (claims 5587, 5588) rather than tenant customers. The claims contain no revenue figures to reconcile against capacity commitments—only megawatt allocations, lease terms, and ownership structures. This absence prevents verification of disclosed revenue against signed capacity, though the signed nature of all material agreements and the explicitness of named parties (Fluidstack USA I, Fluidstack USA III, Core42 Holding US LLC) minimize interpretive ambiguity about binding status. The risk profile shows asymmetric customer concentration mitigated partially by Google's guarantee and the joint venture structure, which aligns incentives through equity ownership.

TeraWulf's disclosed partnerships reveal a clear trajectory from energy-infrastructure alignment to major customer concentration in cloud compute. All named counterparties—Somerset, Core42, Fluidstack, and the New York Power Authority—are identified by their legal entity names, not vague descriptors; no 'strategic partners' or unnamed 'anchor customers' appear in the record.

The Somerset relationship (claims 5587–5588, 2024-10-09) operates at the operational level: Beowulf, a Somerset affiliate, maintains exclusive operator status for TeraWulf's facilities, and Somerset holds board observer rights. This is a landlord-tenant-plus-governance arrangement, distinct from customer revenue contracts.

Core42, a G42 subsidiary, was announced in December 2024 with 70 MW of Data Center Infrastructure for G42's US Operations (claim 1051). Subsequent filings specify 60 MW at La Lupa under long-term lease agreements dated December 23, 2024 (claims 4993, 3628), plus an expansion option for an additional 135 MW exercisable through March 31, 2025 (claim 4994). No revenue figures are disclosed, and the option's exercise status is not clarified in the record.

Fluidstack emergence dominates the latter period. Announced August 13, 2025, the relationship spans three distinct structures: (1) a 200+ MW, 10-year hosting agreement at Lake Mariner and Akela (claims 1054, 3652—378 MW total); (2) a 168 MW pre-leased Abernathy HPC Campus under a 25-year lease to Fluidstack USA III (claim 3642); and (3) a 50.1% minority venture stake in Flash Compute, with Fluidstack holding 49.9% (claims 3966, 3970). Google's backstop guarantee (claims 4509, 4458) for Fluidstack lease obligations suggests credit reinforcement rather than Google as a customer. By February 2026 (claims 3630–3641), Fluidstack contracts dominate TeraWulf's capacity roadmap: 914+ MW across multiple entities and tenures.

Concentration risk is material. Fluidstack accounts for six figures' worth of announced megawatt-years (914+ MW across 10–25 year terms), dwarfing Core42's 60 MW and the NYPA power allocation (90 MW). Revenue arithmetic is absent from public filings; without pricing-per-MW or utilization assumptions, the announcements cannot be reconciled to reported revenues. The record does not disclose Fluidstack's financial stability, default history, or the true economic value of the 49.9% equity interest TeraWulf granted.

TeraWulf's customer portfolio has consolidated dramatically around two named counterparties under signed agreements: Fluidstack and Core42, representing a substantial shift from the Somerset-anchored infrastructure relationship disclosed in 2024 (claims 5587–5588). The largest commitment, spanning multiple signed lease agreements, is Fluidstack: 378 MW at Akela (Lake Mariner) under a long-term lease (claims 3630, 3652), 168 MW at Abernathy under a 25-year lease (claim 3634), and earlier commitments totaling 200+ MW over ten years (claim 1054), collectively positioning Fluidstack for roughly 546 MW of the company's disclosed capacity. Core42, identified as a G42 subsidiary specializing in sovereign cloud, holds 60 MW at La Lupa with an expansion option to 135 MW exercisable through March 2025 (claims 4993–4994, 3651). Both relationships are documented as signed data center lease agreements dated in late 2024 and executed through 2025 (claims 4993, 3630), not frameworks or pilots.

Concentration risk is acute. Fluidstack alone represents approximately 78% of named customer capacity commitments; adding Core42's 60 MW leaves only modest diversification. The Somerset relationship, by contrast, appears to be infrastructure and governance scaffolding—board observer rights and an exclusive operator arrangement (claims 5588)—rather than a customer contract with disclosed capacity or revenue terms.

One credit enhancement is noteworthy: Google has agreed to backstop Fluidstack's obligations under its leases (claims 4509, 4458), mitigating counterparty risk on TeraWulf's largest customer. This arrangement suggests institutional confidence in the Fluidstack relationship despite Fluidstack's relative youth as a company.

The disclosed revenue arithmetic does not reconcile. Across the Fluidstack and Core42 leases, TeraWulf has committed to deliver 606 MW of capacity under long-term contracts, yet no specific revenue figures, pricing per megawatt, or aggregate expected revenue are disclosed in these claims. For a company with material capex obligations tied to build-out (claims 1310, 2072), the absence of revenue guidance tied to these named contracts is conspicuous. The only non-customer revenue-bearing relationship mentioned is NYPA's 90 MW power supply agreement (claims 3627, 2114), which appears to be infrastructure cost recovery rather than customer revenue.

TeraWulf's customer roster presents a clean distinction between executed contracts and formal arrangements, with no discernible MOU or framework-stage partnerships cited in the disclosed record. All named counterparties—Beowulf, Somerset, Core42, Fluidstack, and the New York Power Authority—appear in signed agreements with defined operational or financial terms rather than preliminary discussions. The verbatim citations employ language of execution ('entered into,' 'agreed to,' 'executed') rather than exploratory or conditional phrasing, suggesting these are binding commitments without caveat (claims 5588, 4993, 4508, 3627).

Concentration risk, however, is severe. Fluidstack dominates TeraWulf's customer base, with 746+ MW of named capacity across three distinct agreements: the 378 MW Akela lease (Aug 2025, claims 3630, 3640, 3652), the 168 MW Abernathy HPC Campus (Oct 2025, claims 3641, 3633, 3654), and an earlier 200+ MW AI hosting commitment (claim 1054). Core42, the second-largest named customer, holds only 60 MW of committed capacity (Dec 2024, claims 3628, 3639, 3651). This 12-to-1 ratio underscores near-complete dependency on a single customer for near-term revenue generation.

The portfolio's evolution traces a strategic pivot. Somerset and Beowulf anchor 2024 claims (claims 5588, 5587) as infrastructure or operational partners rather than customers. Core42 entered in December 2024 (claim 4993), representing TeraWulf's first named AI-infrastructure customer. Fluidstack's arrival in August 2025 (claim 4508) marks the inflection point, with subsequent agreements in October and December 2025 (claims 3641, 3970) embedding the relationship into joint-venture structures—50.1% owned by TeraWulf subsidiaries—signaling deeper alignment or capital commitment than pure colocation.

Revenue reconciliation cannot be assessed from the disclosed record. The claims provide capacity (MW) and lease terms (10-year, 25-year) but no pricing, contract value, or implied revenue figures. Without rate cards or financial footnotes tying MW capacity to announced earnings, the arithmetic remains opaque. A 746 MW Fluidstack portfolio and 60 MW Core42 portfolio generate no quantified revenue claim, and no prior-period revenue validates whether these agreements are performing as disclosed. This gap between dramatic capacity announcements and silent financial reality invites scrutiny of whether these represent revenue-accretive contracts or arrangements struck at distressed pricing to secure utilization.

TeraWulf's customer announcements overwhelmingly reference signed data center lease agreements rather than MOUs or framework partnerships, with all major counterparties explicitly named. Beowulf Electricity & Data Inc. appears in 2024 as the exclusive operator under an operative agreement (claim 5588), alongside Somerset's board observer rights (claim 5587), both of which are governance instruments rather than customer contracts. Core42, a G42 subsidiary, emerged as the first customer with a long-term lease signed in December 2024 committing TeraWulf to deliver 60 MW (claims 3628, 3639, 3651), substantiated across multiple subsequent filings. Beginning August 2025, Fluidstack entered into signed leases that grew to dominate TeraWulf's customer base, initially via 200+ MW across a 10-year hosting agreement (claim 1054), followed by a 378 MW three-lease arrangement at Lake Mariner (claims 3630, 3640, 3652) and a 50.1%-owned joint venture at Abernathy for 168 MW pre-leased under a 25-year term (claims 3641, 3654). Google's backstop guarantee of Fluidstack's lease obligations (claims 4509, 4458) signals credit risk mitigation rather than a direct customer relationship. By February 2026, Fluidstack represented approximately 90 percent of disclosed customer megawattage (546 MW of ~606 MW total, including Core42's 60 MW and New York Power Authority's 90 MW power supply allocation). The NYPA agreement, executed February 2022 (claim 3627), functions as infrastructure support rather than a revenue-generating customer. Counterparty concentration has shifted from an operator-and-investor model (Somerset/Beowulf) to near-total dependency on Fluidstack, with Core42 remaining an early adopter but now marginal in scale. Revenue reconciliation raises material questions: TeraWulf announces megawattage commitments across facilities described as under construction or designed for future capacity, yet none of the cited claims quantify actual deployed capacity, revenue per MW, or historical revenue recognized. Language such as 'committed to deliver' and 'designed for' (claims 3634, 3642) references forward promises, not operating performance. The proliferation of identical claims across April–May 2026 filings (claims 2076–2086 repeating content from February 2026 claims 3627–3654) suggests regulatory re-disclosure of the same underlying leases rather than new customer wins, implying limited material customer growth in recent months.

TeraWulf's disclosed partnership activity remains heavily concentrated in a single named counterparty. The July 9, 2026 announcement of an Anthropic engagement (claim 1) represents a binding lease agreement rather than exploratory framework language; the terminology 'secures 20-year lease agreement' signals a committed, operational relationship with defined term structure, not a memorandum of understanding or pilot arrangement. Anthropic is explicitly identified, eliminating ambiguity around counterparty identity that often obscures early-stage partnership claims. The announced deal structure—a long-duration lease spanning two decades—suggests material capital commitment and sustained revenue visibility, though the disclosure lacks specificity on lease economics, data center capacity, or pricing terms needed to assess revenue contribution or reconcile announced commitments against reported financials. With only one customer announcement visible in the record, counterparty concentration appears acute; absent disclosure of other named customers or partnerships, TeraWulf's near-term revenue dependency on the Anthropic relationship is difficult to assess independently. The absence of quantified revenue figures, capacity commitments, or pricing disclosed alongside the lease announcement limits ability to validate whether publicly stated backlog or revenue guidance aligns with the economic substance of announced deals. For a company in high-capital infrastructure services, the gap between signing a long-term lease and achieving cash realization is material—terms, payment structures, and capacity ramp schedules remain opaque. Investors should inquire whether the Anthropic lease represents the sole customer commitment in the current backlog or whether additional named customers exist but remain undisclosed. The long-dated nature of the commitment reduces near-term revenue visibility compared to shorter operational contracts, potentially explaining conservatism in near-term guidance despite material announced commitments.

TeraWulf's capacity story has expanded and narrowed at the same time. The original headline target of 800 MW deployed by 2025 (claims 21125, 21141, 14108) was a bitcoin-mining figure, and the record now grades it delayed. The supporting components shrank on the way. Nautilus was first presented as a 300 MW nuclear-powered facility (14121, 14130), with TeraWulf's share framed as 150 MW (14112). By August 2022 it had been restated as 200 MW initial capacity plus a 100 MW expansion (13715, 13716), with TeraWulf holding a one-third equity stake (13714). The expansion is graded delayed, and the project is now listed as sold. Lake Mariner moved the other way. It started from a 90 MW NYPA supply (14110, 14126) and a 50 MW first building (13881), and now appears as a 410 MW initial and 750 MW campus target tied to an Anthropic lease and a $3.5 billion financing goal. That is a pivot in business model, not a continuation of the original plan.

The dated promises have a mixed record. The New York Q4 2021 start (21138), the Lake Mariner first phase by year-end 2022 (14106) and the Nautilus ASIC installation (14122, 14024) are graded verified. The 50 MW online in 2021 (21124, 21140) never became verifiable. The 58,000-miner delivery commitment was repeated across at least five filings (14247, 14049, 14025, 13978, 13897) without restatement and ended delayed. It also sits awkwardly beside a 78,000-unit procurement figure (14142). Claim 13880, published June 30, 2022, promised 3,300 miners by July 1, a one-day window that reads more like a status report than a forecast.

The supplied record contains no independent corroboration: no permits, no NYISO or PJM queue entries, and no site reporting. The verified grades reflect consistency with later company disclosures, not third-party evidence. The claim log also stops in early 2023. The entire HPC pivot, the Anthropic lease and the Kentucky and Maryland sites therefore rest on the project roster, not on dated, auditable claims.

Over the next 90 days, the key test is whether the $3.5 billion Lake Mariner financing closes on the stated terms. The second is whether NYISO interconnection filings support the step from 410 MW to 750 MW. The third is whether any of the announced Kentucky and Maryland sites, including Morgantown, Hawesville and Boyd-Greenup, produce zoning, air or utility filings that would move them beyond press-release status.

TeraWulf's capacity story has been rewritten more than revised. The original headline, 800 MW deployed by 2025, appears unchanged across claims 21125, 21141 and 14108. It is now marked delayed, and none of the claims on file walk it back or restate it. The near-term building blocks performed better. The New York facility came online in Q4 2021 (21138), and the 100 MW first phase at Lake Mariner landed by year-end 2022 (14106). The 50 MW 2021 figure (21124, 21140) cannot be confirmed from this record. Equipment shows a softer pattern. The 58,000-miner delivery commitment was repeated in at least five filings (14247, 14049, 14025, 13978, 13897) and is now delayed, while a parallel 78,000-unit procurement figure (14142) was never reconciled with it.

Nautilus is the clearest case of quiet shrinkage. The site was pitched at 300 MW (14121, 14130, 14111) with TeraWulf holding 150 MW through the Talen joint venture (14112). It was then restated as 200 MW of initial capacity (13715) with a one-third equity stake (13714). The 100 MW expansion (13716) slipped, and the project is now listed as sold, as is Abernathy. The portfolio has rotated away from self-built bitcoin capacity. It now leans on Lake Mariner, where the Anthropic-leased facility's disclosed size moved from 410 MW to a 750 MW campus target alongside a $3.5 billion financing goal. The claim record supplied here ends in early 2023, so those later figures cannot be matched to dated claim ids or checked for acknowledged revisions.

The file contains no independent corroboration: no permits, no NYISO or PJM queue positions, and no site reporting. That is the central gap. The pre-2023 record shows TeraWulf reliably hits single-building milestones but restates portfolio-scale targets without flagging the change. The 750 MW Lake Mariner figure, and the long list of announced Kentucky and Maryland sites (Justified, Muskie, Hancock, Boyd-Greenup, Chesapeake, Morgantown), all currently rest on company statements alone.

Over the next 90 days, watch for three things. The first is whether the $3.5 billion financing closes on stated terms. The second is energization dates for incremental Lake Mariner tranches beyond the partially operational base. The third is whether any announced site appears in an interconnection queue or a county permitting docket. A Kentucky or Maryland site with no queue entry by year-end would suggest those announcements are optionality rather than pipeline.

The capacity story at TeraWulf has inflated in headline terms while shrinking at the project level. The 2021 merger-era pitch was 800 MW deployed by 2025 (21125, 21141, reaffirmed in 14108). The record flags 21125 as delayed, and that target was never met as a mining footprint. Nautilus was first described as a 300 MW nuclear-powered site (14130, 14111, 14121). Within five months it was reframed as 200 MW of initial capacity with a 100 MW expansion option (13715, 13716). The expansion is marked delayed, the site now sits in the sold column, and the 300-to-200 reset was presented as phasing rather than as a reduction. Lake Mariner grew more modestly, from a 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881) to a verified 100 MW first phase (14106). The project record now attaches 410 MW and then 750 MW figures to that campus under the Anthropic HPC banner. That is a change of business model, not an extension of the mining plan.

On dates, the near-term deliverables held. New York was operational in Q4 2021 (21138), and ASIC installation in Pennsylvania began on schedule (14122, 14024). Equipment was the weak point. The 58,000-miner delivery commitment is marked delayed across five separate restatements (14247, 14049, 14025, 13978, 13897). A parallel 78,000-unit procurement figure (14142) was never reconciled with it.

The independent evidence is thin. Almost every capacity claim here is classed as stated rather than verified. Nothing in this record ties the Kentucky and Maryland sites (Justified, Muskie, Chesapeake, Hancock County, Boyd-Greenup, Morgantown Generating Station) to permits, PJM or NYISO interconnection queue positions, or on-the-ground reporting. The claim list provided also cuts off in early 2023, so the 2023 to 2026 statements behind the 410 MW and 750 MW Lake Mariner figures cannot be checked by id here.

Over the next 90 days, the Q3 2026 report is the first test: it should show energized and leased megawatts at Lake Mariner measured against the 410 MW initial disclosure. Watch also whether the $3.5 billion financing referenced in July closes on its stated terms, and whether NYISO filings support a 750 MW campus. For the announced Kentucky and Maryland sites, any county permit, utility service agreement, or queue entry would be the first hard evidence that they exist beyond press releases.

The record supplied here ends abruptly in early 2023, so this read covers the bitcoin-mining era in detail and the AI-hosting pivot only through the project list. The capacity story falls into two phases. The first phase was an 800 MW target by 2025 (21125, 21141, 14108), which has been marked delayed. The second is a campus-level ambition at Lake Mariner that grew from a 410 MW initial disclosure to a 750 MW target, with a $3.5 billion financing headline in July 2026. The headline number has not shrunk so much as been relabeled. Mining megawatts became HPC megawatts, and the 2025 deadline passed without a clean accounting of how far short the original goal fell.

On dates, the record is mixed. Near-term site milestones generally held: the Q4 2021 New York start (21138), the Pennsylvania ASIC installation by mid-2022 (14122, 14024), and the 100 MW first phase at Lake Mariner by year-end 2022 (14106) are all verified. The larger promises slipped. The 58,000-miner delivery commitment is delayed across five restatements (14247, 14049, 14025, 13978, 13897). A same-day filing cited 78,000 units (14142) against 58,000 (14123) without reconciling the two figures. At Nautilus, capacity was described as 300 MW (14121, 14130) and later as 200 MW initial plus a 100 MW expansion (13715, 13716). The expansion is delayed, and TeraWulf's framing shifted from a 150 MW joint-venture share (14112) to a 33.33 percent stake (13714) without any stated explanation. Nautilus and Abernathy now show as sold, so two flagship capacity sources have left the portfolio.

Independent corroboration is the weakest part of the file. None of the claims here cites permits, NYISO or PJM queue positions, utility tariff filings, or site reporting. The verified tags rest on company disclosures. The new Kentucky and Maryland sites (Justified, Muskie, Hancock, Boyd-Greenup, Chesapeake/Morgantown) carry no stated capacity or cost, so their announced pace cannot be tested at all.

The next 90 days should cover the Q3 2026 earnings cycle. The first thing to look for is whether the $3.5 billion Anthropic-facility financing has closed and on what terms. The second is energized megawatts at Lake Mariner against the 410 MW tranche, as opposed to contracted megawatts. The third is any interconnection or state-commission filings that would give the Kentucky and Maryland sites real load numbers. A complete claim history after 2023 is also needed to judge whether the 750 MW target has its own dated milestones.

The record for TeraWulf shows capacity targets rising at the headline level and shrinking at the asset level. The original merger-era pledge of 800 MW deployed by 2025 (21125, 21141, 14108) was repeated without revision through 2022. It is now marked delayed (21125) against a deadline that has passed. Underneath it, the building blocks were smaller than they first appeared. Nautilus was presented as 300 MW of nuclear-adjacent capacity (14111, 14121, 14130). By August 2022 that had become a 200 MW initial build (13715) with a 100 MW expansion (13716) that later slipped. That asset is now listed as sold. Lake Mariner was framed as a 100 MW first phase (14106) resting on a 90 MW NYPA agreement (14110, 14126) and a 50 MW first building (13881). The current campus framing of 410 MW, rising to a 750 MW target with a $3.5 billion financing headline, is a different business: HPC leasing rather than self-mining. It should not be read as continuity with the 2021 bitcoin plan.

On dates, the near-term operational promises largely held. The New York facility went live in Q4 2021 (21138), Lake Mariner's first phase landed in 2022 (14106), and Pennsylvania ASIC installation began on schedule (14024, 14122). The equipment pipeline is where slippage concentrated. The 58,000-miner delivery commitment is marked delayed across five separate restatements (14247, 14049, 14025, 13978, 13897), and a 78,000-unit procurement figure (14142) sits alongside it unreconciled. The 50 MW-online-in-2021 claim (21124, 21140) was never independently confirmed. The 100 percent zero-carbon target for 2028 (14245, 14026, 13846, 13826, 13439) has been restated verbatim with no interim path disclosed beyond a 90 percent figure (13847).

The supplied record contains no permit filings, interconnection-queue entries, or site reporting, and it is truncated after early 2023. Pace claims for the Anthropic-leased campus and the Kentucky and Maryland sites therefore rest on company statements alone. Most of those newer projects, including Morgantown Generating Station, Hancock County, and Boyd-Greenup, carry no disclosed capacity or cost.

Over the next 90 days, three things matter. The first is whether Q3 results separate energized, revenue-generating MW at Lake Mariner from the 410/750 MW envelope. The second is whether NYISO queue or NYPA allocation filings substantiate load growth beyond the original 90 MW contracted base. The third is whether the announced Kentucky and Maryland sites acquire interconnection positions or permits, or remain capacity-less placeholders.

TeraWulf's capacity story has been re-scoped several times. The anchor figure, 800 MW deployed by 2025, was repeated across the 2021 merger materials and the first 2022 filings (21125, 21141, 14108). It is now marked delayed (21125), and the company's later framing replaced it rather than reconciled it. Nautilus shows the clearest shrinkage. It began as a 300 MW nuclear-powered site (14130, 14121, 14111). It then became a 200 MW initial build with a contingent 100 MW expansion (13715, 13716), and TeraWulf's stake fell to a third (13714). The site is now recorded as sold. Lake Mariner moved the other way, from a 90 MW NYPA allocation (14110, 14126) to the 410 MW and then 750 MW campus now tied to Anthropic. None of the claims in this record tie those later figures to the original targets.

On dates, the record is mixed. Near-term facility milestones mostly held. The New York facility came online in Q4 2021 (21138), the Pennsylvania ASIC installation started on schedule (14122, 14024), and the first 100 MW phase at Lake Mariner landed by year-end 2022 (14106). The softer promises fared worse. The 50 MW online in 2021 (21124, 21140) and the 3,300-miner Q3 2022 target (13880) were never confirmed. The 58,000-miner delivery commitment was restated at least five times before being marked delayed (14247, 14049, 14025, 13978, 13897). The 100 percent zero-carbon target for 2028 is repeated in nearly every filing (14245, 13826, 13439) but is never paired with an interim metric beyond the 90 percent figure (13847).

Independent corroboration is thin. The claims supplied here run only through early 2023. They contain no permit filings, NYISO, PJM, or Kentucky utility interconnection entries, and no site reporting for the newer announced projects in Kentucky and Maryland. That matters because those sites carry no stated capacity or cost.

In the next 90 days, watch whether the $3.5 billion Lake Mariner financing closes on the reported terms, and whether TeraWulf states what share of the 750 MW is energized versus contracted. Also watch for the first queue positions or utility agreements at Morgantown, Hancock County, Boyd-Greenup, and Hawesville. Those filings would be the first evidence that this pipeline is more than a list of sites.

The record supplied here effectively ends in early 2023, so this read rests on the bitcoin-era claims. The HPC-era targets attached to the Lake Mariner Anthropic AI Factory (410 MW initially, then 750 MW, and the $3.5 billion financing target from the July 13, 2026 headline) appear only in project metadata, without claim ids to test. The headline ambition was 800 MW deployed by end-2025 (21125, restated in 21141 and 14108), and 21125 is now marked delayed. Within that envelope, capacity shrank at the project level. Nautilus was pitched at 300 MW of nuclear-backed access (14111, 14121, 14130). By August 2022 it had become a 200 MW initial build (13715) with a contingent 100 MW expansion (13716) that also slipped. TeraWulf's economic share moved from a 150 MW joint-venture framing (14112) to a 33.33 percent equity stake (13714), and the asset has since been sold. Miner procurement shows the same pattern. The 58,000-unit 2022 delivery commitment was repeated across at least five filings (14247, 14049, 14025, 13978, 13897) and ended delayed. Claim 14142 cites 78,000 units in the same window without reconciliation. Near-term site dates fared better. The New York facility went live in Q4 2021 (21138), Pennsylvania ASIC installation began on schedule (14122, 14024), and Lake Mariner's first phase reached its year-end 2022 target (14106), though that phase was framed against a 90 MW NYPA allocation (14110, 14126) while the target called for 100 MW. The 50 MW online in 2021 claim (21124, 21140) remains unverified. The pattern is reliable execution on discrete first phases alongside slippage on portfolio-level aggregates that were restated rather than formally revised. Independent corroboration in this record is limited to the verified flags; no permit filings, NYISO or PJM queue positions, or site reporting are attached. That matters because the current pipeline includes Cayuga, Lake Hawkeye, two Kentucky campuses, Hancock County, Boyd-Greenup, and Morgantown, and all of these carry no stated capacity or cost. Over the next 90 days, watch for energized-MW disclosures at Lake Mariner against the 410 MW tranche and for whether 750 MW is presented as contracted or aspirational. Also watch the closing terms of the $3.5 billion financing, any interconnection queue filings tied to the Maryland and Kentucky sites, and whether the 2028 zero-carbon target (14245, with 90 percent claimed in 13847) survives the shift to larger, grid-dependent loads.

TeraWulf's capacity story has moved through two distinct phases. The merger-era framing set an aggregate target of 800 MW by year-end 2025 (21125, 21141, 14108). It rested on roughly 90 MW of NYPA power at Lake Mariner (14110, 14126) and a Nautilus joint venture sized at 300 MW (14111, 14121, 14130). By August 2022 Nautilus had been recast as 200 MW of initial capacity with a 100 MW expansion option (13715, 13716), with TeraWulf holding a one-third stake (13714). Nautilus is now listed as sold. Lake Mariner has re-emerged as an Anthropic-leased AI facility at 410 MW initially and 750 MW as a campus target, which is a different business from the one the 800 MW promise described. That mining target is marked delayed against its own deadline, and nothing in the record shows the company formally retiring or restating it. It appears to have been superseded rather than acknowledged.

Near-term site dates mostly held. The New York facility going live in Q4 2021 (21138), the Lake Mariner first phase by end-2022 (14106) and the Pennsylvania ASIC start (14122, 14024) are all verified. Equipment was the weak link: the 58,000-miner delivery commitment is delayed across five separate restatements (14247, 14049, 14025, 13978, 13897). The 50 MW online in 2021 (21124, 21140) remains unverifiable, and the Nautilus expansion (13716) slipped.

On independent evidence, the claims supplied here contain no permit filings, interconnection queue entries or on-the-ground reporting. The claim history also ends in early 2023. As a result, the 410-to-750 MW scaling and the $3.5 billion financing rest only on headline-level project data, not on corroborated claims. That gap matters most for the Kentucky and Maryland sites, including the Morgantown Generating Station, all still at announced status with no disclosed capacity or cost.

Over the next 90 days, watch four things. First, whether the Lake Mariner financing closes near the $3.5 billion target and on what terms. Second, whether grid-operator queue records or local filings show power at Lake Mariner beyond the initial 410 MW tranche. Third, whether Morgantown or any Kentucky site produces a concrete interconnection, zoning or capacity disclosure. Fourth, whether the company explicitly reconciles its lapsed 800 MW commitment with the current high-performance computing roadmap.

The claim record supplied here ends in early February 2023, cut off mid-entry, so this read covers TeraWulf's bitcoin-mining build phase. The HPC pivot that now defines the story, including the Lake Mariner Anthropic facility's move from 410 MW to a 750 MW campus target, appears only in the project list and should be checked against the full claim history before publication. Within that window, the headline ambition stayed fixed while the components underneath it shrank. The 800 MW by 2025 target was restated without change across 21125, 21141 and 14108, and is now flagged delayed. Nautilus fell from 300 MW (14121, 14130) to 200 MW of initial capacity with 100 MW of optional expansion (13715, 13716). TeraWulf's stake moved from a 150 MW joint-venture framing (14112) to a one-third equity interest (13714). Nautilus was later sold. Lake Mariner was the more grounded asset. Its 90 MW NYPA allocation (14110, 14126) was a counterparty commitment rather than a company aspiration, and the 100 MW first phase by year-end 2022 (14106) is verified. On timing, the near-dated promises mostly held: the New York facility in Q4 2021 (21138) and the Pennsylvania ASIC installation in mid-2022 (14122, 14024). The longer-dated equipment commitments slipped quietly. The 58,000-miner delivery figure was repeated unchanged from March through June 2022 (14247, 14049, 13978, 13897) with no acknowledgement of delay, and it sat alongside an unexplained 78,000-unit procurement figure (14142). The 2021 target of 50 MW online (21124, 21140) was never independently confirmed. The record contains no independent corroboration of pace from permits, NYISO or PJM queue positions, or site reporting. The verified statuses appear to rest on the company's own filings. This gap matters most for the Kentucky and Maryland sites, which remain announced with no capacity attached. Over the next 90 days, the key test is whether the $3.5 billion Anthropic-linked financing closes and whether the Q3 2026 results reconcile the 410 MW and 750 MW figures with energized megawatts. Readers should also watch NYISO queue filings or NYPA allocations that would cover the incremental Lake Mariner load, and any PJM interconnection entries for Morgantown that would show the Maryland plan is more than a site announcement.

The supplied claim record stops in early 2023 (it breaks off at claim 13411), so this read covers the bitcoin-mining era in detail. The current AI-campus program is visible only in the project list, without claim ids. Within that window, TeraWulf's headline capacity stayed fixed while the underlying building blocks moved. The 800 MW-by-2025 aggregate target was repeated unchanged across claims 21125, 21141 and 14108. Claim 21125 is now marked delayed, and the record contains no restated figure or acknowledgement of the miss. Nautilus shrank quietly. It was first described as 300 MW of nuclear-powered capacity (14121, 14111, 14130), then reframed as a 200 MW initial facility with a 100 MW expansion option (13715, 13716). The expansion carried a May 2024 date and is also now delayed. Fleet figures drifted too: the 58,000-miner delivery commitment (14247, 14049, 14025, 13978, 13897) sits beside a 78,000-unit procurement figure (14142), the record never reconciles the two, and the 58,000 commitment is flagged delayed in every instance. Near-term site milestones held up better. The New York facility going live in Q4 2021 (21138), the Lake Mariner first phase by year-end 2022 (14106) and the Pennsylvania ASIC installation by mid-2022 (14122, 14024) are all verified. The 50 MW-in-2021 claim (21124, 21140) and the 2022 Lake Mariner miner-count milestones (13877, 13880) remain unverifiable, and 13880 carries a deadline that predates the quarter it names. Lake Mariner's 100 MW first phase also outruns the 90 MW NYPA allocation cited in 14126 and 14110, which implies additional supply that the record does not document. On independent evidence, the record includes no permits, interconnection queue positions or site reporting. Every verification status here therefore rests on company disclosures. That gap matters more now, because the project list shows Lake Mariner rising from a 410 MW initial disclosure to a 750 MW campus target, with $3.5 billion in Anthropic-linked financing, alongside several Kentucky and Maryland sites still at the announced stage. Over the next 90 days, the key questions are whether the Lake Mariner financing closes on its stated terms, whether NYISO or NYPA filings support load beyond the original 90 MW allocation at the 750 MW scale, whether county-level permits or utility agreements appear for the Hawesville, Grayson, Boyd-Greenup and Morgantown sites, and whether management formally retires the 800 MW-by-2025 framing rather than letting it lapse.

The record as supplied ends mid-entry in February 2023 (claim 13411), so this read covers TeraWulf's bitcoin-mining era in detail and the later HPC pivot only through the project roster. Within that window, the headline capacity ambition was steady on paper. The 800 MW by 2025 target appears in October 2021 (21125), is repeated in November 2021 (21141) and again in March 2022 (14108), and is never revised even as the pieces underneath it moved. That steadiness is itself the issue. 21125 is now marked delayed, and nothing in the record shows the company formally retiring or restating the target. It simply stops being repeated after early 2022.

The individual sites tell a more mixed story. Lake Mariner's near-term promises largely held: Q4 2021 operations (21138) and a 100 MW first phase by end-2022 (14106) are both verified. Nautilus contracted without a headline change. It was presented as 300 MW of nuclear-powered capacity (14111, 14121, 14130), then reframed in August 2022 as 200 MW initial with a 100 MW expansion (13715, 13716), and that expansion, keyed to May 2024, is delayed. The shift from a 150 MW joint-venture framing (14112) to a one-third equity stake (13714) also changes TeraWulf's attributable capacity without any restated total. Equipment followed the same pattern. The 58,000-miner commitment for 2022 (14247, 14049, 14025, 13978, 13897) is delayed, and a separate 78,000-unit procurement figure (14142) is never reconciled with it.

Independent corroboration is thin in what was provided. The record contains no permit filings, NYISO or PJM queue entries, or on-site reporting, so the verified statuses rest on company disclosures rather than third-party evidence.

The current portfolio is a different business. Abernathy has been sold. Lake Mariner is now organized around an Anthropic-leased AI facility whose disclosed size moved from 410 MW to 750 MW, alongside a $3.5 billion financing target. About ten other sites in New York, Kentucky and Maryland are announced with no stated capacity or cost. Over the next 90 days, watch three things. First, whether the $3.5 billion financing closes, and on what terms. Second, whether the 750 MW figure gets a dated energization schedule and a matching NYISO interconnection position. Third, whether any of the Kentucky or Maryland sites acquire concrete megawatts, permits or a tenant. Given the 2021 to 2023 pattern of targets left unrevised rather than restated, quiet changes to these figures deserve as much scrutiny as new announcements.

TeraWulf's capacity story has expanded, then quietly reset around a different business. The founding headline was 800 MW deployed by 2025 (21125, 21141, 14108), and the record now marks it delayed. Nautilus is a smaller version of the same pattern. It was pitched as a 300 MW nuclear-powered facility (14130, 14111, 14121), then reframed as 200 MW of initial capacity (13715). TeraWulf's exposure fell from a 150 MW joint-venture share (14112) to a one-third equity stake (13714). The 100 MW expansion due by May 2024 (13716) is also marked delayed. None of these revisions came with an explicit walk-back. The new figures simply appeared alongside the old ones.

Near-dated site milestones have mostly held. The New York facility came online in Q4 2021 (21138). Lake Mariner's first phase was online by year-end 2022 (14106). Pennsylvania ASIC installation started by mid-2022 (14122, 14024). Equipment deliveries are the weaker spot. The 58,000-miner commitment for 2022 is repeated across five filings (14247, 14049, 14025, 13978, 13897) and is delayed in every instance. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. The 3,300-unit and 3,000-unit Lake Mariner ramp targets (13880, 13877) remain unverifiable. The 100 percent zero-carbon target for 2028 (14245, 13846, 13826, 13439) has been restated without change, against a stated 90 percent baseline (13847).

The independent evidence is thin. The supplied record contains no permit filings, no NYISO or PJM interconnection queue entries, and no on-the-ground reporting. The verified tags rest on the company's own subsequent disclosures. The claim excerpt also ends in early 2023, while the project roster runs into 2026. That roster shows Abernathy sold, Lake Mariner rebranded as an Anthropic-leased AI factory moving from 410 MW to a 750 MW campus target, and a $3.5 billion financing target. It also shows eight-plus announced sites in New York, Kentucky and Maryland with no capacity or cost attached. The AI-era pace therefore cannot be tested against this record at all.

Over the next 90 days, watch four things. First, whether the $3.5 billion Anthropic-facility financing actually closes, and on what terms. Second, whether the third-quarter report breaks the 750 MW Lake Mariner target into energized megawatts and dated phases. Third, whether interconnection or load-study filings appear for the Kentucky and Maryland sites. Fourth, whether the 800 MW and Nautilus expansion targets are formally retired or silently dropped.

The record supplied here stops in early 2023 (it breaks off at claim 13411), so TeraWulf's current AI-hosting pivot appears only in the project metadata: a 410 MW initial disclosure at Lake Mariner expanding to a 750 MW campus target, tied to a $3.5 billion financing headline. That growth should be read against the mining-era record. The 2021 merger materials set an 800 MW deployment target for 2025 (21125, 21141), restated unchanged through March 2022 (14108). It is now tagged delayed, and nothing in the supplied claims acknowledges a revised date. Component numbers were also unstable. Nautilus was described as 300 MW (14130, 14121), then as a 150 MW TeraWulf share (14112), then as 200 MW initial with a 100 MW expansion (13715, 13716). That expansion missed its May 2024 date. Equipment followed a similar pattern: a 58,000-miner commitment repeated across five filings (14247, 14049, 14025, 13978, 13897) is delayed, while a separate 78,000-unit procurement figure (14142) was never reconciled with it. Near-term operational dates did better. The Lake Mariner first phase (14106) and the Pennsylvania ASIC installation start (14122, 14024) are verified. This points to a company that hits concrete phase-one milestones while its aggregate targets slide quietly. The one target restated most consistently, 100 percent zero-carbon by 2028 (14245, 14026, 13846, 13826, 13439), is a sourcing claim rather than a capacity claim. Independent corroboration is thin. Apart from verification status, the record has no permit filings, NYISO or PJM queue entries, or site reporting. Two mapping problems also limit what can be concluded. Bitcoin-era Lake Mariner claims such as the 90 MW NYPA allocation (14110, 14126) are attached to the Anthropic project, and Nautilus, a Susquehanna-powered Pennsylvania site (14129), is listed under Tompkins County, New York. Both should be fixed before the 410-to-750 MW step-up is credited as continuity. Over the next 90 days, watch for five things. The first is whether the Q3 2026 results tie the 750 MW figure to dated phases. The second is whether the $3.5 billion financing actually closes or remains a headline. The third is whether NYPA or NYISO filings support power beyond the original 90 MW allocation. The fourth is whether the Abernathy sale closes. The fifth is whether any of the Kentucky or Maryland sites, including Morgantown Generating Station, produces a queue position or permit, or whether they stay at announced status.

TeraWulf's capacity story has been rewritten more often than it has been revised. The merger-era framing promised 800 MW deployed by year-end 2025 (21125, 21141, 14108), all for bitcoin hashrate, and the company repeated that figure unchanged through early 2023. Claim 21125 is now marked delayed. Nothing in the record shows the company formally retiring or restating that target. Instead the narrative moved to HPC leasing. The Lake Mariner project now carries a 410 MW initial disclosure, a 750 MW campus target, and a $3.5 billion financing goal tied to an Anthropic lease. That changes what the megawatts are for, not just how many there are. The 750 MW figure should not be read as continuity with the 800 MW mining pledge.

The early record is mixed at the component level. Lake Mariner's first phase and the Pennsylvania ASIC start held (14106, 14024, 14122). The 58,000-miner delivery commitment slipped (14247, 14049, 13978, 13897), and the 100 MW Nautilus expansion tied to May 2024 is delayed (13716). Nautilus itself shrank in the company's own telling, from 300 MW (14130, 14121) to a 200 MW initial build (13715) within five months, with no explicit reconciliation. The company has also announced sites in Kentucky, Maryland and New York and sold Abernathy. That adds optionality but weakens any single delivery promise.

Independent evidence is the gap. The claim history supplied here contains no permits, no interconnection queue positions and no on-the-ground reporting. The visible claim list also ends in February 2023. As a result, the Anthropic-era timeline cannot be checked against the company's own dated promises, let alone outside sources. In this record, Lake Mariner's under-construction status rests on company disclosure alone.

Over the next 90 days, watch four things. First, whether the $3.5 billion financing closes and on what terms. Second, whether NYPA or NYISO filings show load additions consistent with 410 MW rather than the original 90 MW NYPA allocation (14110, 14126). Third, whether the company commits to dates for energizing the first HPC halls. Fourth, whether any of the announced Kentucky and Maryland sites produce a filed interconnection request rather than another press release.

TeraWulf's capacity story has been reset more than once. The 2021 merger materials set an 800 MW deployment target by 2025 (21125, 21141), repeated in 2022 (14108). The record now marks it delayed, and no later claim restates or retires it. Nautilus shows the same pattern at project scale. It was first described as 300 MW of nuclear-powered capacity (14130, 14111), then as a 150 MW joint venture (14112), and later as a 200 MW initial build (13715) with a 100 MW expansion promised by May 2024 (13716), which is also flagged delayed. The company's stake was set at one third (13714), so its attributable megawatts were always well below the headline. The current Lake Mariner disclosure moved from 410 MW to a 750 MW campus target, alongside a $3.5 billion financing target tied to the Anthropic lease. That is the largest upward revision in the record, but the claims supplied here stop in early 2023, so its sourcing cannot be cited by id.

Near-term dates have held better than long-range ones. New York operations in Q4 2021 (21138), Pennsylvania ASIC installation (14122, 14024) and the first Lake Mariner phase by end-2022 (14106) are verified. The 58,000-miner delivery commitment slipped across five successive filings (14247, 14049, 14025, 13978, 13897) without a restated date. The 78,000-unit procurement figure (14142) sat alongside it unreconciled. The 2021 promise of 50 MW online (21124, 21140) was never independently confirmed.

The record contains no permits, interconnection queue positions or site reporting for any project. The only power evidence is the 90 MW NYPA allocation (14110, 14126), far below a 750 MW campus. A dozen sites in Kentucky, Maryland and New York carry an announced status with no capacity or cost attached. Abernathy is listed as sold.

Over the next 90 days, watch four things. First, whether the Anthropic-linked financing closes at or near $3.5 billion. Second, any NYPA or NYISO filings that show grid access beyond the original 90 MW. Third, whether Q3 reporting puts megawatt figures and in-service dates on the announced Kentucky and Morgantown sites. Fourth, whether the company formally retires or restates the 800 MW target rather than letting it lapse.

TeraWulf's capacity story has been rewritten more than it has been reconciled. The founding promise of 800 MW deployed by 2025 (21125, restated in 21141 and 14108) never came close and is now marked delayed. The company has not formally retired that figure; it has been replaced by AI-hosting targets. The Pennsylvania joint venture shows the same pattern at smaller scale. Nautilus was pitched at 300 MW of nuclear-powered capacity (14111, 14121, 14130). TeraWulf's share was framed as 150 MW (14112). The facility was then re-described as a 200 MW initial build (13715), with a 100 MW expansion that later slipped (13716). Each step lowered the headline number without an explicit correction. The current Lake Mariner Anthropic AI Factory runs the other way, from 410 MW at initial disclosure to a 750 MW campus target backed by a $3.5 billion financing goal. The record does not yet tie that growth to a disclosed power source.

Delivery on near-term, site-specific dates has been better than on aggregate targets. The New York facility came online in Q4 2021 (21138), Lake Mariner's 100 MW first phase was verified (14106), and Pennsylvania ASIC installation began on schedule (14024, 14122). Fleet-level promises held up worse. The 58,000-miner delivery commitment was repeated five times through mid-2022 (14247, 14049, 14025, 13978, 13897) and is delayed. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. The 2021 claim of 50 MW online (21124, 21140) remains unverifiable. The 100 percent zero-carbon target for 2028 (14245, 13846, 13439) has been restated consistently, against 90 percent reported usage (13847).

Independent corroboration is thin. This record contains no permit filings, interconnection-queue entries or site reporting for any project, including the eight announced Kentucky, Maryland and New York sites that have no capacity or cost figures. The Lake Mariner NYPA allocation of 90 MW (14110, 14126) is the only documented power agreement at that site. It is far below the AI factory's targets, so the step to 410 or 750 MW depends on interconnection capacity the record does not show. The claim history supplied here also ends in early 2023, so recent statements could not be checked. Nautilus's listed location in Tompkins County conflicts with its Pennsylvania claims and should be corrected.

Over the next 90 days, watch for whether the $3.5 billion Lake Mariner financing closes and on what terms, any NYISO queue or NYPA filing that supports load beyond 90 MW, a dated energization schedule for the first AI phase, and whether any announced site moves to a permit or interconnection filing.

TeraWulf's capacity story has expanded in headline terms while the dated commitments have repeatedly come in late. The original 2021 framing promised 50 MW online that year (21124, 21140) and 800 MW deployed by 2025 (21125, 21141, 14108). The 2025 target is now marked delayed, and nothing in this record shows the company formally restating it. The 2022 build-out was paced by a 58,000-miner delivery commitment that every subsequent filing carried forward unchanged and that is now flagged delayed (14247, 14049, 14025, 13978, 13897). A separate 78,000-unit procurement figure (14142) was never reconciled with it. Lake Mariner's near-term milestones did land: the 100 MW first phase by year-end 2022 is verified (14106). The more granular miner counts for Q3 2022 (13880, 13877), however, remain unverifiable. Nautilus shows real contraction. Disclosures described access to 300 MW (14111, 14121, 14130). They then narrowed to a 200 MW initial facility (13715) with a contingent 100 MW expansion that slipped past its May 2024 date (13716), and to a one-third equity stake (13714). The current headline is a 410 MW initial disclosure that grew into a 750 MW campus target at Lake Mariner, financed against an Anthropic lease at roughly $3.5 billion. That is a pivot from mining to AI hosting rather than a continuation of the 800 MW mining plan, yet the record tags 2022 bitcoin-mining claims (14105, 14140, 14110) to the AI Factory project. That blurs any honest baseline. The project list also places Nautilus in Tompkins County, New York, while its claims describe a Pennsylvania facility powered by Susquehanna (14129, 14113), which suggests a mapping error worth correcting. The supplied claim history stops in February 2023, so revisions from 2023 to 2026 cannot be assessed here. No permits, interconnection-queue entries or site reporting appear in the record. The pace of the 750 MW campus, and of the Kentucky, Maryland and Cayuga sites listed only as announced with no capacity, rests entirely on company statements. The Abernathy sale shows that announced campuses can also leave the portfolio. Over the next 90 days, watch whether the Anthropic financing closes at the stated size and on what terms. Also watch for NYISO interconnection or NYPA allocation filings that support load beyond the legacy 90 MW contract, for energization dates for the first AI phase, and for any permitting activity at Cayuga, Morgantown or the Kentucky sites that would move them beyond announcement.

This record ends partway through a claim from February 2023 (13408). Nothing from the past three and a half years of TeraWulf's disclosures is visible, including the Anthropic-leased Lake Mariner build that now dominates the project list. This read therefore covers the bitcoin-era trajectory, and the conclusions about the current program should be held loosely. On capacity, the headline ambition stayed fixed while project-level figures moved. The 800 MW by 2025 target appears unchanged in 21125, 21141 and 14108, and has since been marked delayed. Nautilus shrank in practice. It was described as a 300 MW nuclear-powered site (14121, 14130, 14111) and then as a 150 MW joint-venture share (14112). By August 2022 it was framed as 200 MW initial capacity with a 100 MW expansion option (13715, 13716), and that expansion is now flagged delayed. Lake Mariner's early figures were modest: a 90 MW NYPA supply (14110, 14126), a 50 MW first building (13881) and a 100 MW first phase (14106). The jump to 410 MW and then 750 MW in the project summary is the largest scaling in the record, but the underlying claims are not in this excerpt to evaluate. On dates, the near-term promises mostly held. New York operations began in Q4 2021 (21138), the first Lake Mariner phase was online by end-2022 (14106), and the Pennsylvania ASIC installation started on schedule (14122, 14024). The weaker pattern is the 58,000-miner delivery commitment. It was restated at least five times through June 2022 (14247, 14049, 14025, 13978, 13897) without any visible acknowledgement of slippage, and a 78,000-unit procurement figure (14142) sits alongside it unreconciled. The 2021 claims of 50 MW online that year (21124, 21140) were never independently confirmed. No permits, interconnection-queue entries or on-the-ground reporting appear anywhere in this record. So the announced pace, especially the 750 MW Lake Mariner target and the several Kentucky and Maryland sites listed as announced with no stated capacity, rests entirely on company statements. In the next 90 days, the priorities are: whether the $3.5 billion Lake Mariner financing closes on the terms reported in July; whether NYISO queue filings or New York permits support the 750 MW figure rather than the 410 MW one; and whether the next quarterly report gives energization dates for leased capacity. Any Kentucky or Maryland site that converts from announced to a filed interconnection request would be the first independent sign of the broader pipeline being real.

The supplied claim log ends abruptly in February 2023 (claims 13412 and 13414). Everything after that, including the AI pivot that defines TeraWulf's current project list, is missing from the evidence. The 410 MW and 750 MW Lake Mariner figures and the $3.5 billion Anthropic financing target appear only as project attributes, not as dated claims. So this read covers the bitcoin-era trajectory, and the AI-era ramp needs a fuller pull before publication.

On capacity, the original anchor was 800 MW deployed by 2025, repeated in claims 21125, 21141 and 14108. It is now correctly marked delayed. The company never formally revised it down in the record. Instead the target faded as the narrative shifted. Nautilus shows the clearest quiet contraction. It was framed as a 300 MW nuclear-powered facility (claims 14111, 14121, 14130). Within five months it was described with 200 MW of initial electrical capacity and a 100 MW expansion contingent on later conditions (claims 13715, 13716), and the expansion is itself now delayed. The 58,000-miner delivery commitment (claims 14247, 14049, 14025, 13978, 13897) was restated five times without adjustment and is flagged delayed. A parallel 78,000-unit procurement figure (claim 14142) was never reconciled with it.

Dates held best at the near term. The Q4 2021 New York start (claim 21138) and the mid-2022 Pennsylvania ASIC installation (claims 14122, 14024) are verified, as is the 100 MW first phase at Lake Mariner (claim 14106). Multi-year targets slipped without acknowledgement in the record. There are also data-hygiene problems. Nautilus is mapped to Tompkins County, New York, although its claims describe Talen's Susquehanna site in Pennsylvania (claims 14113, 14129). Bitcoin-era Lake Mariner claims are tagged to the Anthropic AI Factory project. Claim 21075 is an IKONICS-era 2006 auditor item unrelated to facilities.

The record contains no independent corroboration: no permits, no NYISO or PJM interconnection queue positions, and no on-site reporting. For the Kentucky and Maryland announcements, and for a 750 MW Lake Mariner campus, that absence is material.

Over the next 90 days, watch for three things. First, whether TeraWulf closes the Anthropic-linked financing near the $3.5 billion headline. Second, NYISO filings that support load beyond the 410 MW initial disclosure. Third, whether Q3 disclosures give energization dates for Lake Mariner phases or restate the campus target in a new form.

TeraWulf's record reads as two companies sharing a ticker. The early claims describe a bitcoin miner promising scale it did not reach. The newer project list describes an AI-infrastructure landlord whose headline numbers have no claim ids in this log. The supplied claim history stops in February 2023, so the 410 MW initial disclosure and 750 MW campus target for the Lake Mariner Anthropic AI Factory, and the $3.5 billion financing figure attributed to a July 13, 2026 headline, cannot yet be tested against a dated promise trail.

The earlier capacity story is one of quiet contraction. The 800 MW-by-2025 target appears in claims 21125, 21141 and 14108, and 21125 is now marked delayed. It then drops out of the 2022 claims without any recorded walk-back. Nautilus shows the same pattern within the same year. It was described as 300 MW in claims 14111, 14121 and 14130, then as 200 MW initial capacity (13715), with a 100 MW expansion (13716) that is also delayed. Equipment targets diverge as well: 58,000 miners in claims 14123, 14025 and 13897 sit alongside 78,000 units in claim 14142.

On timelines, the verified items are the modest ones: New York operational in Q4 2021 (21138), Pennsylvania ASIC installation (14122, 14024), and a first Lake Mariner phase (14106). The ambitious items, including the 58,000-miner delivery (14247, 14049, 13978), are uniformly delayed. Near-term operational markers such as claims 21124 and 13880 remain unverifiable rather than confirmed.

The dataset contains no independent corroboration of the announced pace, such as permits, NYISO queue positions or local reporting. It also has linkage problems that weaken attribution. Nautilus is mapped to Tompkins County, New York, while its claims reference Susquehanna in Pennsylvania. Bitcoin-era Lake Mariner claims are filed under the Anthropic project. Claim 21075 is a 2006 IKONICS auditor item. Any read of the current trajectory therefore rests on the company's own word.

Over the next 90 days, the first thing to check is whether third-quarter 2026 disclosures restate the 750 MW figure or revert to 410 MW. Next is whether the $3.5 billion financing actually closes or remains a target. Third is whether NYISO interconnection filings or Somerset-area permits support load growth on that scale. Finally, the many announced-only sites, including Cayuga, Lake Hawkeye, the Kentucky campuses and Morgantown, need any first filing that would move them beyond press-release status.

TeraWulf's capacity story has shrunk in headline terms and then reinvented itself under a different name. The founding target of 800 MW deployed by 2025 (21125, 21141, 14108) was repeated without change through early 2022. It is now marked delayed. The record contains no later filing that formally retires or restates it, and that silence matters more than the miss itself. Nautilus followed the same arc. It was first framed as 300 MW of nuclear-adjacent access (14111, 14121, 14130). By August 2022 it had become a 200 MW initial build (13715) with a 100 MW expansion option (13716) that is also now delayed, and TeraWulf's economic exposure was a one-third stake (13714) rather than the full figure implied earlier. Lake Mariner moved the other way, growing from a 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881) to a project-level disclosure of 410 MW and then a 750 MW campus target under the Anthropic lease. That last step is a change of category, from bitcoin hosting to AI infrastructure. It should not be read as the original mining plan arriving late.

On dates, the verified items are the near-dated ones: the Q4 2021 New York start (21138), the first-phase online target at Lake Mariner (14106), and ASIC installation at Nautilus (14024, 14122). The 58,000-miner delivery commitment was restated four times between March and June 2022 (14247, 14049, 13978, 13897) without any narrowing, even though it ultimately slipped. The 78,000-unit procurement figure (14142) sits alongside it unreconciled. Claim 13880 set a July 1, 2022 deadline in a statement dated June 30, which says little about forecasting discipline.

The supplied record contains no independent evidence: no NYISO queue positions, no building or air permits, no site reporting. The ledger is also truncated after February 2023, so the AI-era claims behind the 410 MW, 750 MW and $3.5 billion figures cannot be cited by id here. Project tagging is unreliable as well. Nautilus is listed in Tompkins County, New York, while its claims reference Susquehanna in Pennsylvania, and 2022 mining claims are attached to the Anthropic facility.

Over the next 90 days, watch for three things. First, whether the $3.5 billion financing closes on the terms headlined in July. Second, NYISO interconnection filings that would support load above the legacy allocation. Third, any explicit company reconciliation of the abandoned 800 MW and Nautilus expansion targets against the new campus figures.

The record splits into two eras. Nearly every dated claim belongs to TeraWulf's 2021-2023 bitcoin-mining buildout, while the project now driving the story, the Lake Mariner Anthropic AI Factory (410 MW initially, then a 750 MW campus target, with a $3.5 billion financing figure from a July 2026 headline), has no dated claims in the excerpt. Several 2022 Lake Mariner mining claims (14106, 14110, 13881) are tagged to that AI project, which ties two different builds together. The Nautilus entries describe a Pennsylvania facility powered by Susquehanna (14113, 14129) but are filed under Tompkins County, New York. Both points mean the record needs cleanup before it can show a trajectory for the AI campus.

Within the mining era, announced capacity grew in headlines and shrank in practice. The 800 MW-by-2025 target was repeated without change (21125, 21141, 14108) and is now marked delayed, with no restated figure in the record. Nautilus fell from a 300 MW nuclear-powered facility (14130, 14111) to 200 MW initial capacity plus a contingent 100 MW (13715, 13716). TeraWulf's position also moved from a 150 MW joint-venture share (14112) to a one-third equity stake (13714). The 58,000-miner delivery commitment was restated five times (14247, 14049, 14025, 13978, 13897) and ended delayed, while a parallel 78,000-unit procurement figure (14142) was never reconciled with it.

Near-term dates held better than long-range ones. New York operations by Q4 2021 (21138), the Pennsylvania ASIC start by mid-2022 (14024, 14122) and Lake Mariner's first phase by year-end 2022 (14106) are verified. The 50 MW-online-in-2021 claim (21124, 21140) remains unverifiable, and the Nautilus expansion date slipped (13716) without a revised target appearing. The pattern is a company that delivers discrete phases while letting aggregate targets lapse quietly.

The record contains no independent evidence for any of these claims: no permits, no NYISO or PJM queue positions, and no site reporting. Announced-only sites in Kentucky and Maryland, including the Morgantown Generating Station, rest entirely on company statements.

Over the next 90 days, the key test is whether the 410 MW to 750 MW expansion at Lake Mariner shows up in NYISO interconnection filings or New York Power Authority allocation changes. A second marker is whether the $3.5 billion financing closes with terms disclosed. A third is whether any Kentucky or Maryland site gains a queue position or local permit that would move it beyond announcement.

The supplied record breaks off partway through claim 13414 in February 2023, so this read covers the bitcoin-mining era in detail and the AI pivot only through the project table. The headline ambition has not shrunk so much as been replaced. The 800 MW-by-2025 target was repeated in claims 21125, 21141 and 14108 and is now marked delayed. The Lake Mariner Anthropic AI Factory figures of 410 MW, then 750 MW, have taken its place, but without per-claim dates we cannot test whether that escalation came with a matching schedule.

Nautilus is the clearest case of quiet contraction. The 300 MW nuclear-powered figure (14111, 14121, 14130) became 200 MW of initial capacity (13715), with the remaining 100 MW recast as an expansion that is now delayed (13716). TeraWulf's exposure narrowed further, from a 150 MW joint-venture share (14112) to a 33.33 percent equity stake (13714). None of this was framed as a revision. The project table also places Nautilus in Tompkins County, New York, while the claims tie it to Susquehanna in Pennsylvania, an attribution conflict editors should resolve before publishing.

The early record is mixed on dates. The New York start (21138), the Pennsylvania ASIC installation (14122, 14024) and the first 100 MW phase at Lake Mariner (14106) were verified. The 58,000-miner delivery commitment, restated five times (14247, 14049, 14025, 13978, 13897), slipped, and a parallel 78,000-unit procurement figure (14142) was never reconciled with it. Lake Mariner's contracted NYPA supply of 90 MW (14110, 14126) sits well below the current 750 MW campus target.

On independent evidence, the record contains no permits, NYISO or PJM queue positions, or site reporting for any project. The pace of the Anthropic build therefore rests on company statements alone. The same holds for the nine announced sites in Kentucky, Maryland and New York, which carry no capacity or cost figures at all.

For the next 90 days, three things matter. First, whether the $3.5 billion Lake Mariner financing closes on the terms in the July 13, 2026 headline. Second, whether NYPA or NYISO filings show new power allocations large enough to support 410 MW, let alone 750 MW. Third, whether any of the Kentucky or Morgantown sites appear in an interconnection queue or a county permit docket. Until those documents surface, the AI-era capacity figures should be read as targets, not as a delivery trajectory.

The record supplied here is lopsided. It stops at claim 13408 in early 2023, yet the project list carries a 2026-era Lake Mariner AI campus quoted at 410 MW and later 750 MW, with a $3.5 billion financing target. That three-year gap matters more than any single claim, because the material describing the current build is not yet in the evidence base. Several Lake Mariner bitcoin-mining claims (14105, 14106, 13880) are tagged to the Anthropic facility. Those claims describe a different asset generation and should not be read as that project's delivery history. Claim 21075, a 2006 IKONICS auditor appointment, is also misattributed noise.

On capacity, the headline 800 MW-by-2025 target was repeated without change (21125, 21141, 14108) and is now marked delayed. Nautilus shows quiet shrinkage. It was framed as 300 MW of nuclear-powered access (14111, 14130, 14121), restated as 200 MW initial with a 100 MW option (13715, 13716), and booked as a 150 MW share (14112). The equity stake then fell to 33.33 percent (13714) with no explicit reconciliation. That 100 MW expansion, promised by May 2024, is delayed. Miner commitments also drifted between 58,000 units (14247, 14049, 13897) and 78,000 (14142), and the 58,000 figure is marked delayed every time it recurs.

Near-term dates held better than long-range ones. New York operations in Q4 2021 (21138), Pennsylvania ASIC installation by mid-2022 (14122, 14024) and the 100 MW first phase at Lake Mariner by end of 2022 (14106) all verify. The 50 MW-in-2021 claim (21124, 21140) remains unverifiable. Equipment-dependent promises slipped, while site-energization promises mostly landed.

Independent corroboration in this record is thin. The 90 MW NYPA allocation (14110, 14126) and the Susquehanna supply term (14113) are contractual disclosures, not interconnection-queue or permit evidence. Nothing here substantiates the step from roughly 100 MW to a 750 MW campus.

Over the next 90 days, four things matter most. The first is whether the $3.5 billion Lake Mariner financing closes, and on what capacity basis. The second is whether NYISO queue positions or upgrade studies support load well above the NYPA allocation. The third is whether 410 MW versus 750 MW is reconciled in filings rather than headlines. The fourth is whether any of the announced Kentucky and Maryland sites, including Morgantown Generating Station, produce permits or utility agreements rather than further announcements.

TeraWulf's capacity story has been rebuilt rather than scaled. The founding target of 800 MW deployed by 2025 (21125, 21141, 14108) was a bitcoin-mining figure split between Lake Mariner and the Nautilus joint venture with Talen. It is now marked delayed, and no claim on record restates or retires it. Nautilus shrank within a single year. It was 300 MW of nuclear-powered access in March 2022 (14111, 14121), then a 200 MW initial facility with a one-third stake by August (13715, 13714). Its 100 MW expansion (13716) passed its May 2024 date. The current headline, 410 MW and then a 750 MW campus target at the Anthropic-leased Lake Mariner facility, appears only in project metadata. The claim record here stops in February 2023 and contains nothing that bridges a 90 MW NYPA allocation (14110, 14126) to a multi-hundred-megawatt AI campus.

On dates, the near-term operational promises mostly held. The first New York facility came online in Q4 2021 (21138), Lake Mariner's first 100 MW phase was verified for end-2022 (14106), and Pennsylvania ASIC installation began on schedule (14122, 14024). The slippage was in equipment and aggregate scale. The company repeated its 58,000-miner delivery commitment in five 2022 filings (14247, 14049, 14025, 13978, 13897) without changing it, and the commitment was later marked delayed. That is restatement, not acknowledged revision. The 50 MW 2021 figure (21124, 21140) was never independently confirmed.

The record contains no permits, interconnection-queue entries or site reporting, so the announced pace rests entirely on company statements. That matters most for the 750 MW target, which implies power well beyond any documented supply agreement. It also matters for the long list of announced Kentucky and Maryland sites, none of which has a stated capacity or cost. The 2006 auditor claim (21075) predates the company's formation and looks misattributed to this issuer; it should be checked before publication.

Over the next 90 days, there are four things to watch. The first is whether the $3.5 billion Lake Mariner financing closes on the stated terms. The second is whether NYISO or NYPA filings show load additions consistent with 410 MW. The third is whether the 750 MW figure is tied to a dated phase schedule. The fourth is whether TeraWulf formally retires the 800 MW mining target or lets it lapse silently.

TeraWulf's capacity story has been rewritten more than it has been tracked. The 800 MW-by-2025 headline (21125, 21141, 14108) anchored the 2021-2022 pitch. It is now marked delayed, and nothing in the record shows the company formally retiring or restating it. Nautilus is the clearest case of quiet shrinkage. It was first framed as 300 MW of nuclear-powered capacity (14111, 14130, 14121), then as a 150 MW TeraWulf share of the Talen joint venture (14112), and finally as a 200 MW initial facility (13715) in which TeraWulf held 33.33 percent (13714). The 100 MW expansion (13716) slipped past its May 2024 marker. Lake Mariner moved the other way. A 90 MW NYPA supply (14110, 14126) and a 100 MW first phase (14106, verified) have since been reframed around an Anthropic-leased AI facility carrying 410 MW and 750 MW targets and a $3.5 billion financing headline. None of those later figures appears as a dated claim in this extract, so their provenance cannot be scored here.

On timing, the early operational markers mostly held. New York went live in Q4 2021 (21138) and the Pennsylvania ASIC installation was verified (14122, 14024). The 50 MW-in-2021 claim (21124, 21140) and the 3,300-miner Q3 2022 target (13880) remain unverifiable. Equipment is the weak point. The 58,000-miner 2022 delivery commitment was repeated five times without revision (14247, 14049, 14025, 13978, 13897) and ended up delayed. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. The 100 percent zero-carbon-by-2028 pledge (14245, 13846, 13826, 13439, 13411) has been restated consistently against a stated 90 percent baseline (13847), but it has not been independently tested.

Independent corroboration is thin. The verified statuses rest on company disclosures, not on permits, NYISO or PJM interconnection queue entries, or on-the-ground reporting. The ledger also effectively stops in early 2023, while the project list has grown to include eight announced sites in Kentucky, Maryland and New York with no stated capacity or cost.

Over the next 90 days, watch for dated, attributable claims that put the 410 MW and 750 MW Lake Mariner figures on the record. Watch also for NYISO large-load or interconnection filings that match that scale, the close and terms of the $3.5 billion financing, and the Q3 2026 disclosures, to see whether energization dates for the Anthropic halls are stated explicitly or left open. Any capacity number attached to the Kentucky and Maryland sites should be checked against utility and county filings before it is treated as more than a placeholder.

The record supplied here breaks off partway through claim 13411 in early February 2023, so this read covers the bitcoin-mining era of TeraWulf's disclosures and cannot cite anything behind the 410 MW and 750 MW Lake Mariner AI campus figures or the $3.5 billion financing headline from July 2026. Those numbers appear only in the project summary, not as dated claims, so they cannot yet be tested against the company's own earlier statements.

Within the 2021 to 2023 window, the headline ambition stayed fixed while the components underneath it moved. The 800 MW by 2025 target was repeated in claims 21125, 21141 and 14108 and is now marked delayed. Nautilus is the clearest case of drifting capacity. It was described as a 300 MW nuclear-powered facility (14121, 14111, 14130), then as a 150 MW Talen joint-venture share (14112), and by August 2022 as 200 MW initial capacity with a 100 MW expansion (13715, 13716). TeraWulf's equity stake was reported at 33.33 percent (13714). The expansion is flagged delayed, and the later stake disclosure does not reconcile cleanly with the earlier 150 MW framing. Equipment followed the same pattern: the 58,000-miner delivery commitment (14247, 14049, 14025, 13978, 13897) slipped without a revised date appearing in the record, alongside a larger 78,000-unit procurement figure (14142) that was never verified.

Near-term execution was better. Lake Mariner's first phase by year-end 2022 (14106), the New York facility's Q4 2021 start (21138) and the Pennsylvania ASIC installation (14122, 14024) are all verified. Smaller interim markers, such as 50 MW online in 2021 (21124, 21140) and the 3,300-miner target (13880), remain unverifiable. Several 2022 mining-era claims are tagged to the Anthropic AI Factory project, which mixes two different facility generations and should be separated before any trajectory is charted.

The record contains no independent evidence: no permits, no NYISO or PJM queue entries, no site reporting. Announced pace therefore cannot be corroborated from this dataset alone.

Over the next 90 days, watch the third-quarter 2026 results and whether they restate the 410 MW or 750 MW figure with an energization schedule. Also watch for confirmation that the $3.5 billion financing has closed rather than merely been targeted, NYISO interconnection filings that match the 750 MW campus claim, and any first permitting activity at the announced Kentucky, Maryland and Cayuga sites, none of which has disclosed capacity.

TeraWulf's capacity story has been rewritten more than revised. The original thesis was a bitcoin-mining footprint of 800 MW by year-end 2025 (21125, 21141, 14108), with 50 MW online in 2021 (21124, 21140). The 2025 aggregate target has lapsed and is flagged delayed, and the record contains no claim that restates or formally retires it. Today's flagship is the Lake Mariner Anthropic AI Factory: 410 MW at initial disclosure, now a 750 MW campus target, with $3.5 billion in financing sought per the July 13, 2026 headline. That is a change of asset class, not just scale. Measured against the only power contract the record documents at Lake Mariner, 90 MW from NYPA (14110, 14126), the 750 MW figure is roughly eightfold larger.

Date discipline was mixed in the mining era. Some near-term milestones held: the New York facility went operational in Q4 2021 (21138), Lake Mariner's first phase came online by year-end 2022 (14106), and Pennsylvania ASIC installation began by mid-2022 (14122, 14024). The supply side slipped. The 58,000-miner delivery commitment was repeated across five filings (14247, 14049, 14025, 13978, 13897) and is now marked delayed. Nautilus shrank quietly. It went from 300 MW of nuclear-powered access (14111, 14121) and a 150 MW TeraWulf share (14112) to 200 MW initial capacity with a 100 MW expansion that slipped past its May 2024 date (13715, 13716), and to a one-third equity stake (13714). We found no explicit acknowledgement of that resizing in the claims provided.

Independent corroboration is thin. The record contains no permits, interconnection-queue entries, or site reporting for any project. That matters most for the long list of announced but unsized sites: Cayuga, Lake Hawkeye, Justified, Chesapeake/Morgantown, Muskie, Hancock and Boyd-Greenup. It also matters that Abernathy, the one Texas HPC campus, is listed as sold before any capacity was disclosed. The supplied claim history also cuts off in early 2023, so the AI-era figures cannot yet be tied to dated claim ids.

Over the next 90 days, watch four things. First, whether the $3.5 billion Lake Mariner financing closes and on what terms. Second, whether TeraWulf discloses power procurement or NYISO interconnection filings that bridge the gap from 90 MW toward 410 MW, let alone 750 MW. Third, whether Akela Data's partial operation becomes a dated energization milestone. Fourth, whether any Kentucky or Maryland site attaches a megawatt figure, utility agreement or local permit to its announcement.

TeraWulf's capacity story has repeatedly been resized, and the record shows the revisions as reframings rather than acknowledged misses. The original anchor, 800 MW deployed by 2025 (claims 21125, 21141, 14108), was repeated across three filings and is now marked delayed, with no claim on record conceding the shortfall or setting a replacement date. Nautilus followed the same pattern on a smaller scale. It began as a 300 MW nuclear-powered facility (14111, 14121, 14130) with TeraWulf holding 150 MW through the Talen joint venture (14112). By August 2022 it had become 200 MW of initial capacity (13715), with the remaining 100 MW recast as a potential expansion (13716) that missed its May 2024 marker, and TeraWulf's stake had fallen to one third (13714). The company's attributable Nautilus exposure therefore shrank by more than half without any single disclosure saying so.

Near-term operational dates have a better record. The New York facility came online in Q4 2021 as promised (21138), Pennsylvania ASIC installation began on schedule (14122, 14024), and Lake Mariner's first 100 MW phase was verified by year-end 2022 (14106). Equipment commitments were weaker. The 58,000-miner delivery target is flagged delayed in five separate claims (14247, 14049, 14025, 13978, 13897), and the parallel 78,000-unit procurement figure (14142) was never reconciled with it. Granular Lake Mariner miner counts (13877, 13880) and the 50 MW 2021 figure (21124, 21140) remain unverifiable.

The claim record ends in early 2023, and that leaves a substantial gap. The Lake Mariner Anthropic AI Factory, with 410 MW initially disclosed, a 750 MW campus target, and a $3.5 billion financing target, has no dated capacity or timeline claims behind it here. The only linked claims are 2022 bitcoin-era items, including the 90 MW NYPA allocation (14126, 14110), which falls far short of the power the new targets imply. No permits, interconnection queue entries, or site reporting in this dataset independently support the announced HPC pace. The ten announced sites in Kentucky, Maryland, and upstate New York are likewise uncorroborated.

Over the next 90 days, the key questions are these. Does the $3.5 billion financing close on the stated terms? Do NYPA or NYISO filings show additional allocation or queue positions sized for 750 MW? Does the November quarterly report attach energization dates to the 410 MW tranche? Do any of the announced Kentucky or Maryland sites produce a permit or a utility agreement rather than another press-release placeholder?

TeraWulf's capacity story has been rewritten around the same footprint. The original pitch was 800 MW deployed by 2025 (21125, 21141, 14108), built on Lake Mariner and a nuclear-powered Nautilus site. That target is now marked delayed, and the record contains no claim that formally retires or restates it. Nautilus shows the clearest contraction. It was first framed as 300 MW (14111, 14121, 14130). It then appeared as a 150 MW TeraWulf share of a joint venture (14112), and later as 200 MW of initial capacity with a contingent 100 MW expansion (13716) that slipped past its May 2024 marker. The equipment narrative wobbles the same way: a 58,000-miner delivery commitment repeated across five filings (14247, 14049, 14025, 13978, 13897), all now delayed, sits beside a 78,000-unit procurement figure (14142) that is never reconciled with it.

Near-term operational promises fared better than headline targets. The New York facility going live in Q4 2021 (21138), Pennsylvania ASIC installation by mid-2022 (14122, 14024) and the first 100 MW phase at Lake Mariner by year-end 2022 (14106) are all verified. The pattern is reliable execution on tranches and soft drift on aggregates, with revisions surfacing as new numbers rather than acknowledged slips.

Independent corroboration in this record is thin. The NYPA 90 MW supply arrangement (14126, 14110) and the Susquehanna power agreement (14113) are counterparty contracts, not permits or interconnection queue positions, and no on-the-ground reporting is cited. Two data-hygiene problems also need resolving before readers can track the pivot cleanly. Legacy bitcoin-mining claims are attached to the Lake Mariner Anthropic AI Factory entity. Nautilus is listed in Tompkins County, New York, even though its stated power source is a Pennsylvania nuclear station. The current AI-hosting capacity figures of 410 MW, later 750 MW, and the $3.5 billion financing target exist only at the project level, with no dated claim ids behind them.

Over the next 90 days, watch for three things. First, whether the $3.5 billion Anthropic-facility financing closes and on what terms. Second, whether TeraWulf files dated, MW-specific energization milestones for Lake Mariner that explain the step from 410 to 750 MW. Third, whether NYISO queue entries or local permits appear that match that scale. The Kentucky and Maryland sites (Justified, Muskie, Chesapeake, Morgantown, Hancock, Boyd-Greenup) remain announcements without timelines. Any capacity attached to them should be treated as aspirational until power agreements or permits surface.

TeraWulf's capacity story has scaled in two distinct phases, and the record on file captures only the first. From late 2021 through early 2023, the company held a consistent 800 MW-by-2025 aggregate target (21125, 21141, 14108). That figure was never formally walked back, and it is now marked delayed. Its equipment commitment grew rather than shrank: 58,000 miners (14247, 14049, 13978) became 78,000 units in a parallel disclosure (14142), and every 58,000-unit delivery claim is flagged delayed. The second phase is the AI pivot. Lake Mariner moved from a 90 MW NYPA allocation (14110, 14126) and a 100 MW first phase (14106) to a 410 MW initial and 750 MW campus target for the Anthropic-leased facility, backed by a $3.5 billion financing target. None of the claims supplied here date from that period, so the step-up cannot be tied to specific dated promises.

On dates, the early near-term commitments mostly held. The New York facility came online in Q4 2021 (21138), the Lake Mariner phase one target was met (14106), and Pennsylvania ASIC installation began on schedule (14122, 14024). Nautilus is where the numbers changed without acknowledgement. It was described as 300 MW (14111, 14121, 14130) with a 150 MW TeraWulf share (14112). It was then restated as 200 MW initial capacity (13715) with a 33.33 percent stake (13714), and the 100 MW expansion slipped past its May 2024 marker (13716). The multi-year 2028 zero-carbon target (14245, 13846, 13411) has simply been repeated.

Independent evidence is absent from this record. No permits, NYISO or PJM queue positions, or site reporting are attached to any claim, so the announced pace rests on company disclosure alone. Some bitcoin-era claims (14140, 14105) are retroactively tagged to the Anthropic project, which can overstate how continuous that project's history is. Nautilus is listed as announced in Tompkins County even though its claims describe a Pennsylvania site with ASICs already installed.

Over the next 90 days, watch for four things. First, whether the $3.5 billion Lake Mariner financing closes on stated terms. Second, whether NYISO filings or NYPA amendments appear that support capacity above the roughly 100 MW historically documented. Third, whether the Q3 report gives energized megawatts for the Anthropic halls as opposed to contracted ones. Fourth, whether any of the Kentucky and Maryland sites (Justified, Muskie, Hancock, Boyd-Greenup, Chesapeake/Morgantown) produce a permit or interconnection filing that would move them beyond announced status.

TeraWulf's capacity story has been rewritten rather than simply missed. The founding target of 800 MW deployed by 2025 (21125, 21141, 14108) is now marked delayed. The company never published a revised number for it. Instead, the narrative moved to a high-performance computing tenant at Lake Mariner, where the figure has grown from 410 MW to a 750 MW campus target, alongside a $3.5 billion financing goal. At Nautilus, the number shrank. Early filings described 300 MW of nuclear-adjacent capacity (14111, 14121, 14130). By August 2022 that had become 200 MW of initial electrical capacity (13715), plus a 100 MW expansion (13716) that is itself delayed. The 150 MW joint-venture framing (14112) sits uneasily between those two figures.

The dates show a mixed record. The near-term milestones the company controlled directly mostly held: New York operational in Q4 2021 (21138), Pennsylvania ASIC installation by mid-2022 (14122, 14024), and Lake Mariner's first 100 MW phase by year-end 2022 (14106). The supply-chain promises did not. The 58,000-miner delivery commitment was repeated in five filings (14247, 14049, 14025, 13978, 13897) and is delayed in every one. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. Claim 13880 describes 3,300 miners online by Q3 2022 but carries a July 1 deadline, which is an internal inconsistency worth correcting at the source. The 100 percent zero-carbon target for 2028 has been restated without change since 2022 (14245, 14026, 13846, 13826, 13439).

This record contains no independent corroboration. It has no permit filings, no NYISO or PJM interconnection queue positions, and no on-the-ground reporting. The only grid-side anchor is the 90 MW NYPA allocation (14110, 14126), which is an order of magnitude below the 750 MW Lake Mariner target. The Kentucky and Maryland sites, including Hancock, Boyd-Greenup, Muskie, Justified, and Morgantown, carry no disclosed capacity or cost at all. The claims supplied here also end in early 2023, so the 410-to-750 MW escalation could not be traced claim by claim.

Over the next 90 days, the first thing to watch is whether the $3.5 billion Lake Mariner financing closes, and on what terms. The second is whether new interconnection or load-study filings appear that could support anything near 750 MW. The third is whether TeraWulf assigns megawatt figures and in-service dates to the announced Kentucky and Maryland sites. Until it does, those sites should be treated as optionality, not as pipeline.

The record supplied here ends partway through early 2023 (the list cuts off at claim 13410), so this read covers the bitcoin-mining era in detail. The current AI-hosting cycle, including the 410 MW to 750 MW Lake Mariner campus and its $3.5 billion financing target, appears only in the project table and has no supporting claims. That gap matters, because the most consequential capacity story in the portfolio is the one this ledger cannot yet test.

The headline capacity target was set early and has not been met. The 800 MW by 2025 figure was repeated across 21125, 21141 and 14108, and 21125 is now marked delayed. Other numbers shrank quietly. Nautilus was framed as 300 MW of nuclear-powered access (14111, 14130, 14121). By August 2022 it was described as 200 MW of initial capacity (13715), with the remaining 100 MW recast as a contingent expansion that later slipped (13716). TeraWulf's share was also smaller than the facility-level figures suggested: 150 MW (14112), or a one-third stake (13714). Equipment figures drifted the same way. The 58,000-unit delivery commitment was restated at least five times (14247, 14049, 14025, 13978, 13897) and is uniformly delayed, while a 78,000-unit procurement figure (14142) briefly appeared alongside it without reconciliation.

Some near-term dates did hold. The New York facility was operational in Q4 2021 (21138), the first Lake Mariner phase was online by year-end 2022 (14106), and Pennsylvania ASIC installation began on schedule (14122, 14024). The pattern is near-term site milestones delivered and aggregate targets missed, with the misses absorbed through restatement rather than explicit revision.

No permit filings, interconnection-queue entries or on-the-ground reporting appear in this record, so the announced pace rests entirely on company statements. There is also a lineage problem. Bitcoin-era Lake Mariner claims (14105, 14106, 14126) are tagged to the Anthropic AI Factory, and Pennsylvania/Susquehanna Nautilus claims sit under a Tompkins County, New York project entry. Both mappings should be corrected before any trajectory scoring.

Over the next 90 days, watch four things. First, the Q3 2026 results, for whether the Anthropic facility is still described at 410 MW or at the 750 MW campus figure. Second, whether the $3.5 billion financing has closed. Third, NYISO queue or utility filings that would independently support load beyond Lake Mariner's original 90 MW NYPA allocation (14110). Fourth, any permitting activity at the Kentucky and Maryland sites, which so far exist only as announcements.

The supplied claim log stops in February 2023, so this read covers TeraWulf's bitcoin-era commitments in detail and can describe the current AI-hosting phase only through the project-level summary. That gap matters, because the headline capacity numbers have changed in kind, not just in size. The original plan was 800 MW of mining load by 2025 (21125, 21141, 14108). Today the stated goal is a Lake Mariner campus for Anthropic with 410 MW at first disclosure and a 750 MW target after that, backed by a $3.5 billion financing goal. The site's anchor has moved from a 90 MW NYPA allocation (14110, 14126) to a campus roughly eight times that size. The record contains no claim that explains the step-up or ties it to new power procurement.

Nautilus is where capacity shrank most clearly. It was first presented as 300 MW of nuclear-powered capacity (14130, 14121, 14111) with TeraWulf holding a 150 MW share (14112). By August 2022 it had been restated as 200 MW initial capacity, with 100 MW of expansion as an option (13715, 13716) and a 33.33 percent equity stake (13714). That reduced TeraWulf's attributable megawatts, and the record shows no reconciliation with the earlier figures. The expansion option is now marked delayed.

The early promises that held were the narrow ones: New York going live in Q4 2021 (21138), Lake Mariner's first 100 MW phase (14106), and the start of ASIC installation in Pennsylvania (14122, 14024). The broad ones slipped. Delivery of 58,000 miners in 2022 is delayed across five restatements (14247, 14049, 14025, 13978, 13897). A one-off claim of 78,000 units (14142) sits alongside them with no explanation. The 800 MW 2025 target is now marked delayed, and the log shows no point where it was formally withdrawn.

The record holds no independent corroboration of the announced pace: no permits, no NYISO or PJM queue positions, and no on-the-ground reporting. Seven Kentucky, Maryland and New York sites are listed as announced with no capacity or cost attached, and Abernathy has already been sold. The project count is therefore a weak proxy for deliverable megawatts.

There are four things to watch over the next 90 days. First, whether the $3.5 billion Lake Mariner financing closes and on what terms. Second, whether NYISO filings show interconnection capacity consistent with 750 MW rather than the legacy allocation. Third, whether Niagara County permitting records match the construction status the company claims. Fourth, whether the next earnings cycle states, rather than implies, that the 800 MW bitcoin target has been retired.

TeraWulf's capacity story has been rewritten more than once, and the record provided here ends in early 2023. Claims after claim 13409 were not available for this read, so the AI-era figures are assessed only from the project summaries. The original thesis was 800 MW of bitcoin mining deployed by year-end 2025 (21125, 21141, 14108). That target is now flagged as delayed, and the company never formally retired it. It was overtaken by a different business model. Nautilus shows the same drift on a smaller scale. It was first pitched as 300 MW of nuclear-powered capacity (14130, 14111, 14121). Within five months that had become 200 MW initial plus a 100 MW option (13715, 13716), and the option's May 2024 date is marked delayed. TeraWulf's position also moved from a Talen joint venture framed at 150 MW (14112) to a one-third equity stake (13714). The latest framing is a Lake Mariner campus for Anthropic, disclosed at 410 MW and later targeted at 750 MW, with a $3.5 billion financing target. That is the largest number TeraWulf has attached to any single site, and it grew between disclosures.

Dates have held best where TeraWulf controlled the site and power. The New York facility came online in Q4 2021 (21138), and Lake Mariner's first 100 MW phase was verified for 2022 (14106). Pennsylvania ASIC installation was also verified (14122, 14024). The repeated 58,000-miner delivery commitment for 2022 (14247, 14049, 14025, 13978, 13897) is marked delayed in every iteration, and the record shows no revised date. The 2021 50 MW figure (21124, 21140) remains unverifiable.

Independent corroboration is thin. Apart from the Lake Mariner and Pennsylvania verifications, nothing in the record points to permits, NYISO or PJM queue positions, or site reporting. That gap matters most for the eight announced sites in Kentucky, Maryland and New York, none of which carries a capacity or cost figure. The Abernathy campus has already been sold, which is a reminder that announced sites can leave the portfolio before delivery.

Over the next 90 days, the main test is whether the $3.5 billion Lake Mariner financing closes on the stated terms. After that, watch for documented energization of tranches toward 410 MW and interconnection filings that would support 750 MW. Also watch whether the Morgantown, Hawesville or Boyd-Greenup sites gain megawatt figures backed by utility agreements or permit filings rather than headlines.

Note first that the claim record supplied here ends partway through February 2023. The pivot that now defines TeraWulf, Lake Mariner becoming an Anthropic-leased AI factory sized at 410 MW and then 750 MW against a $3.5 billion financing target, is not covered by any claim id in this excerpt. This read therefore assesses the bitcoin-era delivery record that the HPC story is built on, and treats the newer figures as unsourced until the later claims are reviewed.

The headline target of 800 MW by 2025 was repeated almost word for word across 21125, 21141 and 14108 and is now marked delayed. Underneath it, announced capacity shrank without much comment. Nautilus was pitched at 300 MW of nuclear-powered access (14111, 14121, 14130), with a 150 MW TeraWulf share through the Talen joint venture (14112). By August 2022 it had become a 200 MW initial facility (13715), with the extra 100 MW reframed as a possible expansion (13716) that has since slipped past its May 2024 date. Lake Mariner moved the other way: a 90 MW NYPA allocation (14110, 14126) supported a verified 100 MW first phase by the end of 2022 (14106).

On timing, the near-term site dates mostly held. New York went live in Q4 2021 (21138), and Pennsylvania ASIC installation began on schedule (14122, 14024). Equipment was the weak link. The 58,000-miner delivery commitment was restated five times (14247, 14049, 14025, 13978, 13897) and is flagged delayed in every one, while a 78,000-unit procurement figure (14142) was never reconciled against it. The two 2021 50 MW claims (21124, 21140) are still not_yet_verifiable years after their deadlines. That points to gaps in our tracking rather than in the company's delivery, and it should be closed.

No permits, interconnection-queue entries, or on-the-ground reporting are attached to any claim here. That matters most for the eight announced sites in Kentucky, Maryland and upstate New York, none of which has a disclosed capacity or cost.

Over the next 90 days, watch for four things. First, whether the $3.5 billion Anthropic facility financing closes and on what terms. Second, whether the Q3 2026 results restate the 750 MW campus figure and attach energization dates to it. Third, whether NYISO large-load or NYPA filings exist that can support load at that scale. Fourth, whether PJM queue activity tied to the Morgantown Generating Station shows that the Maryland site is more than a press-release placeholder.

A caveat first: the claim record supplied here ends in early 2023, partway through an entry. The AI-era figures on the project list (410 MW initially, then 750 MW, plus a $3.5 billion financing target for the Anthropic-leased Lake Mariner build) therefore have no itemized claim ids. This read can only measure the old bitcoin-era promises against what was later recorded.

On capacity, the headline 800 MW-by-2025 target was repeated without change across 21125, 21141 and 14108. The record now marks 21125 as delayed, and no restated figure ever replaced it. Nautilus shows quiet shrinkage. It was first described as 300 MW (14130, 14111, 14121) and later recast as 200 MW initial with a 100 MW expansion (13715, 13716), and that expansion is itself marked delayed past its May 2024 date. TeraWulf's 33.33 percent stake (13714) means its attributable share was always far smaller than the headline number. The jump from 410 MW to 750 MW at Lake Mariner continues the pattern of announcing larger targets before the earlier ones are met.

On dates, the record is mixed. Near-term site milestones held: Lake Mariner's first phase by year-end 2022 (14106), the Pennsylvania ASIC installation (14122, 14024) and the Q4 2021 New York start (21138) are all verified. Equipment and aggregate-scale promises did not hold. The 58,000-miner delivery commitment is delayed in every restatement (14247, 14049, 14025, 13978, 13897). The 50 MW-in-2021 figure (21124, 21140) and the 3,300-miner Q3 2022 target (13880) were never verified, and neither was revised in the record.

The record contains no independent corroboration: no permits, no NYISO or PJM queue positions, no on-the-ground reporting. Status labels such as under_construction and partially_operational come from the company. The geography of the Nautilus project also conflicts across the record, listed in Tompkins County, New York but described in the claims as a Pennsylvania Susquehanna site.

Over the next 90 days, watch for four things. First, whether the $3.5 billion Anthropic financing actually closes and on what tranche schedule. Second, whether NYISO interconnection filings support load beyond Lake Mariner's existing NYPA-backed 90 MW (14110, 14126), which is necessary for 750 MW to be credible. Third, dated energization milestones for the Anthropic halls. Fourth, whether any of the capacity-less Kentucky and Maryland announcements, Morgantown Generating Station in particular, produce state permitting or utility filings.

TeraWulf's record splits into two eras, and the ledger documents only the first. From October 2021 through February 2023 the company repeated an 800 MW aggregate target for 2025 (21125, 21141, 14108); that target is now flagged delayed. Asset-level figures shrank as the details firmed up. Nautilus was pitched at 300 MW of nuclear-powered capacity (14130, 14121, 14111). By August 2022 it had become a 200 MW initial build (13715) in which TeraWulf held a one-third stake (13714), with the remaining 100 MW pushed to an expansion dated May 2024 that is also flagged delayed (13716). Lake Mariner moved from a 90 MW NYPA allocation (14110, 14126) to a 50 MW first building (13881). The 100 MW phase-one goal (14106) is marked verified, which is the strongest point in the early record.

The dates show a consistent pattern. Near-term site milestones generally held: the New York facility was operational in Q4 2021 (21138), and Pennsylvania ASIC installation began on its Q2 2022 schedule (14122, 14024). Equipment and fleet-scale promises did not hold. The 58,000-miner delivery commitment was restated at least five times during 2022 (14247, 14049, 14025, 13978, 13897) and ended up delayed. None of those restatements acknowledged a revised schedule. The 78,000-unit procurement figure (14142) and the 50 MW online-in-2021 claim (21124, 21140) were never independently confirmed.

The record contains no permits, interconnection-queue filings, or on-the-ground reporting for any project. The verifications above come from company disclosures, not third-party evidence. There is also a data-integrity problem. Bitcoin-era Lake Mariner claims are tagged to the Lake Mariner Anthropic AI Factory, and Nautilus is listed in Tompkins County, New York, even though the claims describe the Susquehanna site in Pennsylvania. The project list includes the AI factory's move from 410 MW to 750 MW and a $3.5 billion financing target, plus seven Kentucky, Maryland and New York sites in announced status. None of these has dated claims on record, so the company's current narrative cannot be checked against its own history here.

Over the next 90 days, four items matter. First, whether the $3.5 billion Anthropic-facility financing closes, and on what terms. Second, whether TeraWulf reconciles the 410 MW and 750 MW figures with a phased energization schedule. Third, whether NYISO queue entries or NYPA allocations appear that are large enough to support that load. Fourth, whether any of the announced Kentucky or Maryland sites produce a zoning, air-permit, or utility filing. If the pre-2023 pattern repeats, construction-start dates are likely to hold better than headline capacity totals.

TeraWulf's capacity story has changed shape more than it has changed scale. The 800 MW by 2025 target was repeated with little change across the 2021 merger materials and the 2022 annual filing (21125, 21141, 14108), as was the 100 percent zero-carbon goal for 2028 (14245, 13846, 13439). Those numbers then stopped appearing in the portion of the claim record supplied here, which ends in February 2023. The project-level record shows the replacement. Lake Mariner, originally framed around a 90 MW NYPA allocation (14110, 14126) and a 50 MW first mining building (13881), is now carried as a 410 MW initial and 750 MW campus target for an Anthropic-leased facility. That is a pivot from bitcoin hashrate to leased HPC capacity, not a continuation of the original plan. Nautilus also drifted. Early filings cited 300 MW of nuclear-adjacent access (14111, 14121, 14130). The later JV description gives 200 MW initial with a 100 MW expansion (13715, 13716), and that expansion is marked delayed. TeraWulf's own share was framed as 150 MW (14112) and later as a 33.33 percent equity stake (13714).

On dates, the record is mixed. The verified items are narrow ones: New York operational in Q4 2021 (21138), Pennsylvania ASIC installation by mid-2022 (14122, 14024), and Lake Mariner's first 100 MW phase by year-end 2022 (14106). The 58,000-miner delivery commitment was restated at least five times through mid-2022 (14247, 14049, 14025, 13978, 13897) and is marked delayed in every instance. We found no acknowledgement in the record of how that slip resolved. The 800 MW deadline passed on 2025-12-31 and is still logged as not yet verifiable, and that status should now be resolved one way or the other. Claim 13880 also sets a July 1, 2022 deadline for 3,300 miners while 13877 gives August 1 for 3,000, an internal inconsistency that was never reconciled.

Independent corroboration is thin. The supplied record contains no permit, NYISO or PJM queue, or on-the-ground evidence for the 750 MW Lake Mariner target, nor for the Kentucky and Maryland sites, which are all at announced status. Nautilus is mapped to Tompkins County in the project list, yet its claims describe the Susquehanna site in Pennsylvania, and that tagging should be corrected.

Over the next 90 days, watch for four things: Q3 2026 results and whether they restate Lake Mariner energization milestones in megawatts actually delivered rather than leased; the close and terms of the $3.5 billion financing; NYISO interconnection filings covering load above the legacy NYPA block; and any Maryland PSC or Kentucky siting filings that would move Morgantown, Hawesville or Grayson beyond announcement.

The record supplied here stops in early 2023 (it breaks off at claim 13409), so it covers TeraWulf's bitcoin-mining years. It does not include the claims behind the later HPC pivot, such as the 410 MW to 750 MW Lake Mariner campus or the $3.5 billion Anthropic financing target listed in the project table. Within that window, the trajectory is one of quiet downsizing. The headline target of 800 MW by 2025 (21125, 21141, 14108) was repeated through 2022, but the project-level figures beneath it shrank. Nautilus went from 300 MW (14130, 14111, 14121) to 200 MW initial capacity, with another 100 MW left contingent (13715, 13716). TeraWulf's share fell from a 150 MW half of the joint venture (14112) to a 33.33 percent equity stake (13714). At Lake Mariner, a 100 MW first phase by year-end 2022 (14106) became a 50 MW first building (13881) with no date attached. Equipment claims drifted the same way: 58,000 miners slated for 2022 delivery are all marked delayed (14247, 14049, 14025, 13978, 13897), while a parallel 78,000-unit procurement figure (14142) was never reconciled with them.

Dates followed a familiar pattern. The Pennsylvania facility was set for Q2 2022 (21139), then restated as mid-year 2022 (14107), and the June 30, 2022 ASIC installation date was repeated (14122, 14024) without any acknowledgement that it was slipping. Only the corporate milestones, meaning the IKONICS merger and the Q4 2021 New York start (21138, 21136), were actually verified. Every capacity deadline remains not_yet_verifiable, even though the 2025 date for the 800 MW pledge has now passed. That is a resolution gap the tracker should close.

The record contains no independent corroboration: no NYISO or PJM queue positions, no building permits, no site reporting. The data also has hygiene problems. Nautilus is listed in Tompkins County, New York, while the claims tie it to Susquehanna in Pennsylvania. Legacy 2022 mining claims are mapped onto the Lake Mariner Anthropic AI Factory. Both issues distort any project-level read.

Three things to watch over the next 90 days. First, the Q3 2026 results should say whether the $3.5 billion Lake Mariner financing closed and when the first tranche of the 410 MW is energized. Second, NYISO interconnection filings and Niagara County permits should show whether the 750 MW campus target has grid backing. Third, the Kentucky and Maryland sites still marked announced, including Justified, Muskie, Chesapeake, Morgantown, Hancock and Boyd-Greenup, need a first permit or queue entry before they deserve any weight. Given the 2022 history of shrinking capacity without disclosure, investors should look for stated MW being trimmed between filings.

The record available for this read is truncated in early 2023, so it carries none of the claims behind the Lake Mariner Anthropic AI Factory's 410 MW initial disclosure, its 750 MW campus target, or the $3.5 billion financing headline of July 2026. What it does show is a pattern that should shape how those figures are read. The headline scaling target, 800 MW deployed by 2025, appears in claims 21125, 21141 and 14108 and was never visibly restated, revised or retired before its deadline lapsed. Its status is still listed as not_yet_verifiable, which in practice means the deadline passed without the company accounting for it. Nautilus shows the same drift at project level. It was introduced at 300 MW (14130, 14111, 14121), quietly reframed as a 200 MW initial build with a 100 MW expansion option (13715, 13716), and the 33.33 percent stake in 13714 implies attributable capacity well below the original framing. None of these revisions was acknowledged as such. Dated promises fared poorly. The 58,000-miner delivery commitment is marked delayed across five separate restatements (14247, 14049, 14025, 13978, 13897). A same-day 78,000-unit procurement figure in 14142 conflicts with it. Lake Mariner's 100 MW first phase by end-2022 (14106) sits uneasily against a 50 MW first building (13881) and a 90 MW NYPA supply (14110, 14126). The only verified operational milestone is the generic New York facility going live in Q4 2021 (21138). The paired claim of 50 MW online in 2021 (21124, 21140) remains unconfirmed. The record contains no independent corroboration of any kind: no permits, no NYISO or PJM queue positions, no local reporting. The project registry also has mapping problems that weaken attribution. Early bitcoin-mining claims are tagged to the Anthropic AI Factory, and Nautilus is listed in Tompkins County although its claims describe a Susquehanna, Pennsylvania site. Over the next 90 days, watch the Q3 2026 10-Q for whether the 410 MW and 750 MW figures are reconciled, given dates and tied to energization milestones. Watch also whether the $3.5 billion financing actually closes or remains a target. Check the NYISO queue and Niagara County filings for interconnection and permits that would show load growth beyond the legacy NYPA allocation. Finally, see whether any of the roughly ten announced sites in Kentucky, Maryland and upstate New York acquire a dated capacity figure or quietly drop out of disclosure.

The supplied claim record stops in February 2023 (13408, truncated). The 2026 Lake Mariner Anthropic figures in the project list therefore have no claim ids here, so this read covers the bitcoin-era trajectory that the AI pivot inherits. The headline ambition was steady and large. A target of 800 MW by year-end 2025 appears unchanged across 21125, 21141 and 14108. Project-level numbers, by contrast, narrowed quietly. Nautilus was first framed as 300 MW of nuclear-powered capacity (14121, 14130, 14111). By August 2022 that became 200 MW initial capacity with a contingent 100 MW expansion (13715, 13716), and TeraWulf's position moved from a 150 MW joint-venture slice (14112) to a 33.33 percent equity stake funded by 20,000 miners (13714, 13713). Lake Mariner's scale was also inconsistent within the same month: a 90 MW NYPA supply (14110, 14126) against a 100 MW first phase (14106), later scoped as a 50 MW first building (13881). Fleet size was cited as both 58,000 (14123) and 78,000 units (14142). Dates show a pattern of restatement without revision. The 58,000-miner 2022 delivery commitment repeats verbatim from March through June 2022 (14247, 14049, 14025, 13978, 13897) and is now marked delayed, with no revised date in the record. The New York Q4 2021 start held (21138), but the 50 MW-in-2021 claim (21124, 21140) and the Pennsylvania Q2 2022 ASIC start (21139, 14122, 14024) remain unresolved. The 800 MW-by-2025 deadline has now passed and should be scored rather than left as not yet verifiable. Independent corroboration in the record is thin. The NYPA allocation and the Susquehanna supply agreement (14113, 14129) are counterparty contracts, but there are no permits, NYISO or PJM queue positions, or site reporting. The Nautilus entry is listed in Tompkins County, New York, while every claim places it at Susquehanna in Pennsylvania, which suggests an entity-mapping error. Over the next 90 days, the main question is whether the $3.5 billion Anthropic-facility financing closes on stated terms. A second question is whether 410 MW or 750 MW becomes the figure tied to a dated energization schedule. Before accepting any Lake Mariner ramp beyond the original 90 MW allocation, check NYISO interconnection filings and any expanded NYPA allocation. Q3 2026 results should also show whether the Kentucky and Maryland sites (Justified, Muskie, Chesapeake, Hancock, Boyd-Greenup) carry dates and megawatts or remain undated announcements.

TeraWulf's capacity record is anchored by one headline number repeated without revision: 800 MW deployed by 2025 (21125, 21141, 14108). The project-level figures beneath it shrank quietly. Nautilus was presented as a 300 MW nuclear-powered facility (14111, 14121, 14130). Five months later it became 200 MW of initial capacity, with the missing 100 MW recast as a potential expansion due by May 2024 (13716). TeraWulf's own position moved from a 150 MW joint-venture share (14112) to a 33.33 percent equity stake (13714), which implies roughly 67 MW attributable on the 200 MW base. The company never acknowledged the cut. Lake Mariner moved in the other direction: a 90 MW NYPA allocation (14110, 14126) became a 100 MW first phase targeted for year-end 2022 (14106). Miner procurement was cited as 58,000 units across most filings (14247, 14049, 13897) but as 78,000 in one (14142), and that discrepancy is unexplained.

The early dates slipped. New York was promised operational in Q4 2021 (21138) alongside 50 MW online that year (21124, 21140). By mid-2022 the stated milestone had become 3,000 to 3,300 miners online between July and August 2022 (13880, 13877), well short of 50 MW. Pennsylvania ASIC installation was held at Q2 2022 across several filings (21139, 14122, 14024), with no closing confirmation in this record. The 800 MW-by-2025 deadline has now passed, and the record contains no reconciliation against it.

There is no independent corroboration in what was supplied. Every capacity and timeline claim is either self-stated or marked not yet verifiable. No permit, NYISO queue position, NYPA allocation filing, or site reporting is attached. The project registry also needs correcting. Nautilus is located in Tompkins County, New York, while its claims reference Susquehanna power (14129), which is in Pennsylvania. Bitcoin-era Lake Mariner claims are mapped to the 2026 Anthropic AI Factory, which conflates two distinct delivery stories. Finally, the claims supplied end in early 2023, so the 410 MW to 750 MW campus escalation and the $3.5 billion financing headline have no claim ids here and cannot yet be assessed against stated dates.

Over the next 90 days, watch four things. First, whether the $3.5 billion Lake Mariner financing closes on the announced terms. Second, whether 750 MW is described as contracted load or as an aspirational campus figure. Third, whether NYISO queue entries or NYPA allocations appear that are large enough to support anything beyond the original roughly 100 MW footprint. Fourth, whether the Kentucky and Maryland announcements, including Morgantown Generating Station, produce county permits or utility filings rather than further press releases.

TeraWulf's capacity story has expanded faster than its delivered megawatts. Early filings anchored Lake Mariner to a 90 MW NYPA supply (14126, 14110), a 100 MW first phase due by end-2022 (14106) and a 50 MW first building (13881). Nautilus was pitched at 300 MW of nuclear-powered access (14121, 14130, 14111). By August 2022 that figure had been quietly reframed as 200 MW of initial capacity plus a 100 MW expansion (13715, 13716), with no disclosed reason for the change. The Talen joint venture was described at 150 MW (14112) alongside a 33.33 percent equity stake (13714). Those figures are hard to reconcile into a single net-to-TeraWulf number. The 800 MW aggregate target for 2025 (14108) is now marked delayed. Today the project table describes Lake Mariner at 410 MW, then 750 MW, with a $3.5 billion financing target. No claim ids in this record carry those figures, so the jump from 90 MW to 750 MW cannot be traced through the company's own statements here.

Timelines have consistently slipped. The 3,300 miners due by July 2022 (13880), the 3,000 S19 XPs due by August (13877) and the mid-2022 Nautilus ASIC installation (14122, 14024) are all delayed. So is the 58,000-miner delivery commitment, which was repeated almost verbatim across five filings (14247, 14049, 14025, 13978, 13897). It coexisted with a separate 78,000-unit procurement figure (14142) that was never reconciled. Mining commencement later acquired a retroactive March 2022 date (13409) that earlier filings did not specify. The 2028 zero-carbon target is the one promise restated unchanged, from 14245 through 13343, and it is easy to keep because it is not yet testable.

The supplied record contains no permits, interconnection-queue entries or site reporting. Twelve of the thirteen listed sites, from Abernathy to Boyd-Greenup, carry no capacity or cost at all. Nautilus is listed in Tompkins County, New York, while its claims describe a Pennsylvania facility powered by Susquehanna, a mismatch worth correcting before publication. The announced pace is therefore unsupported rather than contradicted.

Over the next 90 days, there are four things to watch. The first is whether the $3.5 billion Anthropic-linked financing actually closes. The second is whether NYISO queue or NYPA filings support load at Lake Mariner well beyond the original 90 MW. The third is whether the next 10-Q assigns separate dates to the 410 MW and 750 MW tranches. The fourth is whether any of the Kentucky, Maryland or Texas sites gains a capacity figure, a power agreement or a permit filing.

The record supplied here breaks off partway through claim 13345 in February 2023, so this read covers the bitcoin-mining origins of the portfolio and not the later HPC pivot. Neither the 410 MW and 750 MW figures attached to the Lake Mariner Anthropic AI Factory nor the $3.5 billion financing headline is backed by a claim id in the material provided, so they cannot be assessed yet. Within the visible window, announced capacity was reframed rather than plainly shrunk. Nautilus was first presented as 300 MW of nuclear-powered access (14121, 14130, 14111) and then as 200 MW of initial capacity plus a 100 MW expansion (13715, 13716), with the expansion later marked delayed. TeraWulf's share was described as a Talen joint venture worth 150 MW (14112) before settling at a 33.33 percent equity stake (13714). The record also maps Nautilus to Tompkins County, New York, while the claims describe a Pennsylvania facility (14122, 14141), which is a project-tagging inconsistency worth correcting. At Lake Mariner, a 90 MW NYPA supply (14126, 14110) sat beneath a 100 MW first-phase target for year-end 2022 (14106), and the first building was cited at 50 MW (13881). The 800 MW aggregate target for 2025 (14108) is marked delayed. Timelines slipped often and mostly without acknowledgement. The Pennsylvania ASIC start (14122, 14024), the 3,300-miner Lake Mariner milestone (13880), the first S19 XP batch (13877) and the 58,000-miner delivery commitment (14247, 14049, 13978, 13897) all moved without restated dates, while a parallel 78,000-unit procurement figure (14142) was never reconciled with the 58,000 number. The one commitment that stayed constant is the 100 percent zero-carbon target for 2028 (14245, 13439, 13411, 13343), and none of the visible claims measure progress toward it beyond a 90 percent figure (13847). No permits, interconnection queue entries or site reporting appear in this record, so the announced pace is supported only by the company's own statements. Over the next 90 days, the key questions are whether the Anthropic-leased financing closes at the headline size, whether NYPA or NYISO filings support a Lake Mariner load well above the 90 MW contracted in 2022, whether the 750 MW campus target comes with dated phases, and whether the post-2023 claims, once loaded, show the same pattern of slipping without restatement.

TeraWulf's capacity story has changed character more than it has grown in a straight line. The 2022 filings framed an 800 MW aggregate target by end-2025 (14108), anchored by a 300 MW Nautilus position (14121, 14130, 14111) and a 90 MW NYPA allocation at Lake Mariner (14110, 14126). By August 2022, Nautilus had been recast as 200 MW initial plus a 100 MW expansion option (13715, 13716), and TeraWulf held a 33.33 percent equity stake (13714). That stake implies roughly 67 MW attributable, against the 150 MW joint-venture figure cited five months earlier (14112). The headline number held, but the share TeraWulf owned shrank, and the filings did not clearly reconcile the two. The current record moves Lake Mariner to an entirely different scale: 410 MW at first disclosure, then a 750 MW campus target, with a $3.5 billion financing target for the Anthropic-leased facility. That is roughly eight times the original NYPA allocation, and nothing in the supplied claims shows the power contracts behind it.

On dates, the 2022 cohort slipped almost uniformly. The 58,000-miner delivery commitment was repeated at least six times (14247, 14123, 14049, 14025, 13978, 13897) and is marked delayed each time. The Pennsylvania ASIC start (14122, 14024), the 3,300 and 3,000 unit Lake Mariner milestones (13880, 13877), and the 100 MW first phase by end-2022 (14106) all missed. The 2025 aggregate target (14108) is also delayed. Repetition without revised dates looks more like unacknowledged slippage than active re-guidance.

The supplied record contains no independent corroboration: no permits, no NYISO or PJM queue positions, no on-the-ground reporting. The project list also places Nautilus in Tompkins County, New York, while the claims tie it to Susquehanna in Pennsylvania (14129, 14141), which suggests a data-mapping error worth correcting. Eight Kentucky, Maryland and Texas sites carry no capacity or cost at all. The claim list is truncated after February 2023, so the 410-to-750 MW step-up cannot be traced claim by claim here.

Over the next 90 days, watch whether the $3.5 billion financing closes on stated terms. Watch for interconnection or NYPA filings that support load well beyond 90 MW at Lake Mariner, and for the company to say clearly whether 410 MW or 750 MW is the contracted figure. Finally, watch whether any of the capacity-blank sites receive a sized filing.

TeraWulf's capacity story has scaled in two very different ways. In the crypto era, figures moved down quietly. Nautilus was first presented as 300 MW of nuclear-powered capacity (14121, 14130, 14111), while the same filing described a 150 MW Talen joint venture (14112). By August 2022 it had become 200 MW initial with a 100 MW expansion (13715, 13716), and that expansion's May 2024 target is now marked delayed. Lake Mariner's 100 MW first phase by end-2022 (14106) sat on a 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881). That framing outran the contracted power from the start. The 800 MW aggregate target for 2025 (14108) was missed. Today's Lake Mariner Anthropic figures, 410 MW and then 750 MW against a $3.5 billion financing target, are another step up in scale. The supplied claim record breaks off in February 2023, though, so those numbers cannot be tied to dated claim ids here.

Dates have slipped more than they have held. Every hardware and installation milestone from 2022 is now marked delayed: the 58,000-miner delivery commitment (14247, 14049, 14025, 13978, 13897), the 3,300 and 3,000-unit Lake Mariner ramps (13880, 13877), and the June 2022 Pennsylvania ASIC start (14122, 14024). The 58,000 figure sits beside a 78,000-unit procurement claim (14142) with no reconciliation. The February 2023 filings restate commencement and custody dates retroactively (13409, 13412, 13414) rather than acknowledging the earlier slippage. The only promise repeated without drift is 100 percent zero-carbon energy by 2028 (14245, 14026, 13846, 13439, 13343), and it remains untested.

The record contains no independent corroboration. There are no permits, NYISO or ERCOT queue positions, or site reporting for Lake Mariner's expansion. The same gap covers the twelve announced sites from Abernathy to Boyd-Greenup, all of which carry no disclosed capacity or cost. Going from 90 MW of NYPA power to 750 MW would need new allocations or transmission, and that should leave a public trail.

Over the next 90 days, watch for four things. First, whether the $3.5 billion Anthropic financing closes and on what terms. Second, any NYPA allocation increase or NYISO interconnection filing that supports load beyond the original 90 MW. Third, the first capacity or cost disclosures for Abernathy, Cayuga, and the Morgantown Generating Station, including any Maryland PSC docket activity. Fourth, whether quarterly reporting restates the 410 MW figure or quietly retires it in favor of the 750 MW campus number.

The claim record supplied here ends abruptly in February 2023, so this read covers the bitcoin-mining era in detail and the AI-campus pivot only through the project roster. That gap matters. The 410 MW initial and 750 MW campus figures for the Lake Mariner Anthropic facility, and the $3.5 billion financing target, appear in the project summary without claim ids in this log. They cannot yet be tested against the earlier promises.

On capacity, the early trajectory narrowed TeraWulf's attributable footprint while the headline numbers held steady. Nautilus was pitched at 300 MW (14121, 14130, 14111), with TeraWulf's share framed as 150 MW (14112). By August 2022 that had become a 200 MW initial build plus a 100 MW expansion (13715, 13716), with a one-third equity stake (13714). At Lake Mariner, a 90 MW NYPA supply (14126, 14110) sat beside a 100 MW year-end 2022 first phase (14106), and the first building was only 50 MW (13881). The 800 MW aggregate target for 2025 (14108) is marked delayed. The jump to 750 MW at Lake Mariner is therefore a step-change, not a continuation of demonstrated pace.

On dates, slippage is pervasive and rarely acknowledged. The 58,000-miner 2022 delivery commitment was restated almost verbatim across at least five filings (14247, 14049, 14025, 13978, 13897) and ended up delayed. It also coexisted with a 78,000-unit procurement figure (14142) that was never reconciled. Nautilus ASIC installation by mid-2022 (14122, 14024), the Lake Mariner Q3 2022 miner counts (13880, 13877) and the 2024 Nautilus expansion (13716) all slipped. Meanwhile the 2028 zero-carbon target was repeated unchanged (14245, 14026, 13846, 13826, 13439, 13411, 13343). The pattern is stable rhetoric over moving delivery dates.

The record contains no permits, interconnection queue entries or on-the-ground reporting for any project. Twelve of the thirteen rostered sites, including Abernathy, Cayuga, the Kentucky sites and Morgantown, carry no capacity or cost at all. Announced pace is therefore unsupported by independent evidence in this dataset.

Over the next 90 days, watch whether the $3.5 billion Lake Mariner financing actually closes and on what drawdown schedule. Watch for NYISO queue or NYPA allocation filings that would support 750 MW rather than the historical 90 MW grid tie, and whether disclosures reconcile 410 MW against 750 MW. Watch also for the first permit or utility filings at any Kentucky, Texas or Maryland site that would move them beyond announcement.

The claim record supplied here ends abruptly in February 2023 (the last entry, 13344, is cut off). That leaves the 2023 to 2026 pivot from bitcoin mining to AI hosting outside the cited evidence, so this read covers the mining-era delivery pattern and flags where the record runs out. The capacity arc so far is one of contraction dressed as continuity. Nautilus was framed at 300 MW of nuclear-powered access (14111, 14121, 14130), with TeraWulf's joint-venture share at 150 MW (14112). By August 2022 it had become a 200 MW initial facility with a one-third equity stake (13713, 13714, 13715), plus a 100 MW expansion that never firmed up (13716). At Lake Mariner, a 100 MW first phase by year-end 2022 (14106) gave way to a 50 MW first building (13881) running on a 90 MW NYPA allocation (14110, 14126). The 800 MW aggregate target for 2025 (14108) is marked delayed. Dates slipped almost uniformly, and the slips were mostly absorbed rather than acknowledged. The 58,000-miner delivery commitment was restated across at least five filings (14247, 14049, 14025, 13978, 13897) with the deadline unchanged and no reconciliation against a separate 78,000-unit procurement figure (14142). Pennsylvania ASIC installation by mid-2022 (14122, 14024) and Lake Mariner's 3,300 miners by Q3 2022 (13880, 13877) both missed, and later filings described progress without restating the original targets. The 2028 zero-carbon goal (14245, 13411, 13343) is the only date held consistently, and it is also the least operationally testable. The record also tags early bitcoin-mining claims to the Lake Mariner Anthropic AI Factory. That lineage should not be read as continuity of delivery. Independent corroboration is absent from what was provided. There are no permits, NYISO or PJM queue entries, or site reporting, so the announced pace rests entirely on company statements. The headline 410 MW, 750 MW and $3.5 billion figures for the Anthropic-leased facility carry no claim ids here and cannot be assessed against a track record. Over the next 90 days, watch for four things. The first is whether the $3.5 billion financing closes on stated terms. The second is whether NYISO queue or NYPA filings show load beyond the historic Lake Mariner allocation. The third is local permitting in Niagara County. The fourth is whether Q3 2026 results restate the 750 MW target with energization dates or quietly re-phase it, as happened with every mining-era milestone.

The record supplied here is cut off in early 2023, so this read covers only TeraWulf's bitcoin-mining build-out. The headline item, the Lake Mariner Anthropic AI Factory with its 410 MW initial and 750 MW campus figures and a $3.5 billion financing target, appears only as a project summary with no claim ids. It cannot be assessed on this evidence. Capacity framing in the 2022 cohort shifted more than it grew. Nautilus was first described as a 300 MW nuclear-powered facility (14130, 14111, 14121), of which TeraWulf's share was 150 MW (14112). By August 2022 it had been restated as 200 MW initial with a possible 100 MW expansion (13715, 13716), and the record flags that expansion as delayed past its May 2024 marker. The aggregate 800 MW by end-2025 goal (14108) is likewise marked delayed. Equipment figures also drifted without reconciliation: most filings cite 58,000 miners (14247, 14049, 14025, 13978, 13897), while one cites 78,000 units against the same year-end deadline (14142). Dates did not hold. Nautilus ASIC installation by mid-2022 (14122, 14024, 14107), Lake Mariner's 100 MW first phase by year-end 2022 (14106), and the summer 2022 miner-online targets (13880, 13877) all carry delayed status. Later filings shifted to undated language about initial buildout completion and the commencement of mining (13332, 13342), so the slippage was absorbed rather than acknowledged. Lake Mariner's NYPA allocation stayed at 90 MW (14110, 14126), below the 100 MW first-phase promise, and the 50 MW first building (13881) suggests the phase was smaller in practice. The 2028 100 percent zero-carbon target is repeated unchanged throughout (14245 through 13343), with 90 percent reported in mid-2022 (13847). No permits, interconnection queue entries, or site reporting are in this record, so the announced pace has no independent corroboration here. Over the next 90 days, the priority is to load the full post-2023 claim set for Lake Mariner and check whether the move from 410 to 750 MW was presented as a revision or as a separate phase. The other items to watch are the close and terms of the $3.5 billion financing, NYISO interconnection filings and NYPA allocation changes that would support anything above the historical 90 MW, and local building permits. The dozen announced sites in Kentucky, Maryland, Texas and New York with no stated capacity should also be watched for first disclosures, given the company's pattern of headline figures followed by quiet restatement.

The claim record supplied here ends mid-entry in February 2023, so the company's AI-era claims are not citable by id. These include the 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion Anthropic financing headline now attached to Lake Mariner. That gap shapes this read: we can judge the baseline TeraWulf built its credibility on, but not the recent claims themselves. On scale, the early pattern is contraction before expansion. Nautilus was presented as a 300 MW nuclear-powered facility (14121, 14111, 14130). By August 2022 it was restated as 200 MW initial capacity, with the remaining 100 MW recast as a potential expansion (13716), and that expansion has since been marked delayed. TeraWulf's exposure was also a one-third equity stake (13714), not the full figure. Lake Mariner was framed around a 90 MW NYPA allocation (14126, 14110) but promised a 100 MW first phase by year-end 2022 (14106), which slipped, and a 50 MW first building (13881). The line from that 90 MW base to a 750 MW campus is the central question the current record cannot answer. On dates, slippage is consistent and mostly unacknowledged. The 58,000-miner delivery commitment was repeated at least six times through mid-2022 (14247, 14049, 14025, 13978, 13897) without revision before being marked delayed. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. Near-term Lake Mariner miner targets (13877, 13880), Pennsylvania ASIC installation (14122, 14024) and the 800 MW aggregate target for 2025 (14108) all missed. The 2028 zero-carbon goal is restated almost verbatim each filing (14245, 13846, 13411, 13343) with no interim milestones. The record contains no independent corroboration, whether permits, NYISO queue positions or site reporting, for any of these figures. Two mapping problems also need correcting: Nautilus is listed in Tompkins County, New York, while its claims describe a Pennsylvania site on the Susquehanna, and the 2022 bitcoin-mining claims (14140, 14127) are filed under the Anthropic project. Over the next 90 days, watch for interconnection or NYPA filings that would support load well above 90 MW at Lake Mariner, and whether the 410 MW tranche gets a dated energization milestone. Watch also whether the $3.5 billion financing actually closes or remains a target. Finally, any of the eleven announced sites with no stated capacity attracting a first capacity or permit disclosure would be meaningful.

The claim record supplied here breaks off mid-entry in February 2023, so this read covers TeraWulf's crypto-era commitments in detail and treats the 2026 Lake Mariner AI campus figures, carried only on the project line and not as cited claims, as context. The early pattern is consistent. Capacity was announced large and then quietly narrowed. Nautilus was described as a 300 MW nuclear-powered facility in March 2022 (14130, 14111, 14121), with TeraWulf holding a 50/50, 150 MW share (14112). By August 2022 it was a 200 MW initial build (13715) with 100 MW framed as optional expansion (13716), and TeraWulf's stake was 33.33 percent (13714). The record shows no claim acknowledging either reduction. The fleet figures also diverge: 58,000 miners appears across several filings (14247, 14049, 14025, 13978, 13897), while 78,000 appears in another (14142), and every dated version is marked delayed. Dates have slipped almost uniformly. Lake Mariner's 100 MW first phase by year-end 2022 (14106), 3,300 miners by Q3 2022 (13880), the initial S19 XP batch by August 2022 (13877), the June 2022 Nautilus ASIC start (14122, 14024), and the aggregate 800 MW by 2025 target (14108) all missed. Lake Mariner's contracted NYPA allocation was 90 MW (14110, 14126), with a 50 MW first building (13881). That baseline matters because the current campus disclosure of 410 MW, rising to a 750 MW target, implies roughly an eightfold scale-up beyond the only power contract on record. The one commitment restated without revision is the 100 percent zero-carbon target for 2028 (14245, 13846, 13411, 13343). It is a goal rather than a delivery milestone. The record contains no independent corroboration: no permits, no NYISO interconnection queue positions, and no on-the-ground reporting. The under_construction status for the Anthropic-leased facility and the twelve other announced sites, none of which has a stated capacity, therefore rest on company assertion alone. Over the next 90 days, watch four things. First, whether the Q3 2026 10-Q restates 410 MW and 750 MW with energization dates or softens them. Second, whether the $3.5 billion financing actually closes, not merely gets announced. Third, whether there are NYISO queue filings or new NYPA allocations that show incremental power above 90 MW at Lake Mariner. Fourth, whether Niagara County site-plan or building permits match the phasing the company describes. Given the history, unacknowledged revisions are the main risk to track.

TeraWulf's capacity story has moved in two directions. The 2022 bitcoin-era claims rested on modest, power-contracted figures: a 90 MW NYPA supply at Lake Mariner (14110, 14126) and a 50 MW first building (13881), alongside an aggregate 800 MW target for 2025 (14108). Nautilus shrank. It was presented as a 300 MW nuclear-powered facility in March 2022 (14121, 14130). By August it had been restated as 200 MW initial with a 100 MW expansion option (13715, 13716), and TeraWulf's own position had become a 33.33 percent stake backed by 20,000 contributed miners (13713, 13714). That was a quiet reduction, not an acknowledged cut. The project record now lists Lake Mariner at 410 MW initially, a 750 MW campus target, and a $3.5 billion financing headline tied to an Anthropic lease. However, the claim history supplied here ends in February 2023, so those figures cannot be traced to dated claims and are unverified in this read.

On dates, the pattern is consistent slippage. Three Lake Mariner promises all carry delayed status: the 100 MW first phase by year-end 2022 (14106), the August 2022 S19 XP batch (13877), and 3,300 miners by Q3 (13880). The same is true of the mid-2022 Pennsylvania ASIC installation (14122, 14024) and the 58,000-miner delivery commitment. That commitment was repeated across at least five filings (14247, 14049, 14025, 13978, 13897) while it was slipping, which is the clearest case in the record of a date revised by silence. The 800 MW by 2025 target (14108) is also marked delayed, and no later claim acknowledges the miss.

No independent evidence in this record supports the announced pace. No permits, NYISO interconnection queue positions, or on-the-ground reports are attached to any Lake Mariner or Nautilus claim. The twelve other announced sites, from Abernathy to Boyd-Greenup, carry no capacity or cost figures at all.

Over the next 90 days, watch for four things. The first is a dated filing that states the 410 MW and 750 MW figures with energization milestones. The second is NYISO queue or NYPA allocation evidence for load well beyond the original 90 MW. The third is county building permits at Lake Mariner. The fourth is whether the $3.5 billion financing closes on its stated terms. A megawatt figure appearing for any of the dormant announced sites would also show where the pipeline is actually headed.

TeraWulf's capacity story has grown much larger, but the claims shown here do little to support the current narrative. In March 2022 the company framed itself around 90 MW of NYPA power at Lake Mariner (14110, 14126) and a 300 MW nuclear-powered Nautilus position with Talen (14111, 14121, 14130). It then rolled both into an 800 MW aggregate target for 2025 (14108). That figure began shrinking almost immediately. By August 2022, Nautilus was described as 200 MW of initial capacity with a 100 MW expansion option (13715, 13716), and TeraWulf's economic interest was a 33.33 percent stake (13714), consistent with the 150 MW joint-venture figure (14112). Today the project record lists Lake Mariner at 410 MW, rising to a 750 MW campus target tied to an Anthropic lease and a $3.5 billion financing target. That is a large jump from the 50 MW first building described in mid-2022 (13881), and none of the supplied claim ids document how it got there.

Near-term dates slipped repeatedly and quietly. The 100 MW first phase at Lake Mariner by year-end 2022 (14106), the 3,000 to 3,300 miner milestones for mid-2022 (13877, 13880), Nautilus ASIC installation by June 2022 (14122, 14024), and the 58,000-unit equipment delivery commitment (14247, 14049, 14025, 13978, 13897) are all marked delayed. In none of these cases does the record show the company acknowledging the revision. The procurement figure also drifted from 58,000 to 78,000 units (14142) without explanation. The 2028 zero-carbon target (14245, 13439, 13343) is the one commitment restated consistently, and it is not yet due.

The independent-evidence column is empty. This dataset contains no permits, NYISO or PJM interconnection queue entries, or site reporting for Lake Mariner or any of the twelve announced sites in Texas, New York, Kentucky and Maryland. Those twelve all lack disclosed capacity or cost. The record also appears to end in February 2023, so the 2024 to 2026 pivot from bitcoin mining to HPC is asserted in project metadata rather than traced through dated claims.

Over the next 90 days, watch for four things. First, whether the $3.5 billion Lake Mariner financing actually closes and on what terms. Second, whether NYISO filings show load or interconnection consistent with 410 MW, let alone 750 MW. Third, whether the next quarterly report gives energization dates for the Anthropic halls. Fourth, whether any of the announced Kentucky, Maryland or Texas sites gain a disclosed megawatt figure or a permit trail. Absent that, they should be read as optionality, not pipeline.

TeraWulf's capacity story has repeatedly been rescaled, and the scope of what was promised has tended to narrow before it widened again. In March 2022 the company set an aggregate target of 800 MW by 2025 (claim 14108), now marked delayed. Nautilus was first presented as 300 MW of nuclear-powered capacity (14130, 14111, 14121). By August 2022 that had become 200 MW of initial capacity plus a 100 MW expansion described only as potential (13715, 13716), with TeraWulf holding a one-third equity stake (13714). Lake Mariner followed a similar pattern. A 100 MW first phase promised by year-end 2022 (14106) became a 50 MW first building (13881), resting on a 90 MW NYPA allocation (14110, 14126). The current framing of the Anthropic facility, 410 MW initially and then a 750 MW campus target backed by a $3.5 billion financing headline, is roughly eight times the power base the record actually documents.

Near-term dates did not hold. The 58,000-miner delivery commitment was restated at least five times through mid-2022 (14247, 14049, 14025, 13978, 13897) and is delayed in every instance. Nautilus ASIC installation by June 2022 (14122, 14024), 3,300 miners at Lake Mariner by Q3 (13880) and 3,000 S19 XPs by August (13877) all slipped. By February 2023 the filings described an "initial buildout" with no date attached (13332) and dropped the 800 MW figure without saying so. The 2028 zero-carbon target, by contrast, is repeated verbatim (14245, 13439, 13826), and it is the only long-dated commitment that has been held constant.

The record contains no independent corroboration of the announced pace: no permits, no interconnection queue positions and no site reporting. The NYPA allocation is a contracted supply figure, not evidence of grid capacity for 410 or 750 MW. The supplied claim history also ends in February 2023, so it does not cover the disclosures behind the AI pivot or the twelve announced sites in Kentucky, Maryland, Texas and New York, none of which carry capacity or cost figures.

Over the next 90 days, the key questions are whether the $3.5 billion financing closes on stated terms and whether NYISO or NYPA filings show load authorization beyond the original 90 MW. Local permitting in Niagara County that matches the 410 MW phase would also be meaningful. Any announced site moving from announced to a stated capacity with a dated milestone would be the first concrete test of the expanded pipeline.

The visible record for TeraWulf ends in February 2023, so this read reflects the bitcoin-mining era that underlies today's AI pitch. It does not cover the 2026 Lake Mariner Anthropic claims themselves. That gap matters. The project table carries a 410 MW initial disclosure, a 750 MW campus target, and a $3.5 billion financing figure, but no claim ids in the extract support them. Those numbers should be treated as unverified headline figures until the underlying claims are surfaced.

On capacity, the early trajectory is one of contraction and restructuring rather than growth. Nautilus was pitched at 300 MW of nuclear-adjacent capacity (14130, 14121, 14111). It was then recast as 200 MW initial with a contingent 100 MW expansion (13715, 13716), and TeraWulf's position was framed variously as a 150 MW joint-venture share (14112) and a one-third equity stake (13714). Lake Mariner's 2022 baseline was a 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881). The jump from that baseline to 410 MW, and then 750 MW, is the central claim needing independent support.

On dates, slippage was the norm. The 100 MW first phase by year-end 2022 (14106), 3,300 miners by Q3 2022 (13880), the August 2022 S19 XP batch (13877), mid-2022 Nautilus mining (14107), June 2022 ASIC installation (14122, 14024), and the 58,000-unit delivery commitment (14247, 14049, 14025, 13978, 13897) are all flagged delayed. The parallel 78,000-unit figure (14142) was never reconciled with the 58,000-unit commitment. The 800 MW aggregate target for 2025 (14108) also lapsed. Its absence from later filings, rather than an explicit revision, is the closest thing here to a quiet retirement. The 2028 zero-carbon target, by contrast, was restated consistently (14245 through 13343).

The record contains no permits, NYISO queue positions, or site reporting. Pace therefore cannot be independently corroborated. Two tagging problems also need fixing. Nautilus is listed in Tompkins County, New York, although the claims tie it to Susquehanna in Pennsylvania. And the 2022 mining claims are filed under the Anthropic facility, which overstates that project's history.

Over the next 90 days, watch the Q3 2026 results for a dated energization schedule for the first Anthropic phase. Also watch whether the $3.5 billion financing closes on stated terms, and whether NYISO or NYPA filings show load approvals beyond the legacy 90 MW. Finally, watch whether any Kentucky, Maryland, or Texas site acquires a disclosed capacity or cost figure.

TeraWulf's claim record splits into two distinct phases, and the gap between them is the most important feature of this file. Every dated claim was made between March 2022 and February 2023 and concerns bitcoin mining. The current project list, however, describes an Anthropic-leased AI facility at Lake Mariner sized at 410 MW and later 750 MW, with a $3.5 billion financing target, plus twelve other announced sites with no recorded capacity. None of that later scaling is tied to a claim id, so the HPC pivot cannot yet be checked against stated dates.

In the mining phase, capacity mostly shrank or was quietly reframed. The 800 MW aggregate target for 2025 (14108) is marked delayed and does not reappear. Nautilus was described at 300 MW in March 2022 (14111, 14121, 14130), but by August 2022 the figure was 200 MW initial (13715), with 100 MW pushed into a contingent expansion (13716) that also slipped. None of the claims on record explains that change. Equipment figures were also inconsistent: 58,000 miners appears repeatedly (14247, 14049, 14025, 13978, 13897), while a 78,000-unit procurement figure (14142) appears alongside it without reconciliation.

Dates rarely held. Lake Mariner's 100 MW first phase by end-2022 (14106) slipped. So did the near-term miner deployments of 3,300 units by Q3 (13880) and 3,000 S19 XPs by August (13877), and Nautilus ASIC installation by mid-2022 (14122, 14024). By February 2023 the language had softened to "initial buildout completion" at two facilities (13332, 13410). That reads as a revision that was never explicitly acknowledged. The one goal restated consistently is 100 percent zero-carbon power by 2028 (14245, 14026, 13846, 13439, 13411), against a reported 90 percent (13847).

This record contains no independent evidence: no permits, no interconnection queue positions, and no site reporting. The only confirmed power anchor at Lake Mariner is the 90 MW NYPA allocation (14110, 14126). A 410-750 MW campus would need several times that.

Over the next 90 days, four things matter. First, whether the third-quarter 2026 report puts dates and phasing on the 410 MW and 750 MW figures. Second, evidence of new NYISO or NYPA power arrangements beyond 90 MW. Third, whether the $3.5 billion financing actually closes or remains a target. Fourth, whether any of the Kentucky, Maryland, or Texas sites show up in utility or RTO interchange filings. Until then, the history argues for discounting announced pace.

TeraWulf's capacity story has inflated sharply at one site while the rest of the portfolio stays undefined. Lake Mariner was first framed in March 2022 as a 100 MW first phase due by year-end 2022 (14106), resting on a 90 MW NYPA power agreement (14110, 14126) and a 50 MW first mining building (13881). The same site is now carried as the Lake Mariner Anthropic AI Factory, disclosed at 410 MW and later targeted at 750 MW, with a $3.5 billion financing target in a July 13, 2026 headline. The claim ids for that later phase fall outside the portion of the record available here, which cuts off after claim 13335. Nautilus moved the other way. It was presented as 300 MW (14111, 14121, 14130), then as a 200 MW initial facility (13715) in which TeraWulf holds a one-third stake (13714) and 150 MW of joint-venture capacity (14112). The 800 MW aggregate target for 2025 (14108) is marked delayed. The twelve other projects on record, across Kentucky, Maryland, Texas and New York, carry no capacity or cost figures.

Dated promises from the 2022 cycle almost uniformly slipped. These include the 100 MW Lake Mariner phase (14106), the 3,000- and 3,300-miner milestones for mid-2022 (13877, 13880), Nautilus ASIC installation by June 2022 (14122, 14024), and the 100 MW Nautilus expansion due by May 2024 (13716). The 58,000-miner delivery commitment was restated at least five times through mid-2022 (14247, 14049, 14025, 13978, 13897) without a revised date, and a 78,000-unit procurement figure (14142) appeared alongside it unreconciled. The 2028 zero-carbon target has been repeated consistently (14245, 13846, 13343), with 90 percent cited as the current level (13847). It is the one commitment that has not moved.

The record contains no independent corroboration of the current pace: no permits, no NYISO interconnection queue positions, and no on-the-ground reporting. That gap matters because 410 to 750 MW is far above the only documented power arrangement at Lake Mariner, the 90 MW NYPA allocation.

Over the next 90 days, watch whether the $3.5 billion financing closes and on what terms. Watch for NYISO queue or load-study filings that could support capacity well beyond 90 MW, and for third-quarter 2026 disclosures that settle whether 410 MW or 750 MW is the operative near-term figure. A first energized-capacity date for the Anthropic-leased halls would also be meaningful, as would the first capacity figures for any of the twelve placeholder sites.

TeraWulf's claim history reads as two companies sharing a ticker. The bitcoin-mining record that ends in early 2023 is the only substantiated part of this file. The AI-campus story that now headlines Lake Mariner, at 410 MW initially and later 750 MW against a reported $3.5 billion financing target, has no supporting claims in the extract supplied, so its trajectory cannot be checked against the company's own prior statements here.

On capacity, the early figures shrank before they grew. The 800 MW aggregate target for 2025 (claim 14108) is marked delayed. Nautilus was pitched at 300 MW (14130, 14121), reset to 200 MW initial with a 100 MW expansion (13715) that missed its May 2024 date (13716). TeraWulf's position there was described as a 150 MW joint-venture share (14112) and later as a one-third equity stake (13714). At Lake Mariner, a 100 MW first phase by year-end 2022 (14106) gave way to a 50 MW first building (13881) under a 90 MW NYPA allocation (14110, 14126). That 90 MW is the only documented power entitlement for the site, which makes the jump to 410 and 750 MW the central question.

Dates have slipped repeatedly and mostly without acknowledgement. The 58,000-miner delivery commitment for 2022 was restated almost verbatim across at least five filings (14247, 14049, 14025, 13978, 13897) and never formally revised before it lapsed. Nautilus ASIC installation by mid-2022 (14122, 14024) and the 3,000 to 3,300 miner ramp at Lake Mariner (13877, 13880) missed their windows by quarters, not weeks. The 2028 zero-carbon target (14245 through 13343) is the one commitment held steady, and it carries little execution risk.

No independent evidence appears in this record: no permits, no NYISO or PJM queue positions, no site reporting. The twelve other announced projects in Kentucky, Maryland and upstate New York carry no capacity or cost at all.

Over the next 90 days, watch for the Q3 2026 10-Q, expected in November, to state an energization schedule for the first Lake Mariner AI tranche. Also watch whether NYPA or NYISO filings show load beyond the 90 MW allocation, and whether the $3.5 billion financing has actually closed rather than merely been targeted. Any of the dormant Kentucky or Maryland sites gaining a megawatt figure would also be telling.

TeraWulf's record reads as two separate trajectories joined by a shared site name. The supplied claims run from March 2022 to February 2023 and describe a bitcoin miner. The AI-era figures attached to Lake Mariner (410 MW rising to a 750 MW campus target, and a $3.5 billion financing target) appear only in project metadata, with no claim ids. They cannot yet be tested against the company's own earlier statements. The record also ends mid-entry at claim 13346, so anything after February 2023 is out of view.

Within that window, capacity moved in both directions. Nautilus was first framed at 300 MW (14121, 14111, 14130). It was later restated as 200 MW initial with a 100 MW expansion option (13715, 13716), and that option is now marked delayed. TeraWulf's own position narrowed from a 150 MW joint-venture share (14112) to a 33.33 percent equity stake (13714), a reduction that was not presented as one. Miner procurement drifted between 58,000 units (14247, 14049) and 78,000 units (14142). The 800 MW aggregate-by-2025 target (14108) is flagged delayed. At Lake Mariner, the 90 MW NYPA allocation (14126, 14110) sat below the 100 MW first phase promised for year-end 2022 (14106), which slipped, and the near-term miner counts (13877, 13880) missed their summer 2022 dates. A 2023 filing backdates mining commencement to March 2022 (13409) without reconciling the earlier dates. The 100 percent zero-carbon goal for 2028 is the only commitment repeated unchanged (14245, 13846, 13439, 13343).

There is also a data-quality problem. Nautilus is listed in Tompkins County, New York, but its claims reference a Pennsylvania facility powered by Susquehanna (14141, 14129).

The record contains no permits, interconnection-queue entries or on-the-ground reporting, so the announced pace rests entirely on company statements. The twelve other announced sites in Kentucky, Maryland, Texas and New York carry no capacity or cost figures at all.

Over the next 90 days, watch for four things: evidence that Lake Mariner load beyond the historical 90 MW has a matching NYPA allocation or NYISO interconnection basis; the close of the $3.5 billion financing; any revision to the 410 MW and 750 MW figures in third-quarter results; and first capacity numbers for any of the twelve announced sites.

The claim record supplied here ends in February 2023, so this read covers TeraWulf's bitcoin-mining build-out and only gestures at the current AI-hosting pivot. The Lake Mariner Anthropic AI Factory figures of 410 MW, 750 MW and $3.5 billion appear only in the project table, with no dated claim attached. That gap is the most important caveat for anyone using this page.

Within the record, announced capacity shrank before it grew. Nautilus was framed in March 2022 as a 300 MW nuclear-powered site (14111, 14121, 14130). By August it had become 200 MW initial capacity plus a 100 MW expansion (13715, 13716), and TeraWulf's exposure was a 50 percent then one-third interest (14112, 13714). Lake Mariner moved from a 100 MW first phase by year-end 2022 (14106) to a 50 MW first building (13881), against a 90 MW NYPA supply agreement (14110, 14126). The 800 MW aggregate target for 2025 (14108) is marked delayed, and nothing in the record shows it being formally withdrawn.

Dates slipped consistently, and the company rarely acknowledged the slips. The 58,000-miner delivery commitment for 2022 was restated through June (14247, 14049, 14025, 13978, 13897), while a parallel 78,000-unit procurement figure (14142) was never reconciled. The Nautilus ASIC start of mid-2022 (14107, 14122, 14024) and the Lake Mariner Q3 2022 miner counts (13880, 13877) both missed. The 2028 zero-carbon target was repeated verbatim across at least six filings (14245, 14026, 13846, 13826, 13439, 13343) without any interim milestone.

No independent evidence is attached to these claims, meaning no permits, interconnection-queue entries or site reporting. Neither the historical pace nor the current 410-to-750 MW campus figure can be corroborated from this file. The Kentucky, Maryland, Texas and other New York sites carry no capacity or cost at all.

Over the next 90 days, the first thing to watch is whether the $3.5 billion Anthropic-facility financing referenced in the July 13, 2026 headline is shown to have closed, and on what terms. The second is whether third-quarter disclosures explain the 410 MW versus 750 MW framing as phasing or as a scope increase. The third is whether NYISO queue positions or local approvals support load beyond the roughly 90 MW historically contracted at Lake Mariner. The fourth is whether any of the twelve announced-only sites acquires a dated, sized commitment. Given the 2022 record, any new energization date deserves to be tracked against a named filing.

The capacity story at TeraWulf has moved in two directions at once. At Lake Mariner, the original framing was modest: a 90 MW NYPA supply (14110, 14126), a 100 MW first phase targeted for year-end 2022 (14106), and a 50 MW first building (13881). The project record now carries 410 MW at initial disclosure and a 750 MW campus target tied to an Anthropic lease, roughly an eightfold expansion of the original envelope. Nautilus went the other way. It was pitched at 300 MW of nuclear-backed capacity (14130, 14121, 14111), then restated as 200 MW initial with a 100 MW expansion option (13715, 13716), alongside a 33.33 percent equity position (13714) and a 150 MW Talen joint-venture figure (14112). The headline megawatts stayed attractive while TeraWulf's attributable share narrowed. The 800 MW aggregate target for 2025 (14108) is marked delayed.

Near-term dates did not hold. The 3,300 miners at Lake Mariner by Q3 2022 (13880), 3,000 S19 XPs by August 2022 (13877), and mid-2022 ASIC installation at Nautilus (14122, 14024) all slipped. More telling is the 58,000-miner delivery commitment. It was restated across at least five filings (14247, 14049, 14025, 13978, 13897) without visible revision before being marked delayed. The 2028 zero-carbon target has been repeated almost verbatim (14245, 14026, 13846, 13826, 13439, 13411, 13343). That consistency is cheap to maintain because the deadline is distant.

The claim ledger contains no independent corroboration: no permits, no NYISO or PJM queue entries, no site reporting. It also ends in February 2023. As a result, the HPC pivot, the Anthropic facility, and the Texas, Kentucky and Maryland sites have no dated claims to test against. The $3.5 billion financing figure rests on a single July 2026 headline rather than a filing in this record.

Three things matter over the next 90 days. First, whether NYISO interconnection or large-load filings exist that can support Lake Mariner beyond 410 MW, since the 750 MW figure has no grid evidence here. Second, whether the $3.5 billion financing closes, and on what terms. Third, whether the Q3 2026 results report energized, leased megawatts rather than contracted ones. Any first permits or utility agreements for the Kentucky and Maryland sites would be the earliest test of whether the newer pipeline is real or still nominal.

The claim record supplied here ends abruptly in February 2023 at claim 13342, so this read covers the bitcoin-mining era in detail. It can only flag, not assess, the 2026 pivot to an Anthropic-leased AI campus at Lake Mariner. The 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion financing figure appear on the project line without supporting claim ids, and should be treated as unverified until those claims are loaded.

On capacity, the early trajectory is one of quiet contraction. Nautilus was pitched at 300 MW of nuclear-adjacent capacity (14111, 14121, 14130). By August 2022 that had become a 200 MW initial facility (13715) with a 100 MW expansion pushed out and later marked delayed (13716), alongside a one-third equity stake (13714). The JV figure attributed to Talen sat at 150 MW (14112). The 800 MW aggregate target for 2025 (14108) is marked delayed. Lake Mariner moved the other way, from a 90 MW NYPA allocation (14110, 14126) to the AI-era figures, a scale-up of roughly an order of magnitude that the record does not yet document step by step.

On dates, slippage was routine and mostly unacknowledged in the claim text itself. The 100 MW first phase by year-end 2022 (14106), the 3,300 miners by Q3 2022 (13880), the 3,000 S19 XPs by August (13877) and the mid-2022 Nautilus ASIC start (14122, 14024) all lapsed. The 58,000-miner delivery commitment was restated at least five times (14247, 14049, 14025, 13978, 13897) without revision before being marked delayed. The one goal held steady throughout is the 2028 zero-carbon target (14245, 13439, 13343), which is low-cost to repeat.

No independent evidence such as permits, NYISO queue positions or site reporting is attached to any claim. The twelve other announced sites, from Abernathy to Boyd-Greenup, carry no capacity or cost at all. Announced pace is therefore unsupported by third-party corroboration on this record.

Over the next 90 days, watch for the Q3 2026 filing to restate or revise the 410 and 750 MW figures. Also watch whether the $3.5 billion financing actually closes, as opposed to being headlined. NYISO interconnection and Niagara County permit filings would confirm whether incremental load beyond the legacy NYPA block is secured. Finally, check whether any of the dozen dormant sites acquires a megawatt figure or quietly drops out.

The supplied claim record stops in February 2023, so the current headline project, the Lake Mariner Anthropic facility (410 MW initially, later a 750 MW campus target, with a $3.5 billion financing goal), cannot be checked against dated claims here. What the record does show is TeraWulf's early delivery pattern, which is the right baseline for judging the AI pivot.

Capacity claims shrank as they moved from announcement to structure. In March 2022 the company described Nautilus as a 300 MW nuclear-powered facility (14121, 14130, 14111) and set an 800 MW aggregate target for 2025 (14108). By August 2022 Nautilus was framed as 200 MW initial capacity (13715), with the other 100 MW pushed to an expansion now marked delayed (13716), and TeraWulf's share was a one-third equity stake (13714). The headline number held in name while the attributable megawatts fell. At Lake Mariner, a 100 MW first phase by year-end 2022 (14106) gave way to a 50 MW first building (13881) on a 90 MW NYPA supply (14110, 14126).

Dates slipped routinely, and the slips were rarely acknowledged in later language. Pennsylvania ASIC installation by mid-2022 (14122, 14024), 3,300 miners by Q3 2022 (13880), 3,000 S19 XPs by August 2022 (13877) and the 58,000-miner delivery commitment (14247, 14049, 13897) all slipped. By February 2023, the company described 'initial buildout completion' with no date or megawatt figure (13332), not a reconciliation against the 2022 targets. The one steady commitment, 100 percent zero-carbon energy by 2028, has been repeated verbatim (14245, 13846, 13411, 13343). With 90 percent already claimed (13847), that target is cheap to restate.

The record has no independent corroboration: no permits, no NYISO or PJM queue entries, no site reporting. It also lists Nautilus in Tompkins County, New York, while the claims tie it to Susquehanna in Pennsylvania (14129, 14141). That discrepancy should be fixed before anyone cites the project.

Over the next 90 days, the key test is whether the $3.5 billion Anthropic financing closes on the stated terms. A second is whether the jump from 410 to 750 MW is backed by NYPA allocations or NYISO interconnection filings, not just campus language. Watch also for permit or queue activity at the Kentucky and Maryland sites, especially the Morgantown Generating Station in PJM territory. The 2023–2026 claims should be ingested before this read is treated as complete.

The claim record supplied here ends in February 2023, partway through claim 13345, so this read covers TeraWulf's bitcoin-mining buildout and can only bracket the current Lake Mariner AI pivot from the project ledger. On capacity, the early pattern is contraction presented as continuity. Nautilus was framed at 300 MW (14111, 14121, 14130). By August 2022 it had become a 200 MW initial facility (13715). TeraWulf's own share moved from a 150 MW joint-venture slice (14112) to a 33.33 percent equity stake (13714), and the 100 MW expansion slated for May 2024 is marked delayed (13716). Equipment figures also diverged without reconciliation: 58,000 miners in most filings (14247, 14049, 14025, 13978, 13897) against 78,000 units in one (14142). The 800 MW aggregate target for 2025 (14108) did not hold. On timing, nearly every dated 2022 milestone slipped. These include the 100 MW first phase at Lake Mariner by year-end 2022 (14106), 3,300 miners by July (13880), 3,000 S19 XPs by August (13877), and Pennsylvania ASIC installation by mid-2022 (14122, 14024). By February 2023 the company was describing undated 'initial buildout completion' (13332) rather than restating or explaining the missed dates. That is revision by omission, not acknowledgement. The one durable commitment is the 100 percent zero-carbon target for 2028, repeated nearly verbatim (14245, 13846, 13826, 13439, 13411, 13343). The only interim datapoint is a 90 percent figure (13847). The record contains no independent corroboration of any kind, including permits, NYISO or PJM queue positions, and site reporting. The only grid anchor is the 90 MW NYPA agreement (14110, 14126), and that is far below the 410 MW initial and 750 MW campus figures now attached to the Anthropic-leased Lake Mariner facility. Twelve other announced sites carry no disclosed capacity or cost. Over the next 90 days, watch whether the $3.5 billion financing from the July 13, 2026 headline actually closes and on what terms. Watch for NYPA or NYISO filings that substantiate power beyond the original 90 MW. Watch whether the third-quarter report dates the first energized AI capacity and reconciles the 410 MW and 750 MW figures. Watch whether the Kentucky, Maryland, and Texas sites acquire concrete megawatt figures or interconnection filings. Given the 2022 record, a dated milestone backed by a utility or permit document should count for more than any restated headline number.

The claims supplied here run only through February 2023. The current flagship, the Lake Mariner Anthropic AI Factory at 410 MW, later 750 MW, with a $3.5 billion financing target, appears only in the project summary and has no claim ids behind it. The trajectory can be read clearly for the bitcoin-mining era. The AI-era numbers cannot yet be tested against the record.

Capacity has repeatedly been announced large and delivered in smaller pieces. In March 2022 Nautilus was framed as a 300 MW nuclear-powered site (14121, 14111, 14130), with TeraWulf's share put at 150 MW (14112). By August the initial build was 200 MW (13715). The remaining 100 MW became a contingent expansion with a May 2024 date, and that date is now marked delayed (13716). Lake Mariner followed the same pattern. The company promised 100 MW online by year-end 2022 (14106) on a 90 MW NYPA contract (14110, 14126). By June the description had narrowed to a single 50 MW building (13881). The 800 MW aggregate target for 2025 (14108) has lapsed. A 58,000-miner delivery commitment was restated at least six times (14247, 14123, 14049, 14025, 13978, 13897) and never shown as met. A 78,000-unit procurement figure (14142) sat alongside it in the same filing without reconciliation.

Dates slipped and were rarely acknowledged. Near-term miner targets (13880, 13877) missed within weeks. Later filings moved to softer language such as "initial buildout completion" (13332) instead of confirming the original milestones. The 2028 zero-carbon goal has been repeated without change (14245, 13846, 13439, 13343). It is the one target that has held steady.

The record contains no independent corroboration of any kind: no permits, no interconnection queue positions, and no on-the-ground reporting. There are also data-quality problems. Nautilus is listed in Tompkins County, New York, while its claims describe a Pennsylvania facility (14141, 14122). The two Morgantown, Maryland entries appear to be duplicates. Several acquisition claims carry promise dates earlier than the claim itself (14119, 13408). Eleven announced sites, spanning Texas, Kentucky, Maryland and New York, have no capacity or cost figures at all.

Over the next 90 days, watch whether the Q3 10-Q commits to 410 MW or 750 MW with dated energization phases, and whether the $3.5 billion financing actually closes. Also watch for NYISO load or interconnection filings large enough to support the larger figure, and for the first permit or utility filings at the Kentucky, Maryland and Abernathy sites. Given the history recorded here, the headline figure matters less than what is energized on schedule.

The claims record supplied here stops in February 2023, partway through claim 13344. Nothing in it covers the Lake Mariner Anthropic facility's 410 MW initial disclosure, the 750 MW campus target, or the $3.5 billion financing headline. So this read covers the bitcoin-era delivery pattern that the AI pivot now builds on, not the pivot itself. That earlier pattern points to consistent overreach on near-term dates. Headline capacity was revised down quietly while long-dated targets were repeated unchanged. Nautilus shows it most clearly. It was presented at 300 MW (14121, 14111, 14130), with TeraWulf's share framed as 150 MW through the Talen joint venture (14112). By August 2022 the facility was described as 200 MW initial with a 100 MW expansion (13715, 13716), and TeraWulf's stake appeared as 33.33 percent (13714). That means the company's attributable share fell well below the original framing, and none of the claims acknowledges the change. The 800 MW aggregate target for 2025 (14108) is marked delayed. At Lake Mariner, the 100 MW first phase promised by year-end 2022 (14106) slipped, as did the smaller milestones of 3,300 miners by Q3 (13880) and 3,000 S19 XPs by August (13877). The 50 MW first-building figure (13881) suggests the phase was quietly split in half. Miner procurement moved between 58,000 units (14123, 14049, 13897) and 78,000 (14142), and both are delayed. Pennsylvania ASIC installation by June 2022 (14024, 14122) also slipped. By early 2023 the company described its goals as 'initial buildout completion' with no dates attached (13332), which reads as a retreat from dated commitments rather than a restatement of them. The 2028 zero-carbon target, by contrast, recurs without change (14245, 13439, 13343). The record contains no independent evidence at all: no permits, no interconnection queue positions, no site reporting. The only grid anchor is the 90 MW NYPA allocation (14110, 14126), far below either AI-era figure. Twelve other announced sites carry no capacity at all. Nautilus is also filed under Tompkins County, New York, while its claims describe the Susquehanna site in Pennsylvania, which looks like a mapping error worth fixing. Over the next 90 days, watch four things: whether the Q3 2026 filings reconcile 410 MW against 750 MW with dated phases; whether the $3.5 billion financing actually closes; whether NYISO or NYPA filings show power beyond the 90 MW baseline; and whether any Kentucky or Maryland site gets a megawatt figure. Ingesting the missing 2023–2026 claims should come before any published verdict.

The capacity record for TeraWulf shows ambitious headline numbers that were cut back quietly. The March 2022 filings set an 800 MW aggregate target for 2025 (claim 14108), which is now marked delayed. Nautilus was described as a 300 MW facility (14121, 14111, 14130). By August 2022 the same asset was framed as 200 MW of initial capacity plus a 100 MW expansion (13715, 13716), and that expansion has also slipped. TeraWulf's economic stake narrowed from a 150 MW share of a Talen joint venture (14112) to a 33.33 percent equity interest (13714), and no disclosure in this record explains the change. Lake Mariner began as a 90 MW NYPA allocation (14110, 14126), with a 100 MW first phase promised by year-end 2022 (14106) and a 50 MW first building (13881). The current project record jumps to 410 MW and then a 750 MW campus target for an Anthropic-leased facility. None of the claims supplied here documents that step-up, so the pivot from bitcoin hosting to HPC cannot yet be traced claim by claim. The claim set provided ends in February 2023; everything after that date comes from the project record alone.

The early timelines did not hold. Mid-2022 ASIC installation at the Pennsylvania site (14122, 14024), 3,300 miners online at Lake Mariner by July 2022 (13880), and the initial S19 XP batch (13877) all slipped. The 58,000-miner delivery commitment was restated at least five times through June 2022 (14247, 14049, 14025, 13978, 13897) without revision before it was marked delayed. It also sits alongside a 78,000-unit procurement figure (14142) that was never reconciled with it. The 2028 zero-carbon target has been repeated consistently (14245, 13439, 13343), which is the one stable commitment in the record.

The record contains no independent corroboration: no permits, no NYISO or PJM queue entries, and no site reporting. That matters most for the Anthropic facility, because it is the only project marked under construction. There are also data hygiene problems. Nautilus is mapped to Tompkins County, New York, even though its claims tie it to Susquehanna in Pennsylvania. Chesapeake Data and Morgantown Generating Station appear to overlap. Claim 21075 is a 2006 IKONICS-era item. Nine announced sites carry no capacity or cost at all.

Over the next 90 days, the key checks are the following. The third-quarter 2026 results should state whether 410 MW or 750 MW is the contracted figure and give energization dates by phase. There should be evidence of whether the $3.5 billion financing has closed. NYISO interconnection filings or Town of Somerset permits should confirm load growth at Lake Mariner beyond the original 90 MW allocation. Finally, the Kentucky, Maryland and Texas sites should acquire any capacity figures at all.

The capacity story here is less a steady scale-up than a series of resets. In early 2022 TeraWulf framed an 800 MW aggregate target for 2025 (14108), built on Lake Mariner's 90 MW NYPA allocation (14110, 14126) and a 300 MW Nautilus position (14111, 14121, 14130). Even within the same filing window, Nautilus was also described as a 150 MW joint-venture share (14112). By August 2022 it had been restated as 200 MW initial with a 100 MW expansion (13715, 13716), and that expansion is now marked delayed. The 800 MW target is also delayed. Its successor is the Lake Mariner Anthropic AI Factory, whose project record shows a 410 MW initial disclosure followed by a 750 MW campus target. The visible claims contain no dated entries for either figure, so the step-up cannot be tied to specific filings here.

On timing, the early record is consistently missed. The 100 MW first phase by end of 2022 (14106), 3,300 miners online by Q3 2022 (13880), the initial S19 XP batch (13877), and the Nautilus ASIC install by mid-2022 (14122, 14024) all slipped. The 58,000-miner delivery commitment repeated across 14049, 14025, 13978 and 13897 also slipped. It sits awkwardly beside a 78,000-unit procurement figure (14142) that was never reconciled in the record. The one target restated without drift is 100 percent zero-carbon by 2028 (14245, 13846, 13439, 13411, 13343), and it remains open. One data caveat: Nautilus is tagged to Tompkins County, New York, while its claims describe a Pennsylvania site powered by Susquehanna (14129), so the project mapping needs correcting.

The record contains no independent corroboration of the current pace. The NYPA allocation is the only externally anchored figure. There are no permits, NYISO or ERCOT queue entries, or site reporting for Lake Mariner's expansion or for the Kentucky, Maryland, Texas and Cayuga sites. All of those sites carry announced status with no capacity or cost attached. The claims list is also truncated after February 2023 (13344), so the 2024 to 2026 pivot toward HPC is not visible at claim level.

Four things are worth watching over the next 90 days. First, whether the $3.5 billion Anthropic-facility financing closes on the terms in the July 13 headline. Second, whether the Q3 2026 results reconcile the 410 MW and 750 MW figures with dated energization phases. Third, whether NYISO interconnection filings or Town of Somerset permits support anything beyond the existing allocation. Fourth, whether any of the dozen announced sites receives a disclosed capacity figure or a regulatory filing.

The record supplied here ends in February 2023 (claim 13333 is cut off mid-entry). None of the dated claims cover the AI-factory era, so the 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion Anthropic-linked financing appear only in the project summary and cannot be tested against claim-level history. What the record does show is the delivery pattern that preceded the pivot, and it is not encouraging. Announced capacity shrank quickly once projects met physical constraints. The 800 MW aggregate target for 2025 (14108) is marked delayed. Nautilus was framed as 300 MW of nuclear-adjacent capacity (14111, 14121, 14130). Within five months it was restated as 200 MW initial with a contingent 100 MW expansion (13715, 13716), and TeraWulf's exposure fell from a 150 MW joint-venture share (14112) to a one-third equity stake (13714). The project list places Nautilus in Tompkins County, New York, but the claims describe a Pennsylvania site supplied by Susquehanna (14113, 14129). That mapping error should be fixed before publication. Lake Mariner followed the same arc. A 100 MW first phase by year-end 2022 (14106) gave way to a 50 MW first building (13881) against a 90 MW NYPA allocation (14110, 14126), and the near-term miner milestones (13877, 13880) both slipped. Dates were frequently revised without acknowledgement. The 58,000-unit equipment commitment was repeated across at least six filings (14123, 14049, 14025, 13978, 13897, 14247) and is marked delayed each time. A parallel 78,000-unit procurement figure (14142) was never reconciled with it. Claim 13409, filed in February 2023, carries a March 2022 commencement date, which reads as retroactive framing rather than a met promise. The only commitment that held steady was the 100 percent zero-carbon target for 2028 (14245, 13343), and it has no interim milestone attached beyond a 90 percent figure (13847). The record contains no independent evidence at all: no permits, NYISO or PJM queue positions, or site reporting. Nothing here corroborates the announced pace at any stage. Over the next 90 days, watch four things. First, whether the $3.5 billion financing actually closes, and on what terms. Second, whether NYPA or NYISO filings show load approvals large enough to support 410 MW, let alone 750 MW, well above the historical 90 MW allocation. Third, whether lease disclosures specify energization dates. Fourth, whether the company explains the 410-to-750 MW step-up or simply lets the larger number replace the smaller one, as happened with the equipment figures.

TeraWulf's record shows a company whose stated capacity has moved in both directions at once. Early disclosures set an 800 MW aggregate target for 2025 (claim 14108), now marked delayed. Nautilus was first described as 300 MW (claims 14111, 14121, 14130), with TeraWulf's slice framed as 150 MW under the Talen joint venture (claim 14112). By August 2022 that had become a 200 MW initial build with a 100 MW expansion (claims 13715, 13716), and the company's position had become a 33.33 percent equity stake (claim 13714). Neither change was presented as a reduction. Lake Mariner started from a 90 MW NYPA supply (claims 14110, 14126) and a 50 MW first building (claim 13881). The project file now carries 410 MW initial and 750 MW campus figures plus a $3.5 billion financing target, but none of these is tied to a claim id in the supplied record, which ends in February 2023. The path from 90 MW to 750 MW therefore cannot be traced from this record.

On timing, nearly every dated 2022 promise slipped. These include the 100 MW first phase by year-end (claim 14106), the 3,000 and 3,300 miner milestones (claims 13877, 13880), Nautilus ASIC installation by mid-2022 (claims 14122, 14024), and the 58,000-miner delivery commitment, which was repeated across five filings (claims 14247, 14049, 14025, 13978, 13897). A 78,000-unit procurement figure (claim 14142) sits in the same March 2022 filing as the 58,000 figure without reconciliation. The one target restated consistently is 100 percent zero-carbon power by 2028 (claims 14245, 13846, 13411), with 90 percent reported as achieved in mid-2022 (claim 13847).

The record contains no independent corroboration: no permits, no NYISO or PJM interconnection queue positions, and no on-the-ground reporting. That gap matters most for the 750 MW figure, which would require power far beyond the documented 90 MW NYPA allocation. It also matters for the twelve other listed sites in New York, Kentucky, Maryland and Texas, which carry no capacity or cost at all.

Over the next 90 days, watch four things. First, whether the $3.5 billion Anthropic-linked financing closes and on what milestones. Second, any filing that documents incremental grid or NYPA power at Lake Mariner to support capacity beyond the original allocation. Third, whether the next quarterly report restates the 410 MW initial figure or its energization date. Fourth, whether any Kentucky or Maryland site acquires a disclosed megawatt figure backed by a utility or queue filing rather than a press release.

TeraWulf's capacity story at Lake Mariner has grown roughly eightfold. It began as a 90 MW NYPA power allocation (14110, 14126), was framed as a 100 MW first phase (14106), and was physically anchored by a single 50 MW mining building (13881). The project record now carries a 410 MW initial disclosure and a 750 MW campus target tied to an Anthropic lease and a $3.5 billion financing effort. The claim ledger supplied here ends in early 2023, so none of the AI-era figures can be traced to a dated claim id. The step-up from bitcoin-mining megawatts to hyperscale tenancy therefore rests on the project summary, not on an auditable claim trail. Nautilus shows the opposite drift. It opened at 300 MW (14111, 14121, 14130), was restated as 200 MW initial capacity with a one-third equity stake (13715, 13714), and its 100 MW expansion window lapsed (13716). Nautilus now sits in the record as merely announced, with no capacity attached. On dates, the early pattern is consistent slippage without explicit restatement. The Lake Mariner year-end 2022 phase (14106), the 3,000 and 3,300 miner milestones (13877, 13880), the mid-2022 Pennsylvania ASIC installation (14024, 14122), and the 800 MW aggregate-by-2025 target (14108) are all flagged delayed. Procurement also wavered between 58,000 miners (14123, 14049, 13897) and 78,000 (14142) inside a single month. The 100 percent zero-carbon-by-2028 target (14245, 13846, 13343) has been repeated verbatim across filings and remains untested. It sits awkwardly beside a pipeline that includes the Morgantown Generating Station, historically a fossil-fired plant site. The IKONICS-era entries (21075, 14119) look like legacy shell artifacts rather than delivery signals. The record contains no permits, NYISO or ERCOT queue positions, or site reporting, so the announced pace is unverified from independent sources. That gap matters most for the eleven Kentucky, Maryland, Texas and New York sites that carry no capacity or cost at all. Over the next 90 days, watch four things. First, whether the $3.5 billion Anthropic facility financing closes on the stated terms. Second, whether the Q3 2026 10-Q reports energized critical-IT megawatts at Lake Mariner rather than contracted or targeted megawatts. Third, whether NYISO interconnection filings support moving beyond 410 MW toward 750 MW. Fourth, whether Abernathy, Cayuga or any Kentucky site acquires a disclosed capacity figure, a utility agreement, or a local permit that would move it beyond a name on a list.

TeraWulf's Lake Mariner record reads as two different companies on the same site. The 2022 filings describe a bitcoin miner promising 800 MW of aggregate capacity by end-2025 (claim 14108), 100 MW of Lake Mariner online by year-end 2022 (claim 14106), and 58,000 to 78,000 miners delivered within 2022 (claims 14247, 14142, 14049). Every one of those dated targets is now marked delayed, as are the mid-2022 Lake Mariner milestones of 3,300 and 3,000 operating units (claims 13880, 13877). The pattern is consistent: capacity was announced at the scale of the ambition and delivered at the pace of power and equipment. Nautilus shows the same thing on a smaller scale. The headline 300 MW (claims 14111, 14130) was restated months later as 200 MW initial plus a 100 MW expansion (claims 13715, 13716), and the expansion's May 2024 date is also recorded as delayed. The company framed that change as phasing, not a reduction, but it cut the near-term commitment by a third. The AI pivot resets the baseline. The project record now shows a 410 MW initial disclosure, a 750 MW campus target, and a $3.5 billion financing target tied to an Anthropic lease. However, the claim extract supplied here ends in early 2023, so those figures cannot be tied to specific dated claims or checked against any promised energization dates. That is a material gap. Moving from 90 MW of NYPA supply (claims 14110, 14126) to a 750 MW target is exactly the kind of step change that needs dated milestones before anyone can judge it. The record also contains no independent evidence: no NYISO interconnection queue entries, no NYPA allocation filings, no building permits, no on-the-ground reporting. The announced pace is therefore uncorroborated here, not contradicted. Still, none of the dated capacity and delivery promises in this record held on schedule, so the burden of proof sits with the company. The next 90 days offer four tests. First, whether the $3.5 billion financing closes and on what terms. Second, whether TeraWulf publishes phase-by-phase energization dates for the 410 MW tranche rather than a campus-level target. Third, whether queue or utility filings show power commitments well beyond the original 90 MW. Fourth, whether any of the long list of announced sites from Kentucky to Maryland gains a disclosed capacity or cost figure, or whether they remain names without numbers.

The record here runs from March 2022 to February 2023, so it describes the bitcoin-mining buildout that preceded the Lake Mariner AI campus. It cannot say much directly about the 410 MW and 750 MW Anthropic-era figures, which appear only in project metadata and are not tied to any dated claim. Read that way, the early trajectory is one of ambition compressed by execution. The headline aggregate target of 800 MW by end-2025 (14108) was carried as delayed, and it does not reappear in the February 2023 filings. Those filings describe only an 'initial buildout' at two sites (13332, 13410). Lake Mariner's footprint moved from a 90 MW NYPA supply (14110, 14126) and a 100 MW first phase due by end-2022 (14106, delayed) to a single 50 MW first building (13881). Nautilus contracted most visibly. It was described as a 300 MW nuclear-powered facility (14111, 14121, 14130), then as 200 MW initial capacity with a contingent 100 MW expansion (13715, 13716, delayed). TeraWulf's economic share was stated at 150 MW (14112) and later as a one-third equity stake (13714). The project register places Nautilus in Tompkins County, New York, but every claim ties it to Pennsylvania and Susquehanna (14122, 14129, 14141). That entry should be checked. Dated promises rarely held. Pennsylvania ASIC installation by mid-2022 slipped (14122, 14024). So did the 3,000 and 3,300 miner energization targets at Lake Mariner (13877, 13880) and the 58,000-unit delivery commitment (14247, 14049, 13978), which sits alongside an unreconciled 78,000-unit procurement figure (14142). The 2023 filings restate the buildout without dates rather than acknowledging the misses. Only the 2028 zero-carbon target has been repeated consistently (14245, 14026, 13439, 13343). The record contains no permits, no NYISO or PJM queue entries, and no site reporting, so none of the announced pace can be independently corroborated here. That gap matters most for the 750 MW campus target, which is roughly eight times the only documented Lake Mariner power agreement. Over the next 90 days, watch whether the $3.5 billion Anthropic financing closes on stated terms. Look for interconnection or NYPA allocation filings that support load well above 90 MW, and for any dated energization milestone for the first AI halls. Also watch whether the ten capacity-less Kentucky, Maryland, Texas and New York sites acquire megawatt figures. Given this history, figures published without a date or a utility counterparty deserve discounting.

TeraWulf's claim record ends in February 2023, while the project it now attaches to, the Lake Mariner Anthropic AI Factory at 410 MW growing to 750 MW with a $3.5 billion financing target, rests on a July 2026 headline rather than any scored claim. The trajectory can be judged only on the bitcoin-era commitments, and those point to optimistic sizing and quiet date erosion. Capacity moved in both directions. The 800 MW aggregate target for 2025 (14108) is marked delayed. Nautilus was presented at 300 MW with a 150 MW TeraWulf share (14111, 14112, 14130). By August 2022 it had been restated as 200 MW initial plus a 100 MW expansion (13715, 13716), and TeraWulf's stake had fallen to one-third (13714). Lake Mariner was pitched as 100 MW online by end-2022 (14106) on a 90 MW NYPA supply (14110, 14126). That mismatch was never reconciled, and the first dedicated building was later described at 50 MW (13881). Dates rarely held. The 58,000-miner delivery commitment was repeated at least five times through mid-2022 (14247, 14049, 14025, 13978, 13897) and slipped each time, while a parallel 78,000-unit procurement figure (14142) was never explained. Nautilus ASIC installation by mid-2022 (14122, 14024) and the Lake Mariner miner targets for July and August 2022 (13880, 13877) also slipped. By February 2023, filings described initial buildout completion with no dates attached (13332, 13342), which amounts to a revision without acknowledgement. The one constant is the 100 percent zero-carbon target for 2028 (13826, 13411, 13343), which is too soft to test. No independent evidence appears in the record: no NYISO queue position, no local permits, no reporting from the site that would support a jump from a 90 MW NYPA allocation to 410 or 750 MW. The project list also carries data-quality problems. Nautilus is mapped to Tompkins County, New York, although the underlying claims tie it to Pennsylvania and Susquehanna (14129), and eleven other sites have no capacity or cost at all. Over the next 90 days, watch whether the $3.5 billion financing closes on disclosed terms, whether the Q3 2026 filing gives dated, per-phase megawatt figures for Anthropic, and whether interconnection or NYPA allocation filings appear that show power beyond the original 90 MW.

TeraWulf's capacity story has grown sharply in ambition while the dated record behind it has thinned. The early Lake Mariner baseline rested on a 90 MW NYPA power arrangement (14110, 14126), with a first 50 MW building (13881) inside a 100 MW first phase promised by year-end 2022 (14106). The project file now carries a 410 MW initial disclosure and a 750 MW campus target for the Anthropic-leased facility, plus a $3.5 billion financing target. None of that expansion appears as a dated claim in this record, so the jump from 90 MW of contracted power to 410 to 750 MW is not traceable to a stated power source or schedule. Nautilus moved the other way. It opened as a 300 MW nuclear-powered site (14111, 14121, 14130) with a 150 MW TeraWulf share (14112), then was restated as 200 MW of initial capacity (13715) with a 100 MW expansion, a 33.33 percent equity stake (13714) that sits awkwardly against the earlier half-share, and an expansion date that lapsed (13716). Miner procurement shrank quietly from 78,000 units (14142) to 58,000 (14247, 14049, 13897). The 800 MW aggregate target for 2025 (14108) is marked delayed.

Nearly every dated operational promise from 2022 slipped: Nautilus ASIC installation by mid-2022 (14122, 14024), 3,300 Lake Mariner miners by Q3 (13880), and 3,000 S19 XPs by August (13877). By February 2023 the language had shifted to undated references to initial buildout completion (13332) rather than any restatement of the missed milestones, which amounts to revision without acknowledgement. The only commitment reiterated consistently is the 2028 zero-carbon target (14245, 14026, 13826, 13439, 13343), which is a sourcing claim, not a delivery one.

This record contains no independent corroboration: no NYISO queue positions, no local permits, and no site reporting for Lake Mariner's expansion or for the twelve other announced sites, none of which carries a capacity or cost figure. The announced pace is therefore unverified rather than contradicted.

Over the next 90 days, watch whether the $3.5 billion financing actually closes and on what terms. Also watch whether TeraWulf identifies the power supply that bridges 90 MW of NYPA allocation to 410 MW and beyond, whether NYISO filings or state permits surface for the added load, and whether the Q3 2026 report attaches energization dates to the 410 MW and 750 MW figures. A further signal is whether any of the Kentucky, Maryland, Texas, or New York sites acquires a stated megawatt figure.

TeraWulf's claim record for this cluster ends in February 2023. The current framing of Lake Mariner as a 410 MW, later 750 MW, AI campus backed by a $3.5 billion Anthropic-linked financing target has no dated claims behind it in this dataset. Those figures come from project-level metadata and a July 2026 headline, not from sourced statements. Readers should treat the AI-era scale as asserted but not yet traceable.

Within the traceable window, capacity was reframed downward. The March 2022 aggregate target of 800 MW by end of 2025 (claim 14108) is now marked delayed. Nautilus was first presented as 300 MW of nuclear-powered capacity (claims 14111, 14121, 14130). By August 2022 it had become a 200 MW initial facility, with 100 MW described only as a potential expansion (claims 13715, 13716), and TeraWulf holding a one-third stake (claim 13714). That cut TeraWulf's attributable Nautilus capacity well below the 150 MW joint-venture figure it had cited earlier (claim 14112). Lake Mariner's 100 MW first phase (claim 14106) sat on a 90 MW NYPA power agreement (claims 14110, 14126), and the first building was described at 50 MW (claim 13881).

Dates slipped repeatedly and largely without explicit acknowledgement. The 58,000-miner delivery commitment (claims 14247, 14049, 14025, 13978, 13897) and the 78,000-unit procurement figure (claim 14142) both missed their year-end 2022 targets. The Pennsylvania ASIC install target of June 30, 2022 (claims 14122, 14024) was missed. So were the Lake Mariner ramp targets of 3,000 to 3,300 miners by mid-2022 (claims 13877, 13880). The 2023 filings shift to undated language about initial buildout completion (claim 13332), which quietly replaces the dated milestones rather than restating them. The IKONICS items flagged as delayed (claims 14119, 13980) look like a dating artifact in the record, not a real slip. The one consistently restated commitment is the 100 percent zero-carbon target for 2028 (claims 14245, 13846, 13343), with 90 percent already reported in 2022 (claim 13847).

The record contains no independent corroboration of the announced pace, such as permits, NYISO or PJM interconnection queue entries, or site reporting. That is the central gap for the 750 MW story.

There are four things to watch in the next 90 days. First, whether the Q3 2026 results reconcile 410 MW against 750 MW with dated energization phases. Second, whether the $3.5 billion financing actually closes. Third, whether NYISO queue or load-interconnection filings match the claimed Lake Mariner expansion. Fourth, whether the twelve other announced sites acquire any capacity or cost figures at all.

The claim record supplied here stops in early February 2023, partway through claim 13332, so the path from a 90 MW bitcoin site to the 410 MW and then 750 MW Anthropic-anchored Lake Mariner campus in the project header is not documented claim by claim. That gap is the first thing to note. The headline figures and the $3.5 billion financing target from July 2026 sit on top of a history that the record shows only in its early, crypto-mining phase.

That early phase is a clear pattern of scaling up and then missing dates. Lake Mariner was framed around a 90 MW NYPA allocation (14110, 14126), with a 100 MW first phase promised by year-end 2022 (14106). Near-term milestones then shrank to a 50 MW first building (13881) and a few thousand machines by mid-2022 (13877, 13880), and both of those also slipped. The 58,000-miner delivery commitment was repeated at least five times (14247, 14049, 14025, 13978, 13897) and is marked delayed each time, and a larger 78,000-unit procurement figure (14142) went the same way. Nautilus is the sharpest restatement. It was first presented as 300 MW (14121, 14111, 14130) and a 50/50, 150 MW joint venture (14112), then quietly recast as 200 MW initial capacity plus a 100 MW option (13715, 13716) with a one-third equity stake (13714), and no acknowledgement of the change appears in the claims. The 800 MW aggregate target for 2025 (14108) is marked delayed. The only commitment held consistently is the 2028 zero-carbon goal (14245, 13846, 13439, 13343 and others), and it is untested because its deadline has not yet arrived.

The record contains no independent corroboration of any of this. There are no NYISO queue positions, no Town of Somerset or Niagara County permits, no NYPA allocation amendments, and no on-the-ground reporting. The dozen other announced sites, from Abernathy to Morgantown to the Kentucky campuses, carry no stated capacity or cost at all.

Over the next 90 days, the key test is whether the $3.5 billion financing actually closes and on what terms. The Q3 2026 results should also show whether the 410 MW is described as energized, contracted, or merely planned. Readers should check whether the 750 MW target acquires dated phases and a matching interconnection filing, and whether any of the unquantified sites, particularly Cayuga and Morgantown, receive capacity figures that can be checked against state utility-commission dockets. Given the 2022 record, dates should be weighted well below capacity headlines until a filing supports them.

The record supplied here breaks off at claim 13332 in February 2023. Nothing from the AI-hosting era is cited below: not the 410 MW and 750 MW Lake Mariner figures, not the $3.5 billion Anthropic financing, and not the twelve newer announced sites. This read covers the bitcoin-era baseline those later claims build on, and that baseline is not encouraging. Lake Mariner entered the record as a 90 MW NYPA power arrangement (14110, 14126). Its first phase was pitched at 100 MW online by year-end 2022 (14106), but the first building was sized at 50 MW (13881). Near-term miner targets of 3,000 units by August 2022 (13877) and 3,300 by Q3 (13880) both slipped. Nautilus shrank in a similar way. It was framed as 300 MW of nuclear-backed capacity (14130, 14121, 14111), then restated as 200 MW initial with a 100 MW expansion (13715, 13716). The expansion carried a 2024 date that is now marked delayed. TeraWulf's economic interest also moved from a 150 MW half-share (14112) to a one-third equity stake (13714), and that change was never framed as a reduction. The 800 MW aggregate target for 2025 (14108) did not hold. Procurement numbers drifted between 58,000 units (14247, 14049, 14025, 13978, 13897) and 78,000 (14142) without reconciliation. The Nautilus ASIC start slipped past mid-2022 (14122, 14024). The one constant is the 100 percent zero-carbon target for 2028, restated almost verbatim across filings (14245, 14026, 13846, 13826, 13439, 13411, 13343). That target costs nothing to repeat and has no interim milestones. The pattern is ambitious nameplate figures followed by quieter operational restatements, and dates that lapse rather than get revised. Nothing in the supplied record comes from permits, NYISO or other interconnection queue filings, or on-the-ground reporting, so the announced pace cannot be independently tested here. The jump from a 90 MW NYPA allocation to a 750 MW campus target is exactly where that evidence matters most. Over the next 90 days, watch four things. First, whether the $3.5 billion financing closes on disclosed terms. Second, whether any NYISO interconnection or NYPA allocation beyond the original 90 MW becomes public. Third, whether Town of Somerset or Niagara County site-plan and building approvals for new halls appear. Fourth, whether any of the twelve capacity-less announced sites receives a megawatt figure with a date attached. Given the 2022 record, the first quarterly update that restates the Lake Mariner energization date is the key test.

TeraWulf's record on this cluster runs almost entirely through its 2022-2023 bitcoin-mining phase, and that history argues for caution when reading the current AI-factory figures. The claims excerpt supplied stops in February 2023. The 410 MW and 750 MW Lake Mariner targets and the $3.5 billion Anthropic financing appear only in the project header, with no claim ids, so they cannot yet be traced against the company's own sequence of statements.

On capacity, the earliest Nautilus framing was a 300 MW nuclear-powered facility (14121, 14130, 14111), with TeraWulf's share described as 150 MW (14112). By August 2022 that had become 200 MW initial capacity (13715) plus a possible 100 MW expansion (13716), and TeraWulf's position had become a 33.33 percent equity stake (13714). That is a material narrowing, and the record shows no explicit reconciliation. The 800 MW aggregate target for 2025 (14108) is tagged delayed.

On dates, the pattern is consistent slippage. The Lake Mariner first phase was promised at 100 MW by year-end 2022 (14106), and the 3,000- and 3,300-miner milestones (13877, 13880) all slipped. Nautilus ASIC installation by mid-2022 (14122, 14024) was also missed. The 58,000-unit miner delivery commitment was repeated across five filings (14247, 14049, 14025, 13978, 13897) and was delayed each time. A same-day 78,000-unit procurement figure (14142) sits unexplained beside those. The one steady commitment is the 100 percent zero-carbon target for 2028, restated without revision from 14245 through 13343.

On independent evidence, the record contains no permits, interconnection-queue entries or site reporting. The only grid anchor is the 90 MW NYPA supply at Lake Mariner (14110, 14126). That figure sits far below either the 410 MW or the 750 MW target, so it is not evidence for the announced pace. The same gap applies to the ten-plus sites in Kentucky, Maryland, Texas and New York that carry no stated capacity or cost.

Over the next 90 days, watch for four things. First, whether the $3.5 billion financing actually closes and on what terms. Second, whether NYISO queue positions or NYPA allocations appear that are sized to the 410 MW figure, not just the 90 MW. Third, whether energization dates for the first Anthropic data halls are disclosed. Fourth, whether the 750 MW campus target is dated or quietly re-scoped the way Nautilus was.

The record supplied here ends in February 2023, so it does not show how TeraWulf's crypto-era facilities became the AI campus now on file. It holds no dated claims for the 410 MW initial disclosure, the 750 MW campus target, or the $3.5 billion financing tied to the Anthropic lease. Those figures appear only in the project header, and this read cannot date or source them. What the record does show is the promise pattern that the AI pivot inherits. Early capacity ambition ran well ahead of delivery. The 800 MW aggregate target for 2025 (claim 14108) is marked delayed. At Lake Mariner, the 90 MW NYPA allocation (claims 14110, 14126) supported a 100 MW first phase promised for year-end 2022 (claim 14106), which also slipped. The miner-level milestones behind it, 3,300 units by Q3 2022 (claim 13880) and 3,000 S19 XPs by August 2022 (claim 13877), both went past their dates. Nautilus shows capacity shrinking without comment. It was pitched at 300 MW with TeraWulf's joint-venture share at 150 MW (claims 14121, 14112). By August 2022 the facility was framed as 200 MW initial with a possible 100 MW expansion (claims 13715, 13716), and TeraWulf's stake was a one-third equity position (claim 13714). None of the claims acknowledges that reduction. Equipment figures were also inconsistent, with 58,000 miners committed (claim 14247) against 78,000 procured (claim 14142), and both lapsed. The Pennsylvania ASIC start set for mid-2022 (claims 14122, 14024) was missed as well. The one figure that stayed fixed is the 2028 zero-carbon target, restated at least six times (claims 14245, 13846, 13826, 13439, 13411, 13343), with 90 percent reported as current (claim 13847). The record contains no permits, interconnection-queue entries or site reporting for any project, so the announced pace is uncorroborated. The eleven other announced sites in Kentucky, Maryland, Texas and New York carry no capacity or cost figures at all. Over the next 90 days, watch four things. First, whether the $3.5 billion Anthropic-facility financing closes on the terms in the headline. Second, whether NYISO or NYPA filings show load or interconnection approvals large enough for 750 MW at Lake Mariner, which is far beyond the historical 90 MW allocation. Third, whether any AI hall is energized on a stated date. Fourth, whether the Morgantown Generating Station and Kentucky sites get Maryland PSC or state-level permitting records that would turn announcements into measurable capacity.

The claim record supplied here runs from March 2022 to February 2023, and the final entry (13332) is cut off, so this read covers the bitcoin-mining origin of what is now called the Lake Mariner Anthropic AI Factory. The 410 MW initial figure, the 750 MW campus target and the $3.5 billion financing headline from July 2026 are not supported by any dated claim in this set. That gap is itself the main finding: the current AI-factory scale cannot be traced back through the record provided.

Within that window, announced capacity mostly contracted or stalled. The 800 MW aggregate target for 2025 (14108) is marked delayed. Nautilus was pitched at 300 MW (14121, 14111, 14130), with a 150 MW Talen joint-venture share (14112). By August 2022 it had been restated as 200 MW initial plus a 100 MW expansion (13715, 13716), with TeraWulf holding a one-third stake (13714). That is a material reduction in attributable megawatts, and the restatement came without an explicit acknowledgement. Lake Mariner's 100 MW first phase by year-end 2022 (14106) is delayed, and the first building was described at 50 MW (13881). The site's firm power basis remained the 90 MW NYPA allocation (14110, 14126).

Dates slipped repeatedly and in sequence. The Nautilus ASIC installation target of mid-2022 (14122, 14024) slipped. So did 3,300 miners by Q3 2022 (13880) and 3,000 S19 XPs by August 2022 (13877). The 58,000-unit delivery commitment was reiterated at least five times before being marked delayed (14247, 14049, 14025, 13978, 13897), and it sits awkwardly beside a 78,000-unit procurement figure (14142) that was never reconciled. The 2028 zero-carbon target (14245, 13846, 13343) has been restated consistently, so it at least has not drifted.

None of these claims comes with independent corroboration: no permits, no NYISO interconnection queue entries, no site reporting. The jump from a 90 MW NYPA allocation to 410 or 750 MW would need new load-serving arrangements that should be publicly verifiable. That is the central open question. The eleven other announced sites across Kentucky, Maryland, Texas and New York all carry no capacity or cost figures.

Over the next 90 days, the main things to watch are four. First, whether the $3.5 billion Anthropic-linked financing actually closes and on what terms. Second, NYISO queue or NYPA filings that would substantiate power above 90 MW at Lake Mariner. Third, whether the Q3 2026 report reconciles 410 MW with 750 MW and gives energization dates. Fourth, whether Morgantown or Abernathy receive their first quantified capacity claims.

The record supplied here ends in February 2023 (it breaks off at claim 13332), so the later disclosures behind the Lake Mariner 410 MW and 750 MW figures and the $3.5 billion Anthropic financing appear only in the project summary, not as citable claims. With that limit stated, the early history sets a useful baseline. Lake Mariner began as a 90 MW NYPA power arrangement (14110, 14126), with a 50 MW first building (13881). The current campus target of 750 MW is roughly an eightfold increase, and it marks a shift in purpose from bitcoin mining to AI hosting. Nautilus moved the other way. It was framed as 300 MW (14130, 14121, 14111) and then restated as 200 MW initial with a 100 MW expansion (13715, 13716). TeraWulf's position also narrowed, from a 150 MW half-share of the Talen joint venture (14112) to a 33.33 percent equity stake (13714). The quieter revisions are in equipment. Procurement was cited as 78,000 units (14142) alongside the repeated 58,000-miner commitment (14247, 14049, 14025, 13978, 13897), and the discrepancy was never reconciled. On dates, the 2022 promises largely slipped. These include the 100 MW Lake Mariner first phase by year-end 2022 (14106), the 3,000 and 3,300 miner targets for mid-2022 (13877, 13880), mid-2022 ASIC installation at Nautilus (14122, 14024), and the 800 MW aggregate target for 2025 (14108). Some delay flags look like tagging artifacts rather than real misses. Mining commencement was reported in March 2022 (14127) yet carries a lapsed date (13409), and the IKONICS closing (13980) is marked delayed against a pre-announcement date. The only commitment restated without revision is the 2028 zero-carbon target (14245, 14026, 13846, 13826, 13439, 13411, 13343). The record contains no permits, NYISO queue positions or site reporting, so the pace of the 750 MW expansion cannot be independently corroborated here. The long list of announced sites in Kentucky, Maryland, Texas and New York carries no capacity or cost figures, which reads as optionality rather than a pipeline. Over the next 90 days, watch four things: whether the $3.5 billion financing closes on the stated terms; whether Lake Mariner reports energized, leased megawatts, not just contracted ones; whether an NYISO interconnection filing or state permit supports the move from 410 to 750 MW; and whether any satellite site receives a first capacity number.

The claim record supplied here stops in February 2023; claim 13332 is cut off mid-entry. The 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion financing figure for the Anthropic-leased Lake Mariner build appear only in the project summary, with no claim ids attached. That limits what can be said about the AI pivot. The mining-era record, however, sets a clear baseline for how TeraWulf's capacity statements have tracked against delivery. In early 2022 the company framed an 800 MW aggregate target by end-2025 (14108), built on a 90 MW NYPA allocation at Lake Mariner (14110, 14126) and 300 MW of nuclear-backed capacity at Nautilus (14111, 14121, 14130). Within five months Nautilus was restated as 200 MW initial with a 100 MW option (13715, 13716), and TeraWulf's economic share was a third of that (13714, 14112). The shrinkage was never presented as a revision. Lake Mariner followed the same pattern: a 100 MW first phase by year-end 2022 (14106) became a 50 MW first building (13881). Timelines fared worse than capacity. Every dated operational promise in the record carries a delayed status. That includes the June 2022 Nautilus ASIC start (14122, 14024), the 3,000 to 3,300 miner ramps at Lake Mariner (13877, 13880), the 800 MW goal and the Nautilus expansion. Equipment counts also drifted without explanation: 58,000 units recurs across 14247, 14049, 14025, 13978 and 13897, while 14142 cites 78,000 for the same year. The one steady figure is the 2028 zero-carbon pledge, repeated verbatim (14245, 13846, 13439, 13343) while 13847 reports 90 percent achieved. The record contains no independent corroboration of any of this: no permits, no NYISO queue positions, no site reporting. The NYPA contract is the only third-party-anchored input. Scaling Lake Mariner from roughly 90 MW of contracted supply toward 750 MW would require visible interconnection filings and new supply agreements, and none appear in this dataset. The same applies to the eleven announced sites with no stated capacity, from Abernathy to Morgantown. Over the next 90 days, watch four things: whether the $3.5 billion financing actually closes or remains a target; whether NYISO queue or NYPA filings support the 410 MW tranche; whether Q3 2026 results put energization dates on the 750 MW figure; and whether any of the placeholder sites acquire permitted megawatts.

The claim log supplied here covers only TeraWulf's 2022 to early 2023 bitcoin-mining buildout and breaks off in February 2023. The AI-hosting pivot shown in the project summary cannot be tested against dated claims in this read. That pivot comprises 410 MW initially, a later 750 MW campus target and a $3.5 billion financing target for the Anthropic-leased Lake Mariner facility.

The early record shows ambition shrinking in quiet steps. The headline goal was 800 MW of aggregate capacity by end-2025 (14108), now marked delayed. Nautilus was presented as a 300 MW nuclear-powered site (14111, 14121, 14130). By August 2022 it had become 200 MW initial capacity with a possible 100 MW expansion (13715, 13716), in which TeraWulf held a 33.33 percent stake (13714) rather than the 150 MW share stated earlier (14112). Miner procurement moved between 58,000 units (14123, 14049) and 78,000 (14142) without any reconciliation. Lake Mariner itself was anchored by a 90 MW NYPA allocation (14110, 14126).

Dated promises have a poor record. Almost every dated milestone in the log slipped:

Pennsylvania ASIC installation by mid-2022 (14122, 14024). 3,300 miners at Lake Mariner by July 2022 (13880). The first S19 XP batch by August (13877). 100 MW online by year-end 2022 (14106). Delivery of 58,000 miners during 2022 (14247, 13978, 13897). The Nautilus expansion (13716).

Revisions usually arrived as softer wording rather than acknowledged misses. The mid-2022 Nautilus start (14107) became sites "under development" (14053). The only claim repeated unchanged is the 100 percent zero-carbon target for 2028 (14245, 14026, 13846, 13439, 13411), with 90 percent reported in mid-2022 (13847).

Independent corroboration is thin. The record contains no permits, NYISO interconnection queue positions or on-the-ground reporting. The NYPA allocation is the only externally anchored figure. That gap matters more now that Lake Mariner is sized at 410 to 750 MW, several times the capacity the documented power supply supports.

Over the next 90 days, five things are worth watching:

Whether the $3.5 billion financing closes, and on what terms. Whether new NYPA allocations or NYISO queue filings back capacity beyond the original 90 MW. Whether local site-plan and building approvals in Niagara County match the stated construction pace. Whether the move from 410 to 750 MW is given dated phasing rather than a campus-level aspiration. Whether the capacity-less announced sites in Kentucky, Maryland and Texas acquire concrete figures, or stay as placeholders.

The 2022 pattern of optimistic dates followed by quiet downsizing is the baseline to test against.

The record supplied here is uneven. It is dense for 2022 and early 2023 and then breaks off at claim 13332, so the current-era figures at Lake Mariner (410 MW initially, 750 MW as the campus target, and a $3.5 billion financing for the Anthropic-leased facility) appear only in the project summary and cannot be tied to dated claims. Any read of the AI-era trajectory has to be built on how TeraWulf handled its bitcoin-era commitments.

That history shows capacity figures being reset downward without being formally withdrawn. Nautilus was presented as 300 MW of nuclear-powered capacity (14130, 14121). Five months later it became 200 MW initial capacity with a 100 MW expansion (13715, 13716), alongside a one-third equity stake (13714). The expansion was later marked delayed. The 800 MW aggregate target for 2025 (14108) also lapsed as delayed. Miner procurement was cited as both 58,000 (14247, 14123) and 78,000 units (14142) within two weeks. The 58,000 figure was then repeated through June 2022 (14049, 14025, 13978, 13897) before slipping.

Dates slipped with some regularity. Lake Mariner's 100 MW first phase for year-end 2022 (14106) was delayed. So were the 3,000- and 3,300-miner milestones set for mid-2022 (13877, 13880), and Nautilus ASIC installation by June 2022 (14122, 14024). The contracted base was more modest than the headline numbers: 90 MW under the NYPA agreement (14110, 14126) and a 50 MW first building (13881). The only target restated without change is 100% zero-carbon by 2028 (14245, 13439, 13343), and it remains well in the future.

The record contains no independent evidence, such as permits, NYISO queue positions, or site reporting, for either era. The move from 410 MW to 750 MW at Lake Mariner therefore rests on company statements alone. The dozen other announced sites, across Kentucky, Maryland, Texas and New York, carry no stated capacity or cost.

Over the next 90 days, watch four things. First, whether the $3.5 billion financing closes on the terms in the headline. Second, whether third-quarter disclosures tie the 750 MW figure to specific energization dates and a power source beyond the original 90 MW NYPA base. Third, whether interconnection filings support the expansion. Fourth, whether any of the unsized sites receive a capacity number, and whether it holds once it does.

The record supplied here stops at claim 13332 in February 2023, so the current Lake Mariner figures (410 MW initially, a 750 MW campus target and a $3.5 billion Anthropic-linked financing) cannot be tied to any claim id. What the record does show is how TeraWulf announced and delivered capacity during its bitcoin phase, and that history is the best baseline for judging the AI pivot.

Capacity moved in both directions without much explanation. Nautilus was first presented as a 300 MW nuclear-powered facility (14130, 14121, 14111), with TeraWulf's joint-venture share described as 150 MW (14112). By August 2022 that had become 200 MW of initial capacity with a 100 MW expansion option (13715, 13716), and TeraWulf's equity was stated as 33.33 percent (13714), which is below what the earlier 150 MW framing implied. Miner procurement appeared as 78,000 units (14142) and as 58,000 units (14247, 14049) in filings weeks apart. The 58,000 figure was then repeated through June 2022 (13897) with no reconciliation of the two numbers. Lake Mariner's 90 MW NYPA allocation (14110, 14126) sat under a 100 MW first-phase promise (14106), and the first building was only 50 MW (13881).

Dates slipped far more often than they held. The Pennsylvania ASIC start (14122, 14024), the Lake Mariner miner ramps (13880, 13877), mining commencement itself (13409), the IKONICS close (14119, 13980) and the 800 MW-by-2025 aggregate target (14108) are all marked delayed. Most were carried forward rather than formally revised. The only consistently repeated target is the 100 percent zero-carbon pledge for 2028 (14245, 13343), and it has no interim checkpoints.

The record contains no independent corroboration: no permits, no NYISO or PJM queue positions, no site reporting. For 2022 claims that gap can be forgiven. For a 750 MW AI campus that must exceed the historical NYPA allocation by a wide margin, it matters much more.

Over the next 90 days, watch four things. First, whether the Q3 filings reconcile the 410 MW and 750 MW figures and state a date for the first energized tranche. Second, whether the $3.5 billion financing actually closes, and on what terms. Third, whether NYPA or NYISO documents show incremental load approvals at Lake Mariner. Fourth, whether any of the eleven capacity-less announced sites, from Abernathy to the Kentucky and Maryland assets, gains a megawatt figure backed by a utility or regulator.

The record supplied here is truncated. It stops mid-entry at claim 13332 in February 2023, so the analysis below covers only TeraWulf's bitcoin-mining era. It cannot trace the path to the Anthropic-leased Lake Mariner build, which the project ledger lists at 410 MW initially and 750 MW as a campus target, with a $3.5 billion financing target. That gap matters because the capacity step-up is roughly eightfold over the 90 MW NYPA allocation first tied to the site (14110, 14126).

Within the visible window, capacity claims contracted as details firmed up. Nautilus was presented as a 300 MW nuclear-powered facility (14121, 14111, 14130), with TeraWulf's exposure framed as a 150 MW joint-venture share (14112). By August 2022 it had become a 200 MW initial build with a contingent 100 MW expansion (13715, 13716), and TeraWulf's stake was 33.33 percent (13714). The 800 MW aggregate target for 2025 (14108) is flagged as delayed, and nothing in the record shows it being formally withdrawn.

Dates slipped repeatedly, and the slips were often not acknowledged. The 58,000-miner delivery commitment was restated with an unchanged year-end 2022 deadline across five filings from March to June 2022 (14247, 14049, 14025, 13978, 13897), and all of them ended up delayed. The 78,000-unit procurement figure (14142) met the same fate. At Lake Mariner, a 100 MW first phase by year-end 2022 (14106) gave way to a 50 MW first building (13881). The near-term targets of 3,000 and 3,300 installed miners (13877, 13880) also missed. Pennsylvania ASIC installation by mid-2022 (14122, 14024) and a March 2022 mining start (13409) both slipped. The one constant is the 100 percent zero-carbon target for 2028, repeated unchanged from 14245 through 13343.

The record contains no independent corroboration of any of these claims: no permits, no NYISO or PJM queue positions, and no site reporting. The twelve other announced sites, spanning Texas, Kentucky, Maryland and New York, carry no stated capacity or cost.

Over the next 90 days, the key test is whether the $3.5 billion financing closes on the terms in the headline. Beyond that, watch for any interconnection or NYPA filing that supports load well above the original 90 MW, for local building permits at Lake Mariner, and for dated energization milestones for the 410 MW tranche. Any queue filings that attach real megawatts to the Kentucky, Maryland or Abernathy sites would also be informative. Given the 2022 pattern of repeating targets past their feasibility, an unchanged schedule without supporting filings should be read cautiously.

TeraWulf's record here is effectively a 2022 bitcoin-mining ledger with a 2026 AI headline attached, and the gap between them is the most important fact. Every dated facility claim on file ends in February 2023. The Lake Mariner Anthropic AI Factory's 410 MW initial and 750 MW campus figures, and the $3.5 billion financing target, appear only in the project summary and carry no claim ids. The current AI-era pace therefore cannot yet be tested against the company's own prior statements.

On capacity, the trajectory runs from ambitious to narrower. The March 2022 filings set an 800 MW aggregate target for 2025 (14108), which is marked delayed. Nautilus was framed as 300 MW of nuclear-adjacent capacity (14121, 14111, 14130). By August 2022 that had become 200 MW initial with a 100 MW expansion (13715, 13716), and the expansion slipped past its May 2024 marker. Lake Mariner started from a 90 MW NYPA allocation (14110, 14126). Its first phase was pitched at 100 MW by year-end 2022 (14106), but the first building was later described at 50 MW (13881). The move to 410 MW, and then 750 MW, is roughly an eightfold expansion over that original power base, and nothing in this record yet explains where the additional power comes from.

On dates, slippage is the pattern, and it was often unacknowledged. The 58,000-miner delivery commitment was restated almost verbatim from March through June 2022 (14247, 14049, 14025, 13978, 13897) before disappearing from later filings. Its only visible successor is a 20,000-unit contribution to Nautilus (13713). The June 2022 Nautilus ASIC start (14122, 14024) and the Lake Mariner miner milestones (13880, 13877) all lapsed. By February 2023 the language had retreated to an undated 'initial buildout completion' (13332). One target was held steadily: 100 percent zero-carbon by 2028 (14245 through 13343), with 90 percent reported mid-2022 (13847).

No permits, interconnection queue entries or on-the-ground reporting are in this record. The announced pace is therefore unverified rather than contradicted.

Over the next 90 days, four things matter. First, the claims behind the 410 MW and 750 MW figures, and whether each carries an energization date. Second, whether the $3.5 billion financing closes or is only described as a target. Third, any NYISO queue filings or utility agreements that take Lake Mariner meaningfully beyond its original 90 MW. Fourth, whether any of the Kentucky, Maryland or Texas sites moves beyond a bare announcement.

The TeraWulf record splits into two very different eras, and the gap between them is the most important thing to understand about the Lake Mariner trajectory. Every dated capacity and timeline claim on file ends in February 2023, when the company was still a bitcoin miner. The 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion Anthropic-linked financing appear only in the project summary, with no claim ids behind them. Any read of the AI-factory pace therefore rests on the company's earlier delivery behavior, not on a documented record of the current build.

That earlier behavior is instructive. Lake Mariner began as a 90 MW NYPA allocation (14126, 14110), with a first 100 MW phase promised for year-end 2022 (14106). By mid-2022 the near-term framing had narrowed to a single 50 MW building (13881), and both miner-energization milestones slipped (13877, 13880). Nautilus shows a similar contraction. It was pitched at 300 MW (14130, 14121, 14111) and restated in August 2022 as 200 MW initial capacity, with the remaining 100 MW deferred into an expansion that also slipped (13715, 13716). TeraWulf's economic share was a one-third equity stake (13714).

The larger portfolio promises faded rather than being formally withdrawn. The 800 MW aggregate target for 2025 (14108) was never restated after March 2022. The 58,000-miner delivery commitment was repeated through June 2022 (14247, 14049, 13978, 13897) and then disappeared, as did the 78,000-unit procurement figure (14142). By 2023 the language had shifted to an unquantified "initial buildout completion" (13332). The one target held consistently is 100 percent zero-carbon power by 2028 (14245 through 13343), and it is the least operationally demanding of the set.

The record contains no independent corroboration of any kind: no permits, no NYISO interconnection queue positions, and no on-the-ground reporting. Nothing on file supports a move from 90 MW of contracted NYPA power to 410 or 750 MW of deliverable load.

Over the next 90 days, four things matter. The first is whether the $3.5 billion financing closes, and on what draw conditions. The second is NYISO queue or NYPA allocation filings that would substantiate the additional load. The third is local site-plan and building approvals in Niagara County. The fourth is whether the company reconciles 410 MW against 750 MW with dated phase-energization targets. The Kentucky and Maryland sites (Justified, Muskie, Chesapeake) remain capacity-free announcements and should be treated as options, not pipeline.

The shape of this record is a company whose announced scale has outrun its record of hitting dates, though its misses read as slippage rather than abandonment. The 2022 framing was a fleet-wide 800 MW by year-end 2025 (14108), assembled from a 90 MW NYPA allocation at Lake Mariner (14110, 14126), a first 50 MW building (13881), and a 300 MW Nautilus position (14121, 14111, 14130). That Nautilus figure was quietly restated by August 2022 as 200 MW initial with 100 MW recast as potential expansion (13715, 13716) — a de-scoping presented as optionality, which is the single most instructive move in the file. Today the same Lake Mariner site alone carries a 410 MW initial disclosure scaling to a 750 MW campus target, meaning one location now nearly absorbs the entire prior fleet ambition while 14108 has never been formally retired.

The date discipline is weak but not erratic. Equipment commitments of 58,000 units by year-end 2022 were restated at least six times (14247, 14123, 14049, 14025, 13978, 13897) and once inflated to 78,000 (14142) inside the same window before landing delayed. Energization promises slipped in weeks-to-quarters, not years: 3,000 machines by August 2022 (13877), 3,300 by Q3 (13880), Pennsylvania ASIC installation by June 2022 (14122, 14024), Nautilus mid-year start (14107). The IKONICS combination was restated against a December 2021 date across three filings (14119, 13980, 13408). The only claim held verbatim for four years is the 2028 zero-carbon target (14245, 13846, 13826, 13439, 13343), which is durable precisely because nothing tests it yet.

Independent corroboration is the gap. Nothing in this record ties the 410 or 750 MW figures to an interconnection queue position, Article VII filing, or local permit, and neither number carries a promised-by date at all — they arrive as project metadata rather than as a trackable commitment. Twelve announced sites, from Abernathy to Morgantown Generating Station to Boyd-Greenup, disclose no megawatts and no cost; treat those as site control, not capacity.

Over the next ninety days, watch three things: whether the $3.5 billion financing against the Anthropic lease actually closes and on what tenant-credit terms; whether 410 MW is broken into dated energization tranches distinct from the 750 MW aspiration; and whether any announced site converts into a signed power agreement with a disclosed number attached.

The through-line here is a company that has been consistently good at contracting for power and consistently late at putting equipment behind it. The contractual anchors held: the NYPA supply arrangement at Lake Mariner (14110, 14126) and the eleven-year Susquehanna offtake underpinning Nautilus (14113, 14129) were real, counterparty-backed instruments. Nearly everything downstream of them slipped. The 2022 miner-delivery commitments were restated five times in four months (14247, 14049, 14025, 13978, 13897) and inflated from 58,000 to 78,000 units within the same filing window (14142), all landing in the delayed column. The ASIC installation date for Pennsylvania (14024, 14122), the first 100 MW phase at Lake Mariner (14106), and the modest 3,000-to-3,300 unit energization targets (13877, 13880) were missed at a scale that suggests the dates were aspirational at announcement rather than derailed afterward.

On capacity, the pattern is headline-high then quietly trimmed. Nautilus was disclosed three separate times at 300 MW (14111, 14121, 14130) before appearing as 200 MW of initial electrical capacity with the balance recast as a potential 100 MW expansion (13715, 13716) that has been past due since May 2024 without visible acknowledgement. The 800 MW aggregate target for 2025 (14108) simply lapsed. It is worth noting that the present 410 MW initial disclosure scaling to a 750 MW campus target is attached to the same Lake Mariner site whose historical claims here are all bitcoin-era; the AI framing is new, the substation is not, and the 750 MW figure currently carries no date at all.

Independent corroboration is the weak point in this record. There are no permits, interconnection queue positions, or field observations in evidence, and the claim history goes quiet after February 2023 while the project roster expanded to thirteen sites, eleven of which carry neither capacity nor cost. That is announcement velocity outrunning substantiation.

Over the next ninety days, the useful signals are narrow: whether the 3.5 billion dollar financing actually closes and against what collateral; whether 410 MW is described as energized rather than contracted; whether NYISO queue entries appear for the incremental Lake Mariner load; and whether Abernathy, Cayuga, or the Kentucky and Maryland sites acquire megawatt figures or remain place names.

TeraWulf's capacity disclosures have grown roughly an order of magnitude while the unit of account quietly changed underneath them. The 2022 cohort measured itself in mining megawatts and machine counts: an 800 MW aggregate target (14108), 90 MW of NYPA supply at Lake Mariner (14110, 14126), a 50 MW first building (13881), and a Nautilus stake sized variously at 200 MW installed (13715), 300 MW accessible (14111, 14121, 14130), with a 100 MW expansion option (13716). The current framing — 410 MW initially disclosed at Lake Mariner rising to a 750 MW campus target against a $3.5 billion financing — is larger but not obviously comparable, because nothing in the record specifies whether these are gross interconnection, utility-delivered, or critical IT load. That ambiguity is the single most important thing to resolve before treating the increase as growth rather than reclassification.

On dates, the pattern is restatement without acknowledgement. The 58,000-miner delivery commitment was re-promised for year-end 2022 at least five times across 14247, 14049, 14025, 13978 and 13897, each instance carrying the same deadline and each now flagged delayed. Nautilus ASIC installation was set for mid-2022 twice (14122, 14024) and slipped. Lake Mariner's 100 MW first phase (14106), its 3,300-miner Q3 target (13880) and its 3,000-unit S19 XP batch (13877) all missed. Even the closed IKONICS transaction appears with drifting promise dates of 2021-12-13 and 2021-12-14 (14119, 13980, 13408), which suggests the timeline record is reconstructed after the fact rather than maintained against original guidance.

Independent corroboration is the gap. Eleven of the thirteen listed sites — Abernathy, Cayuga, La Lupa, Akela, Lake Hawkeye, the Kentucky and Maryland properties — carry no capacity or cost figure whatsoever, meaning they are announcements, not projects, until a queue position or permit says otherwise. Nothing in this record ties any announced site to a filed interconnection request.

Over the next ninety days, watch whether the 410 MW figure is restated to 750 in the Q3 filing and on what basis; whether NYISO queue entries appear for Cayuga, La Lupa or Lake Hawkeye; whether the Maryland and Kentucky sites produce any state-level permit activity; and whether the $3.5 billion financing closes or is re-announced at a revised size.

The dominant pattern here is that capacity headlines have scaled roughly an order of magnitude faster than the delivery record beneath them. Lake Mariner began as a 90 MW NYPA supply arrangement (14110, 14126) with a first building sized at 50 MW (13881); the same site is now presented at 410 MW initially and 750 MW as a campus target. The land and grid position make that physically plausible, but note the arithmetic: the 750 MW single-site figure now exceeds the 800 MW aggregate portfolio target the company set for the whole business by year-end 2025 (14108), which the record carries as missed. A number that once described the entire company now describes one campus, and the shortfall has been absorbed into a larger story rather than reconciled against the old one.

On dates, nearly every claim with an explicit deadline has slipped, and the slippage is concentrated in exactly the claims the company controlled. The 58,000-unit delivery commitment was restated at least six times across 2022 (14247, 14049, 14025, 13978, 13897) and never met; the 78,000-unit variant (14142) went the same way. Pennsylvania ASIC installation by mid-2022 (14024, 14122), the first 3,000 to 3,300 miners at Lake Mariner (13877, 13880), and the Nautilus expansion to 100 MW by May 2024 (13716) all lapsed. The IKONICS combination has been re-dated repeatedly (14119, 13980, 13408). What has held are the counterparty-backed items: the NYPA agreement, the Talen joint venture and the eleven-year Susquehanna supply term (14112, 14113, 14129). The distinction is informative. When a third party is on the other side of the contract, the claim survives; when the date is self-declared, it moves.

Independent corroboration is thin in this record. Twelve announced sites — Abernathy, Cayuga, La Lupa, Akela, Lake Hawkeye, Justified Data, Chesapeake, Muskie, Morgantown, Hancock County, Boyd-Greenup, Nautilus — carry no capacity and no cost, which is the profile of site-control vehicles rather than funded construction. The $3.5 billion figure remains a financing target, not a close.

Over the next ninety days, the things worth watching are narrow. Whether that financing actually closes and at what cost of capital. Whether the next quarterly filing describes 410 MW as contracted and energized or as targeted, and whether 750 MW appears anywhere outside investor materials. Whether any NYISO interconnection position appears for Lake Mariner capacity beyond the existing envelope. And whether any of the unsized sites converts from a named entity into a filed interconnection request or a county permit. Absent that, they remain announcements.

TeraWulf's capacity disclosures have moved in one direction on paper and a different one in practice. The Lake Mariner story began as a 90 MW NYPA power arrangement (14110, 14126), sat inside an 800 MW aggregate target for 2025 that the record itself marks as missed (14108), and now reappears as a 410 MW initial Anthropic-leased disclosure stepping up to a 750 MW campus target against a $3.5 billion financing objective. That is genuine growth in ambition, but the escalation happens without the 800 MW shortfall being reconciled anywhere in the record. Nautilus shows the same pattern in miniature and more clearly: 300 MW was asserted repeatedly in early 2022 (14111, 14121, 14130), then restated in August 2022 as 200 MW initial electrical capacity with the remaining 100 MW recast as a potential expansion carrying a 2024 date that also slipped (13715, 13716). The number did not shrink in any announcement; it was re-sliced.

Dates have fared worse than capacity. Essentially every near-term commitment from 2022 is flagged delayed: ASIC installation at the Pennsylvania site by mid-2022 (14024, 14122), the first 100 MW Lake Mariner phase by year-end (14106), 3,000 and 3,300 miners energized in the summer (13877, 13880), and the 58,000 and 78,000 unit procurement commitments (14025, 14049, 13897, 13978, 14142). Corporate milestones drifted similarly, including the IKONICS combination against its December 2021 target (14119, 13980, 13408) and both NYDIG agreements (13412, 13414). The one durable long-dated claim, 100 percent zero-carbon energy by 2028 (14245, 14026, 13846, 13439, 13411), has been restated unchanged for four years, which tells us little either way.

On corroboration, this record contains no permits, interconnection queue positions, or site reporting at all. Eleven projects, from Abernathy to Boyd-Greenup, carry announced status with no capacity and no cost disclosed. Until NYISO, ERCOT, and PJM queue entries and county-level permits are attached to them, they are site names rather than pipeline.

Over the next 90 days, watch whether the $3.5 billion financing actually closes, whether the 750 MW target ever acquires an energization date, and whether any announced-only site graduates to a disclosed megawatt figure. A new site announcement arriving before Lake Mariner's first Anthropic tranche is confirmed energized would continue the established pattern.

The striking feature of this record is the discontinuity between what is documented and what is currently being sold. Every timestamped claim in the file dates from 2022 and early 2023 and concerns bitcoin mining; the Anthropic-linked figures now anchoring the equity story — 410 MW at initial disclosure, a 750 MW campus target, and a $3.5 billion financing referenced to a July 13, 2026 headline — appear only as project metadata, unattached to any dated, sourced disclosure. That asymmetry is itself the finding: the escalation from an 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881) to a 750 MW campus has happened faster than the disclosure trail supporting it.

On dates, the earlier record is unambiguous and unflattering. The 100 MW first phase by year-end 2022 (14106), the 3,000 and 3,300 miner milestones (13877, 13880), the Pennsylvania ASIC installation start (14024, 14122), and the 58,000-unit delivery commitment restated across at least five filings (14247, 14049, 14025, 13978, 13897) all slipped. The 800 MW aggregate target for 2025 (14108) is now nine months past due and has been functionally replaced, not retired, by the single-site 750 MW framing. Nautilus shows the same pattern in miniature: 300 MW asserted three times in March 2022 (14111, 14121, 14130), then quietly recast in August as 200 MW initial plus 100 MW "potential" (13715, 13716), with the potential tranche subsequently delayed. Revision by reframing, rather than acknowledgement, is the house style.

Independent corroboration is thin by construction. Eleven of thirteen listed projects carry no capacity and no cost — Abernathy, Cayuga, La Lupa, Akela, Lake Hawkeye, Justified Data, Chesapeake, Muskie, Morgantown, Hancock County, Boyd-Greenup. Several share the Lake Mariner address, which is consistent with entity registrations and site control rather than construction. Nothing here should be counted as pipeline until it appears in a queue.

Over the next 90 days, the load-bearing checks are NYISO interconnection positions and Town of Somerset permitting for Lake Mariner against the 410-versus-750 gap, whether the $3.5 billion financing actually closes or converts to a term sheet, ERCOT large-load filings for Abernathy, and Maryland PSC activity at Morgantown. Also watch whether the 100% zero-carbon-by-2028 pledge, reaffirmed six times (14245, 13846, 13826, 13439, 13411, 13343), survives an ERCOT and PJM footprint.

The arc here is a company whose announced ceiling has risen much faster than its demonstrated throughput. The 2022 framing was an 800 MW aggregate target by 2025 (claim 14108), assembled from a 90 MW power agreement at Lake Mariner (claims 14110, 14126) and a 300 MW share of the Talen joint venture (claims 14111, 14121, 14130). That Nautilus figure was quietly cut to 200 MW initial with a 100 MW option by August 2022 (claims 13715, 13716), and the option itself is logged as missed against a May 2024 date. So the component numbers shrank while the headline grew: Lake Mariner is now presented at 410 MW with a 750 MW campus target, a single site larger than the entire 2022 fleet ambition, even though the 2025 800 MW milestone lapsed without a public reconciliation.

On dates, the pattern is slippage followed by re-baselining rather than correction. The 100 MW first phase due end-2022 (claim 14106), the 58,000-miner delivery commitments (claims 14247, 14049, 14025, 13897, 13978), the larger 78,000-unit procurement (claim 14142), the first 3,000 to 3,300 machines due mid-2022 (claims 13877, 13880), and the Pennsylvania ASIC installation due June 2022 (claims 14024, 14122) all ran late. Notably, the February 2023 filings restate mining commencement and buildout completion with the dates removed entirely (claims 13332, 13342, 13409), which is how a missed deadline disappears without being acknowledged. The one constant is the 2028 zero-carbon pledge, repeated near-verbatim across six disclosures (claims 14245, 14026, 13846, 13826, 13439, 13343), while claim 13847 concedes actual performance at 90 percent.

Independent corroboration is absent from this record, and that absence matters. There are no permits, interconnection queue entries, or site reporting attached to the 410 MW or 750 MW figures, and eleven of the thirteen listed projects carry no capacity or cost at all. That is announcement surface, not delivery evidence.

Over the next 90 days, watch for NYISO interconnection and local permit filings sized to 410 MW at Lake Mariner, whether the 750 MW target ever acquires a date, whether the $3.5 billion financing actually closes, and whether the Kentucky and Maryland sites receive any capacity figure. Also worth noting: this claims record stops in February 2023 while the announcements run to mid-2026, so the recent period is currently unaudited here.

The shape of this record is a company that has repeatedly been directionally right about where it was going and consistently wrong about when it would get there. Lake Mariner's disclosed capacity has ratcheted upward in a way that is normal for a site under genuine development: a first building at 50 MW (13881), a 90 MW NYPA supply arrangement (14110, 14126), a 100 MW first phase (14106), and now a 410 MW initial AI disclosure against a 750 MW campus target. That escalation is credible precisely because it is incremental and tied to an anchor tenant and a stated $3.5 billion financing target. The countervailing signal sits at Nautilus, where 300 MW was asserted three separate times in a single 2022 filing (14111, 14121, 14130) and then quietly restated as 200 MW of initial electrical capacity with the residual 100 MW recast as a potential expansion (13715, 13716) that itself slipped. Nobody reconciled the two numbers; the larger figure simply stopped appearing.

On dates, the pattern is unambiguous. The 58,000-miner delivery commitment was restated at least five times across 2022 and flagged delayed every time (14247, 14049, 14025, 13978, 13897). ASIC installation by mid-2022 slipped (14122, 14024), as did the 3,300- and 3,000-unit energization milestones (13880, 13877) and the March 2022 mining start (13409). Most telling is the 800 MW aggregate target for year-end 2025 (14108): now past due, unmet, and apparently superseded by AI-era framing rather than formally retired.

Independent corroboration is where this file is thinnest, and that is itself the finding. Twelve of the thirteen listed sites carry no capacity, no cost, and no date. Abernathy, Cayuga, La Lupa, Akela, Lake Hawkeye, Justified Data, Chesapeake, Muskie, Morgantown, Hancock County and Boyd-Greenup read as site-control and single-purpose-entity announcements, not permitted projects. Only Lake Mariner is under construction.

Over the next ninety days the useful tests are narrow: energized megawatts versus contracted megawatts in the next quarterly disclosure; whether the $3.5 billion financing actually closes and on what recourse terms; whether 410 MW is restated toward 750 MW with a date attached rather than without one; and whether any named shell site acquires its first interconnection queue position or local approval. Absent that, treat the portfolio count as optionality, not pipeline.

TeraWulf's headline capacity has grown roughly an order of magnitude since 2022, but the growth is not continuous — the denominator changed. The 2022 framing was an 800 MW aggregate target by year-end 2025 (14108), built from miner counts rather than critical IT load, and that target is on the record as missed. The current framing is an Anthropic-leased AI factory at Lake Mariner disclosed at 410 MW and subsequently described against a 750 MW campus, financed toward $3.5 billion. Those are not the same kind of number, and the shift lets a missed mining-era target retire quietly rather than be reconciled.

The date record is the more useful signal. Lake Mariner's first 100 MW phase was promised by year-end 2022 (14106) and slipped; the 3,300- and 3,000-unit energization milestones (13880, 13877) slipped; every version of the 58,000-miner delivery commitment slipped across four separate restatements (14247, 14049, 14025, 13978, 13897), as did the 78,000-unit variant (14142). Nautilus is the clearest case of a quiet downward revision: 300 MW was asserted repeatedly (14111, 14121, 14130), the facility came in at 200 MW initial electrical capacity (13715), and the 100 MW expansion that would close the gap was itself pushed and missed (13716). The 100 percent zero-carbon-by-2028 pledge has been restated verbatim for four years without interim benchmarks (14245, 13846, 13826, 13439, 13343), which makes it unfalsifiable until the terminal date — and harder to square with the Kentucky and Maryland sites now on the books.

Independent corroboration is where this record is thinnest. Thirteen sites sit at announced status with no capacity or cost attached, several under names that read as project entities rather than disclosed campuses. Announcement without an interconnection position is not a schedule.

Over the next 90 days, the things worth checking are specific: whether the $3.5 billion financing actually closes or converts to a commitment letter; whether 410 MW resolves into dated phase energizations rather than a campus aspiration; NYISO queue movement at Lake Mariner and Cayuga, PJM at Morgantown, ERCOT at Abernathy; and county-level building and air permits in Boyd, Greenup, Hancock and Charles counties. If capacity numbers attach to those announced sites before permits do, treat the sequence as the finding.

The claim record supplied here stops in February 2023, so it documents TeraWulf's bitcoin-mining origins rather than the Anthropic-leased AI build now on the project list. The 410 MW initial disclosure, the 750 MW campus target and the $3.5 billion financing figure appear only in the project summary and cannot yet be traced to dated claims. That gap is the first thing a reader should weigh.

Within the visible window, the pattern is ambitious headline capacity followed by quiet re-scoping. The company set an 800 MW aggregate target for 2025 (claim 14108), now marked delayed. The Nautilus JV was framed at 300 MW (claims 14121, 14130, 14111), with TeraWulf's share at 150 MW (claim 14112). By August 2022 it had been restated as a 200 MW initial facility with a 100 MW expansion (claims 13715, 13716) and a one-third equity stake (claim 13714). The expansion itself later slipped. Lake Mariner followed the same arc. A 100 MW first phase by year-end 2022 (claim 14106) sat against a 90 MW NYPA power agreement (claims 14110, 14126). It then narrowed to a single 50 MW building (claim 13881), and the near-term miner counts of 3,000 and 3,300 units (claims 13877, 13880) both slipped. The 58,000-miner delivery commitment was repeated across at least five filings (claims 14247, 14049, 14025, 13978, 13897) and is uniformly delayed. It sat alongside a larger 78,000-unit procurement figure (claim 14142). Neither was ever reconciled in the record. The 2028 zero-carbon target (claims 14245, 13411, 13343) is the only commitment restated consistently, and it is not yet testable.

The record contains no independent corroboration: no permits, no NYISO or PJM queue entries, no site reporting. Nautilus is also mapped to Tompkins County, New York, while its claims describe a Pennsylvania facility powered by Susquehanna. That mismatch should be resolved before any Nautilus claim is used in trajectory work. The Kentucky and Maryland sites carry no capacity or cost data at all.

Over the next 90 days, watch four things. First, whether the $3.5 billion Anthropic financing closes on the reported terms. Second, whether filings commit to 410 MW or 750 MW as the contracted figure. Third, whether NYPA allocation or NYISO interconnection disclosures support power beyond the historical 90 MW Lake Mariner agreement. Fourth, whether any of the announced Kentucky or Maryland sites acquires a dated capacity claim. Given the delay rate on every dated promise above, each of these deserves independent verification rather than acceptance at announcement.

The record supplied here is truncated. The itemized claims stop in February 2023, so the Lake Mariner AI repositioning, with its 410 MW initial disclosure, 750 MW campus target and $3.5 billion financing headline, appears only in the project summary and cannot be traced to dated claim ids. What follows reads the documented bitcoin-era trajectory and treats the AI-era figures as unverified in this dataset.

Capacity rhetoric was trimmed early. The March 2022 aggregate target of 800 MW by year-end 2025 (14108) sat on a 90 MW NYPA supply at Lake Mariner (14110, 14126) and a Nautilus facility pitched at 300 MW (14121, 14130, 14111). By August 2022 Nautilus had been restated as 200 MW initial, with the other 100 MW recast as a contingent expansion (13715, 13716). TeraWulf's position had also become a one-third equity stake (13714) rather than the 150 MW joint-venture share described earlier (14112). Lake Mariner's first increment narrowed from a 100 MW phase (14106) to a 50 MW first building (13881). Miner procurement also moved between 58,000 units (14247, 14025, 13897) and 78,000 (14142) without reconciliation.

Dates held poorly. The Nautilus ASIC installation slated for mid-2022 (14122, 14024) slipped. So did the 3,000-unit Lake Mariner batch due by August 2022 (13877) and the 3,300 miners promised for the third quarter (13880). The 100 MW year-end 2022 phase (14106) and the 2024 Nautilus expansion (13716) are both marked delayed. By early 2023, filings described "initial buildout" at both sites without a date (13332), which is a quiet softening rather than an acknowledged revision. The 2028 zero-carbon target has been restated consistently (14245, 13846, 13411, 13343), making it the one stable commitment on record.

No permits, interconnection queue entries or site reporting appear in this record, so the announced pace is unsupported by independent evidence here. Most of the newer sites, including Abernathy, Cayuga, the Kentucky and Maryland properties, and Morgantown Generating Station, carry no capacity or cost at all. Nautilus's location (Tompkins County, NY) also conflicts with its Pennsylvania and Susquehanna description in claims 14112 and 14113, which should be corrected.

Over the next 90 days, the key question is whether the $3.5 billion Anthropic-linked financing closes on disclosed terms. Beyond that, watch for NYISO queue or NYPA allocation filings sized to the 410 MW, and later 750 MW, figures. Third-quarter results should also be checked for a dated energization schedule instead of a campus-level target.

The record supplied here ends in February 2023, so this read covers TeraWulf's bitcoin-mining era in detail and can only bracket the current AI-campus story. The 410 MW initial and 750 MW campus figures for Lake Mariner, and the $3.5 billion financing target reported on July 13, 2026, appear only in the project roster. No dated claims in this extract carry them, so neither their consistency nor their revision history can be judged yet.

What the early record shows is a pattern of ambitious sizing followed by quiet contraction. The aggregate 800 MW by 2025 goal (14108) is flagged delayed. Nautilus was pitched repeatedly at 300 MW (14111, 14121, 14130), then recast as 200 MW initial with a 100 MW expansion (13715, 13716), and the expansion date also lapsed. TeraWulf's own exposure narrowed further: from a 150 MW joint-venture share (14112) to a 33.33 percent stake backed by a 20,000-miner contribution (13713, 13714). At Lake Mariner the first phase was framed as 100 MW online by year-end 2022 (14106), yet the concrete disclosures are a 90 MW NYPA allocation (14110, 14126) and a 50 MW first building (13881). Near-term milestones slipped in sequence: 3,300 miners by Q3 (13880), 3,000 S19 XPs by August (13877), Pennsylvania ASIC installation by mid-2022 (14024, 14122), and 58,000 unit deliveries (14049, 14247). The 78,000-unit procurement figure (14142) was never reconciled with the 58,000 commitment. The one stable promise is 100 percent zero-carbon power by 2028 (14245, 13343, 13439), restated verbatim for a year.

No permits, interconnection queue entries, or site reporting are present, so the announced pace cannot be independently corroborated. There is also a mapping problem. The Nautilus claims describe a Susquehanna, Pennsylvania facility (14113, 14129), while the roster places Nautilus in Tompkins County, New York. In addition, 2022 mining claims are filed under the Anthropic AI Factory, which conflates two distinct projects on the same site.

Over the next 90 days, the key test is whether the $3.5 billion financing closes on stated terms. The quarterly filing will show whether 410 MW or 750 MW is the contracted figure, and NYISO and NYPA records should show whether incremental load at Lake Mariner is actually queued. For the Kentucky and Maryland sites, which are announced with no capacity, the thing to watch is whether any of them attaches a megawatt figure to a utility agreement.

TeraWulf's record at Lake Mariner reads as a small site repeatedly reframed at larger scale, with delivery dates mostly failing to keep up. The earliest anchor is the 90 MW NYPA supply (14110, 14126), tied to a 100 MW first phase promised by year-end 2022 (14106) and a first 50 MW building (13881). The project header now lists 410 MW, then a 750 MW campus target, plus a $3.5 billion financing target for the Anthropic-leased facility. That is roughly an eightfold increase over the original power basis. The claim log supplied here ends in February 2023, so these AI-era figures cannot be tied to specific claim ids. The step-up is visible only through the project summary, not through dated, attributable statements.

The dated history does not support treating the new targets at face value. Nearly every 2022 commitment with a deadline is marked delayed: the Lake Mariner first phase (14106), the 3,300-miner Q3 2022 goal (13880) and the related S19 XP batch (13877), mid-2022 ASIC installation at Nautilus (14122, 14024), the 800 MW aggregate target for 2025 (14108), and the Nautilus expansion due by May 2024 (13716). Some revisions were never acknowledged. Nautilus appeared as a 300 MW facility (14111, 14121, 14130). By August 2022 it had become 200 MW initial plus 100 MW potential (13715, 13716), in which TeraWulf held a one-third stake (13714). Miner procurement was cited as 58,000 units in most filings (14123, 14049, 13897) but 78,000 in one (14142). The only target restated consistently is the 2028 zero-carbon goal (14245, 13846, 13343). It carries little delivery risk given the 90 percent level already claimed (13847).

The record contains no independent evidence for the current pace. It includes no NYISO interconnection queue positions, no NYPA allocation above 90 MW, no local site-plan or building permits, and no on-the-ground reporting. Eleven other listed sites, from Abernathy to Boyd-Greenup, have no stated capacity or cost, so they cannot yet be assessed as pipeline.

Over the next 90 days, watch four things. First, whether the $3.5 billion financing closes, and on what terms. Second, whether TeraWulf files or discloses a NYISO interconnection request or new NYPA power allocation large enough to support 750 MW rather than 410 MW. Third, whether Q3 2026 reporting sets dated energization milestones for the first Anthropic tranche. Fourth, whether the 410 MW figure is quietly replaced by the 750 MW target without a phased schedule, repeating the Nautilus pattern.

The claim record supplied here breaks off mid-entry in February 2023, so this read covers TeraWulf's bitcoin-mining era in detail. The current AI-hosting program at Lake Mariner appears only in the project header, without claim ids. That gap matters, because the headline numbers for that program (410 MW, then a 750 MW campus target, and a $3.5 billion financing tied to the Anthropic-leased facility) cannot yet be tested against a claim-level history.

On scale, the early pattern is ambition first, then quiet trimming. The 800 MW aggregate target for 2025 (14108) is marked delayed. Nautilus was repeatedly described as a 300 MW nuclear-powered site (14111, 14121, 14130). By August 2022 it had been reframed as 200 MW initial capacity, plus a 100 MW expansion that itself slipped (13715, 13716). That is a one-third reduction presented as phasing. Miner procurement is internally inconsistent: most filings say 58,000 units (14247, 14049, 14025, 13978, 13897), while 14142 cites 78,000 against the same year-end 2022 deadline. Both are marked delayed.

On timing, the near-term milestones almost uniformly slipped. These include the 100 MW first phase at Lake Mariner by year-end 2022 (14106), the 3,000 and 3,300 miner online targets for mid-2022 (13877, 13880), and the Pennsylvania ASIC installation by June 2022 (14122, 14024). None were visibly re-baselined in later claims. Several 2023 filings restate completed agreements with backdated promise dates (13409, 13412, 13414), which inflates the delayed count mechanically rather than signaling new slippage. The one claim restated consistently is the 100 percent zero-carbon target for 2028 (14245, 13846, 13439, 13343). It sits against a self-reported 90 percent figure (13847) and has no dated interim milestones.

Independent evidence is absent from this record. The only hard power figure is the 90 MW NYPA allocation (14110, 14126), with a 50 MW first building (13881). Moving from 90 MW to 410 MW, and then 750 MW, at Lake Mariner requires NYISO interconnection and large-load studies, Somerset and Niagara County site-plan and SEQRA approvals, and substation work. None of these is documented here.

Over the next 90 days, watch four things. First, whether the $3.5 billion financing closes and on what draw schedule. Second, NYISO queue or large-load filings sized to the 750 MW figure. Third, local permit activity for new Lake Mariner buildings. Fourth, whether any of the eleven capacity-less announced sites, from Abernathy to Morgantown to the Kentucky campuses, receives a disclosed megawatt figure or is quietly dropped.

The claim record supplied here stops in March 2023, so it cannot confirm or refute the 2026 AI-factory figures in the project list. That list shows Lake Mariner growing from a 410 MW initial disclosure to a 750 MW campus target, with a $3.5 billion financing target, but no claim ids support those numbers. They should be treated as headline-level until the underlying filings are attached. What the visible record does show is an early pattern that is useful context. In March 2022 TeraWulf framed 800 MW of aggregate capacity by end-2025 (14108), a 100 MW first phase at Lake Mariner by year-end 2022 (14106), and 300 MW at Nautilus with a 150 MW joint-venture share (14112, 14121). By August 2022 Nautilus had become a 200 MW initial facility with a contingent 100 MW expansion (13715, 13716), and TeraWulf's stake was 33.33 percent (13714), a meaningful reduction in attributable capacity that was not framed as a revision. Hardware promises followed the same arc. A 58,000-miner delivery commitment for 2022 was repeated at least six times (14247, 14049, 14025, 13978, 13897), alongside a 78,000-unit procurement figure (14142). Meanwhile the June 2022 operational targets were 3,000 to 3,300 machines and a 50 MW first building (13877, 13880, 13881). By early 2023 the dated language had given way to undated references to initial buildout completion and buildout in progress (13332, 13158). The 100 MW and 800 MW targets were not reaffirmed and not explicitly withdrawn. The delayed tags on 13409, 13412, 13414 and 13408 attach to retrospective descriptions of already-completed events, so they read as a classification artifact rather than evidence of slippage. The one target restated consistently is 100 percent zero-carbon energy by 2028 (14245 through 13343). No permits, interconnection-queue entries or site reporting appear in the record, so the announced pace is unverified on every project. Over the next 90 days, the most useful signals are whether the $3.5 billion Anthropic financing actually closes and on what terms. Watch for NYISO queue or NYPA allocation filings that would support 750 MW at Lake Mariner, and for the Q3 2026 results in November. The key question there is whether 750 MW is described as contracted power or aspiration. Also worth watching are any ERCOT, Kentucky PSC or PJM filings that would move Abernathy, the Kentucky sites or Morgantown beyond the announced stage.

The claim record supplied for this read ends mid-entry at March 2023, so it covers TeraWulf's bitcoin-mining era but not the dated claims behind the Lake Mariner Anthropic AI Factory. The 410 MW, 750 MW and $3.5 billion figures appear only in the project header, so the AI-era trajectory cannot yet be tested against its own promise history.

What the record does show is a pattern of headline capacity being quietly narrowed. The 2022 filings framed Nautilus as 300 MW of nuclear-powered access (14111, 14121, 14130), with TeraWulf's share described as 150 MW of the joint venture (14112). By August 2022 the facility was 200 MW initial, with 100 MW of expansion only 'potential' (13715, 13716, 12940), and TeraWulf's position was a 33.33 percent equity stake (13714). The miner procurement figure moved from 78,000 units (14142) to a consistently repeated 58,000 (14123, 14049, 14025, 13897, 13978). The aggregate target of 800 MW by year-end 2025 (14108) does not recur in any later claim shown. By 2023 the language shifts to unquantified 'initial buildout' at two sites (13332, 13158). That reads as a revision without acknowledgement rather than a stated reset.

Dates were soft from the start. Pennsylvania ASIC installation by June 30, 2022 (14122, 14024) and mid-2022 Nautilus operations (14107) sit against a later expansion horizon of May 2024 (13716). The Lake Mariner miner targets of 3,300 units by July 1, 2022 (13880) and 3,000 S19 XPs by August 1, 2022 (13877) were issued weeks apart with overlapping goals. Several claims tagged delayed (13408, 13409, 13412, 13414) carry promise dates that predate their filings, so the tag likely reflects backward-looking disclosure rather than confirmed slippage.

Independent corroboration is thin. The strongest anchors are counterparty contracts: the 90 MW NYPA agreement (14110, 14126) and the eleven-year Susquehanna supply arrangement (14113). The record contains no permits, no NYISO or PJM queue entries, and no site reporting. That gap matters far more at 750 MW than it did at 90.

Over the next 90 days, watch for three things. First, whether the $3.5 billion Anthropic facility financing closes, and on what capacity and energization schedule. Second, whether 410 MW and 750 MW are distinguished as phased, dated tranches in the Q3 10-Q. Third, whether NYISO interconnection filings or NYPA allocation changes support load beyond Lake Mariner's historical contract. Separately, note whether any Kentucky, Maryland or Texas site moves from announced to a permitted or queued status.

The claim record supplied for this read ends in March 2023, so the Anthropic-era trajectory can only be framed against TeraWulf's earlier bitcoin-mining disclosures. That limits what can be verified, and the analysis says so where it matters. On capacity, Lake Mariner was first presented as a 90 MW NYPA allocation with 410 MW of expansion potential (12912, 12913), built out in 50 MW buildings (13881, 12914, 12915). The same 410 MW figure now serves as the initial disclosure for the Anthropic AI factory, and the campus target has since grown to 750 MW. No claim in this record documents power rights beyond 410 MW, so the larger number currently rests on company framing. Nautilus moved the other way. TeraWulf's stake fell from 33.33 percent (13714) to 25 percent (12872, 12938), and its power allocation settled at 50 MW (12917) inside a 200 MW facility (13715, 12940). The optional 100 MW expansion to 300 MW is flagged delayed (12873, 12941). On dates, the early record shows a pattern of slippage that went largely unacknowledged. The 58,000-unit delivery commitment for 2022 was repeated four times (14049, 14025, 13978, 13897) and then left the narrative. By March 2023 the company reported 2.8 EH/s (12881), which appears well short of what that fleet would imply. The Pennsylvania installation start (14024), the 3,300 and 3,000 miner milestones (13880, 13877) and the mining commencement date (13409) all passed without restated targets, and 13409 and 12942 carry delayed flags. The one steady commitment is the 2028 zero-carbon target (14026, 13439, 13343), which is too distant to test. On independent evidence, the record contains no permits, NYISO interconnection queue entries or site reporting for Lake Mariner's AI buildout or for any of the eleven announced sites in Kentucky, Maryland, Texas and New York. None of those sites carries a stated capacity or cost. The $3.5 billion financing figure is sourced to a headline, not a filing. Over the next 90 days, the key question is whether that financing closes on disclosed terms. The Q3 10-Q should show energized critical IT megawatts against the Anthropic lease schedule. NYPA or NYISO filings would be needed to substantiate power beyond 410 MW. Any announced site should also receive a first megawatt figure backed by a utility or permitting record. Ingesting the 2024–2026 claims should be the priority before this read is refreshed.

The claims supplied here end abruptly on 2023-03-30, partway through claim 12883, so this read rests on TeraWulf's bitcoin-mining era. None of the disclosures behind the Lake Mariner Anthropic figures (410 MW, 750 MW, the $3.5 billion financing target) appear in the claims shown, and conclusions about the current AI-campus trajectory should wait until that later record is attached. Capacity at Lake Mariner has been elastic from the start. The anchor has always been the 90 MW NYPA power supply agreement (12842, 12912). Around it, expansion headroom was framed at 500 MW in one disclosure (12843) and 410 MW in another on the same date (12913). The later 750 MW campus target therefore extends a ceiling that was never fixed, rather than one that was contracted. Nautilus shows the reverse pattern. It was described as 200 MW expandable to 300 MW (13715, 13716, 12873), and one filing even cited 200 MW of expansion potential (12866). TeraWulf's own allocation, however, was only 50 MW (12917), and its equity stake fell from 33.33 percent (13714) to 25 percent (12872, 12938) without much explanation. Timelines slipped repeatedly and were marked delayed rather than formally reset. Initial mining was promised by March 2022 (13409), Nautilus commencement by February 2023 (12927, 12868), and Lake Mariner Building 2 and 110 MW of mining capacity by Q2 2023 (12918, 12858, 12834). The one date that has never moved is the 100 percent zero-carbon target for 2028 (13846, 13826, 13439, 13411, 13343), with 90 percent claimed as current (13847). That consistency will be tested as the project list grows to include Kentucky sites and the Morgantown Generating Station, a former fossil plant. No permits, interconnection queue entries or site reporting are in the record, so the announced pace has no independent corroboration here. The Nautilus project entry lists Tompkins County, New York, while the claims describe the Talen joint venture in Pennsylvania, which looks like a mapping error worth correcting. Over the next 90 days, the key test is whether the $3.5 billion Anthropic facility financing closes and what megawatt figure the closing documents commit to. Watch also for any NYPA or New York grid filings that move Lake Mariner's contracted power beyond 90 MW, and for interconnection or permit activity at Abernathy, Morgantown and the Kentucky sites. Without those filings, those sites remain announcements with no stated capacity.

The record supplied here stops at the March 2023 10-K cycle. The Lake Mariner figures most relevant today, the 750 MW campus target and the $3.5 billion Anthropic-linked financing, appear only in the project header and have no dated claim behind them. That limits any reading of the recent trajectory. What the claims do show is how the capacity story was framed early on. Lake Mariner began as a 90 MW NYPA allocation (12842, 12912), with 60 MW operating (12882) and two 50 MW mining buildings (12914, 12915). In the same filing, TeraWulf described expansion potential as both 410 MW (12913) and 500 MW (12887). The ceiling was elastic from the start. The current 750 MW figure extends that pattern rather than breaking from it. Until we see the source claim, we cannot tell whether it rests on a new power agreement or a restated aspiration.

Nautilus shows the opposite motion. Facility capacity held at 200 MW initial, with a 100 MW option by May 13, 2024 (13715, 13716, 12873, 12941). TeraWulf's own position shrank, though. Its equity went from 33.33 percent (13714) to 25 percent (12938, 12921), and its allocation from 50 MW to 50 MW (12917, 12883). The 2024 expansion date and the June 2023 transition-services date (12942) are both flagged delayed, and no later claim acknowledges the slip.

Several entries marked delayed need care. The March 2022 mining start (13409), the NYDIG agreements (13412, 13414) and the IKONICS combination (13408) carry target dates that had already passed when TeraWulf filed in February 2023. Some of these may be completed events recorded with historical dates rather than genuine slippage, and they should be checked before being counted against management. The 100 percent zero-carbon by 2028 target has been restated consistently (13439, 13411, 13343).

No independent evidence is in the record, meaning no permits, NYISO interconnection queue positions or site reporting. That is the central gap. A jump from a 90 MW contracted base to 750 MW at a single Upstate New York site would normally leave a trail in NYPA allocation filings and the NYISO queue.

Over the next 90 days, three things matter. First, ingest the 2024 to 2026 claims that underpin the 750 MW and $3.5 billion figures. Second, check the NYISO queue and NYPA board minutes for Lake Mariner load increases. Third, look for Q3 2026 disclosure on financing close and energized megawatts, as distinct from contracted megawatts, at the Anthropic-leased buildings.

The record supplied here is truncated at March 2023, so what follows reads the mining-era baseline against the current project listing rather than the full AI-era claim history. That limits any judgment on the 2026 Anthropic figures, which appear in the project record but in none of the visible claims.

On capacity, the most useful thread is continuity. The 410 MW now labeled the initial disclosure for the Lake Mariner Anthropic AI Factory first appears in claim 12913 as NYPA expansion potential for a bitcoin site. At that time the site had a 90 MW initial power agreement (12912) and 60 MW operating (12882). The AI pivot inherited that ceiling rather than creating it. The later 750 MW campus target goes beyond anything the 2023 power arrangements describe, so it implies new supply that should be independently documented before it is treated as bankable. Nautilus followed the opposite path. A 200 MW facility expandable to 300 MW (12873, 12940, 12916) was only ever a 50 MW, 25 percent position for TeraWulf (12883, 12938), and the 100 MW expansion option (12941, 12922) is marked delayed with no sign of exercise.

On timing, the slippage is extensive but mostly acknowledged. Mining commencement promised for March 2022 (13409) and Nautilus operations for February 2023 (12868) were logged as past-due in the filings themselves. Lake Mariner Building 2 and the 110 MW Q2 2023 target (12858, 12918, 12834) and the 44,450-miner delivery commitment (12900) also slipped. Several of these were repeated across February and March 2023 filings without firm new dates, which is closer to silent revision than open resetting. The 100 percent zero-carbon target for 2028 (13439, 13411, 13343) has been repeated unchanged and remains untested.

No independent evidence (permits, NYISO queue positions, or site reporting) is attached to any claim here. The announced pace of the AI build is therefore unverified in this dataset. The same is true of the eleven other announced sites, which carry no capacity or cost figures at all.

Over the next 90 days, the key test is whether the $3.5 billion Anthropic-linked financing actually closes, with terms disclosed. Watch the Q3 2026 results for energized versus contracted megawatts at Lake Mariner. Check NYISO and NYPA records for power above 410 MW that would support the 750 MW target. Also look for whether any Kentucky, Texas or Maryland site gets a megawatt figure backed by a utility or queue filing.

The claim history supplied here ends in August 2023 and appears truncated, so this read covers TeraWulf's bitcoin-mining build-out far better than the AI-hosting pivot now anchoring the Lake Mariner record. That gap is itself the central finding. Nothing in the visible claims ties a date, phase schedule or energization milestone to the Anthropic-leased facility, the 750 MW campus target, or any of the eleven announced sites from Abernathy to Morgantown, none of which carry a disclosed capacity or cost.

On capacity, Lake Mariner's ambition has grown steadily. It began as a 90 MW NYPA supply agreement (12912, 12890), described at the same time as expandable to 410 MW (12913) and elsewhere to 500 MW (12843, 12887). The project record now cites 750 MW. That escalation came without a matching disclosure of new power allocation. Nautilus was framed consistently as 200 MW initial and 300 MW potential (12940, 12873, 12916), with TeraWulf's share fixed at 50 MW (12917, 12883). The expansion option is described as 100 MW in some filings (12941, 12922) and 200 MW in another (12866), a discrepancy the company never reconciled.

On timing, the Q2 2023 targets of 110 MW at Lake Mariner and 160 MW in aggregate (12834, 12833), along with Building 2 completion (12918, 12858), were paired with 60 MW actually operating at the time (12882). The August 2023 statements that miners were received and capital commitments funded (12266, 12267, 12268) are consistent with those targets broadly holding. However, they stop short of confirming energized megawatts. The Nautilus commencement date of February 1, 2023 (12868, 12927) was stated retrospectively, so it is not a test of forecasting.

The supplied record contains no independent corroboration: no NYISO queue positions, NYPA board allocations, local site-plan approvals, or on-the-ground reporting for any site. The announced pace, particularly the move from 410 MW to 750 MW and the $3.5 billion financing, therefore rests entirely on company and headline statements.

Over the next 90 days, the Q3 2026 10-Q due in November should show whether the Anthropic facility carries dated delivery phases and whether the $3.5 billion financing has closed. Beyond that filing, the key tests are whether NYPA or NYISO filings show power allocation above the historical 90 MW base for the 750 MW figure, and whether any Kentucky, Texas or Maryland site produces a permit, interconnection request or disclosed megawatt figure that would move it beyond announced status.

The ledger for TeraWulf splits into two eras, and only the first is documented. Every dated claim here runs from July 2023 to March 2024 and concerns bitcoin mining. The AI-era figures on the project record, 410 MW initially and 750 MW later at the Anthropic-leased Lake Mariner facility, with a $3.5 billion financing target reported July 13, 2026, have no claim ids behind them. Any read of the current trajectory therefore rests on the older mining disclosures plus an unsourced headline.

In the mining era, capacity language was stable at the top and less stable underneath. Lake Mariner's 500 MW scalable ceiling was repeated without change (12115, 12097, 12080, 12033, 12054), as was 110 MW of operating capacity (12124, 12095, 12078). The near-term expansion, however, shrank quietly. The third building was described as 43 MW through March 7, 2024 (12113, 12096, 12079), and twelve days later it became a 35 MW expansion due by mid-2024 (12025, 12070, 12053). No filing on record explains the 8 MW reduction. Nautilus is murkier. The expansion option was framed as 50 MW (12121) and as 100 MW (12100) with the same May 13, 2024 deadline. It was then described as exercised at 100 MW (12087, 12059) while a separate 50 MW step was pushed to 2025 (12026, 12060). Reported hashrate sat at 5.5 EH/s across three filings (12258, 12165, 12127), even as miner deliveries were being booked (12365, 12111). Some statements also cut against each other. Construction was called substantially complete in August 2023 (12269), yet buildout was said to be continuing in the same filing (12265).

No independent evidence is on record. The ledger contains no permits, NYISO queue entries, or site reporting that could corroborate either the mining pace or the jump from a 500 MW ceiling to a 750 MW campus target. The only disclosed power contract is the 90 MW NYPA agreement (12114, 12098, 12056). That is well short of either AI-era figure, and nothing on file shows how the gap will be closed. The eleven other announced sites, spanning Kentucky, Maryland, Texas and New York, carry no capacity or cost at all.

Over the next 90 days, three checks would narrow these gaps. The first is whether the 2025 and 2026 filings ingested into this ledger confirm that the 35 MW mid-2024 target was met. The second is whether a 410 MW or 750 MW figure appears as a dated company claim rather than a headline. The third is whether the $3.5 billion financing closes on stated terms. Beyond those, look for new NYPA or NYISO interconnection filings sized to the AI load, and for any capacity number attached to the announced Kentucky and Maryland sites.

The capacity record for TeraWulf's Lake Mariner campus has grown in several steps, and the steps are not always reconciled with each other. Through late 2023 the company described 110 MW operating at Lake Mariner (12124) and 50 MW at Nautilus (12126). It paired those figures with a 500 MW scalability ceiling for Lake Mariner, repeated almost verbatim across filings (12115, 12097, 12080, 12033, 12054). Within the same March 2024 cycle, operating capacity at Lake Mariner was stated as 110 MW (12078) and then as 160 MW (12037) twelve days later, with no bridging disclosure. The project list now carries an AI campus sized at 410 MW initially and 750 MW subsequently. That target exceeds the 500 MW ceiling the company held for years, and none of the claims supplied here explain how the site's power basis changed to support it. The last supplied claim (12053) is cut off mid-entry, so claims covering the pivot to hosting AI workloads are absent from this record.

Dates and sizes have drifted quietly. The third Lake Mariner building was described as 43 MW (12113, 12096, 12079). A later claim cites a 35 MW expansion due by mid-2024 (12025), and nothing in the record says whether that is the same building, resized. The Nautilus expansion option was first described as 50 MW by May 2024 (12121), then as 100 MW against the same date (12100), then as exercised (12087, 12059). A separate 50 MW Nautilus expansion is still pencilled for 2025 (12060, 12026), and the record does not say how it relates to the exercised option. The 18,500-miner deployment promised by February 2024 (12111) has no follow-up claim confirming it.

The record contains no permits, interconnection queue entries, or on-site reporting. None of these figures can be checked against outside evidence here, and the dozen other announced sites in New York, Kentucky and Maryland carry no stated capacity or cost at all. The NYPA supply agreement was fixed at 90 MW (12114, 12027). That is far below the 410 MW and 750 MW figures, so any growth beyond it depends on power arrangements that are not disclosed in this record.

Over the next 90 days, watch four things. First, whether the $3.5 billion financing tied to the Anthropic lease closes on its stated terms. Second, whether the third-quarter 10-Q reconciles the 410 MW and 750 MW figures and sets a date for energizing the first phase. Third, whether NYISO queue filings or amendments to the NYPA allocation appear that could support load above 500 MW. Fourth, whether local site-plan or building approvals in the Town of Somerset keep pace with the construction status now claimed.

The record supplied here stops mid-entry at claim 5586 in October 2024, so the nearly two years of claims behind the current Lake Mariner figures are missing. Read what follows with that gap in mind. The shift from a bitcoin-mining footprint to a 410 MW initial disclosure and a 750 MW campus target cannot be traced claim by claim from this set.

On capacity, the visible pattern is a stable ceiling and then a large step change. From late 2023 through March 2024, TeraWulf repeatedly put Lake Mariner's maximum scalable capacity at 500 MW (12115, 12097, 12080, 12033). The current project record goes past that ceiling, but no claim in this set explains the change. Operating capacity moved from 110 MW on March 7, 2024 (12078) to 160 MW on March 19 (12037, 12015). That jump is roughly consistent with the 43 MW third building (12079), but the company never stated that the building had been completed.

On timing, the Nautilus sequence is the clearest example of a quiet revision. The expansion option was first tied to a May 2024 date (12121, 12100), then an earlier February 2024 exercise date (12020), and then described as exercised (12087, 12059). The actual 50 MW build was then pushed to 2025 (12026, 12060). Seven months later the company sold its Nautilus stake (5646). The record does not reconcile that sale with the 2025 expansion commitment. Two other promises also have no visible follow-up confirming delivery: the 35 MW Lake Mariner expansion due mid-2024 (12070, 12025) and the 18,500-miner deployment due February 2024 (12111).

No independent evidence appears in the record: no permits, no NYISO interconnection queue positions, and no local reporting. The one grounded power figure is the 90 MW NYPA agreement (12044). A 750 MW campus would need grid access far beyond that, so a load interconnection filing is the most important external check. The October 2024 Somerset lease claims (5581, 5582) are tagged to the Anthropic project but predate it. Claim 5587's share figure looks like a parsing error.

Over the next 90 days, watch for four things. First, whether the $3.5 billion financing reported in July closes on stated terms. Second, whether third-quarter results give dated energization milestones for the 410 MW tranche, or restate the 750 MW target without a schedule. Third, whether NYISO or Town of Somerset filings match the announced pace. Fourth, whether any of the eleven capacity-less announced sites, from Abernathy to the Kentucky campuses, gets a megawatt figure or a date.

TeraWulf's Lake Mariner facility, the only project under active construction, illustrates a pattern of capability expansion paired with execution delays. Initial disclosures cited 410 MW capacity; by May 2025, claims 4997 and 4505 settled on 750 MW maximum capacity with 500 MW near-term expansion. That positioning has held through February 2026 (claim 11760), suggesting the company has stabilized its vision at a 750 MW campus served by Somerset-Brookings infrastructure (claim 4505). However, the builings-level timeline reveals material slippage. CB-2B was promised delivery by March 1-31, 2026 for 21 MW (claims 11754, 11730); as of September 13, 2026, that deadline is five to six months overdue. CB-3 energization was promised by May 15, 2026 for 42 MW (claim 11755), now four months past. CB-4 energization carried a July 1, 2026 target (claim 11756) for 168 MW, now 74 days overdue. CB-5 is promised by October 1, 2026 (claim 11757) for 168 MW—imminent risk territory. Notably, the same claims offer conflicting dates: CB-3 pairs the May 15 energization with a July 31 lease-start target (claim 11731), and CB-4 pairs July 1 energization with December 31 lease-start (claim 11732), suggesting either dual-tracking or unannounced scope shifts. Beyond Lake Mariner, TeraWulf has announced eleven additional sites with minimal transparency: Abernathy, Cayuga, and Justified Data carry capacity targets (168 MW, 350 MW, 384 MW respectively) but lack concrete timelines, permits, or interconnection queue evidence. The February 25, 2026 portfolio slide claimed 2,400 MW total controlled capacity and 2,900 MW multi-regional platform by implication, yet only Lake Mariner shows construction progress. The July 13, 2026 $3.5 billion financing headline underscores ongoing capital intensity. Critical 90-day monitors: CB-5 energization outcome (due Oct 1); any revised guidance for CB-2, CB-3, CB-4 without explicit acknowledgement of prior delays; and whether interconnection approvals or power contracts materialize for Cayuga or Justified Data sites. Independent validation—permits, queue position, utility filings—remains absent from TeraWulf's public record.

TeraWulf has pursued an aggressive expansion strategy centered on Lake Mariner, nearly doubling the facility's initial 410 MW announcement to a 750 MW target across three tranches (claims 4995–4997), while simultaneously announcing a portfolio of 12 additional regional projects purporting to reach 2,900 MW of platform capacity (claim 11762). This portfolio-wide approach masks significant execution risk concentrated at a single site.

The Lake Mariner delivery timeline warrants scrutiny. The company promised sequential building energizations through 2026: CB-2B by March 1, CB3 by May 15, CB4 by July 1, and CB5 by October 1 (claims 11754–11757). We are now in mid-September 2026. The absence of announced completion confirmations for the first three milestones—which have already passed—is conspicuous. No independent evidence via grid interconnection filings, utility reports, or construction permitting records appears to corroborate whether these 21 MW, 42 MW, and 168 MW tranches came online as promised. The CB5 delivery promise of October 1 arrives in 18 days, but the track record of prior claims creates uncertainty about execution consistency.

Outside Lake Mariner, the announced pipeline remains largely skeletal. Justified Data in Kentucky carries a 2027-12-31 target for 384 MW (claim 11734), while Abernathy's Fluidstack joint venture missed an implied 2025-10-01 deadline without public acknowledgement (claim 11753). The Kentucky brownfield acquisition, promised by March 1, 2026, already shows a [delayed] flag (claim 11758), suggesting slippage is accumulating. Chesapeake Data and associated generation assets remain undated.

Independent validation is absent from the record. No FERC interconnection queue positions, state utility commission filings, or engineering permits are cited. All 79 claims in the data set are marked [stated], meaning internal announcement only. For a capital-intensive infrastructure company, this gap between announcement cadence and external corroboration is material.

Over the next 90 days, track the October 1 CB5 energization claim closely—confirmation or delay will signal whether Lake Mariner's 2026 timeline is holding. Monitor utility interconnection filings for Justified Data and Abernathy sites. Any further acquisition date slippages without revision would indicate systemic execution headwinds beneath the portfolio growth narrative.

TeraWulf's capacity roadmap has undergone dramatic expansion, particularly in scope rather than density. The Lake Mariner facility evolved from an initial 410 MW disclosure in October 2024 (claim 5501) to 500 MW near-term expansion (claim 4996) and then 750 MW maximum capacity with transmission upgrades (claim 4997), all announced by May 2025. That represents a significant upward revision in a short window. By February 2026, the company had broadened its ambitions substantially: a 2,900 MW multi-regional platform (claim 11762) comprising a 2,400 MW controlled HPC platform (claim 11724), with fresh projects announced across New York, Texas, Maryland, and Kentucky. However, capacity commitments for most sites remain unspecified, with Cayuga the exception at 320 MW potential (claim 11761) down from the 350 MW proposed just months earlier (claim 1633).

Timeline adherence shows warning signs. Several near-term construction milestones from the February 2026 filing have now passed or are imminently due. CB-2B construction delivery was promised for March 31, 2026 (claim 11730); CB-3 lease start for July 31, 2026 (claim 11731); CB-3 energization for May 15, 2026 (claim 11755); and CB-4 energization for July 1, 2026 (claim 11756). These now sit as historical claims without clear public acknowledgment of completion or delay. The Abernathy joint venture with Fluidstack, promised for October 1, 2025 (claim 11753), did not materialize until December 2025, a two-month slip not explicitly called out in subsequent statements. Lake Mariner CB-5 energization remains promised for October 1, 2026 (claim 11757), just over two weeks away.

Independent verification remains sparse. The record shows significant announced sites but minimal evidence of interconnection queue entries, permits filed, or grid operator confirmation of timelines. The February 2026 filing heavily emphasizes potential and partnerships—a Google backstop guarantee (claims 4509, 4458), Fluidstack arrangements (claims 1052-1054), lease options—rather than locked capacity. Justified Data received a 2027-2028 delivery window (claim 11734), and Abernathy construction is characterized as ongoing without energization dates, suggesting material slippage from the original October 2025 joint venture deadline.

Watch CB-5 completion by October 1, 2026, and any updates on Justified Data and Abernathy's actual construction pace relative to announced commitments.

TeraWulf's Lake Mariner Anthropic AI Factory exhibits capacity repackaging rather than sustained linear growth. Initial disclosures set 410 MW as the baseline, reframed by May 2025 into 500 MW near-term and 750 MW contingent on transmission upgrades (claims 4996, 4997). By August 2025, announcements shifted to 200 MW of dedicated HPC via Fluidstack and 160 MW via a CB-5 expansion (claims 4507, 1053, 4455), displacing earlier framings. Most significantly, February 2026 saw withdrawal of a 606 MW contracted-capacity figure (claim 3623, withdrawn), suggesting announced totals may not correspond to secured bookings. Whether this represents client attrition, reclassification, or initially overstated claims remains unexplained in subsequent disclosures.

Timeline discipline has weakened measurably. A December 2024 commitment to file an S-3 registration statement remains unfulfilled as of September 2026 (claim 5586, delayed). The Hawesville brownfield site acquisition promised by February 2, 2026 has also slipped (claim 3877, delayed). Against this backdrop, CB-5 carries the only remaining hard deadline: December 31, 2026 completion, now less than four months away (claim 4456).

The company has seeded a portfolio of secondary projects—Cayuga (350–400 MW depending on claim), Abernathy in Texas (168 MW via 50.1% Fluidstack joint venture), Morgantown (210 MW), and unspecified capacity at Hawesville—yet none carries credible delivery dates. This broad pipeline may signal genuine optionality or reflect a need to offset cooling near-term progress at Lake Mariner.

Independent verification is sparse. The New York Power Authority supply agreement for 90 MW provides some commercial grounding (claim 3627), and Fluidstack partnerships offer counterparty credibility (claims 1054, 3654). However, interconnection queue filings, local building permits, or third-party construction milestones do not appear in this record. By mid-2026, after years of facility announcements, the absence of such evidence suggests permitting and grid coordination remain immature.

Monitor CB-5's December 2026 deadline closely over the next quarter. Track Cayuga's permitting status and any quantified progress metrics from the Abernathy-Fluidstack joint venture. Any re-announcement of the withdrawn 606 MW figure, or clarification of actual signed customer commitments at Lake Mariner, would signal material course changes.

TeraWulf has materially shifted Lake Mariner's capacity guidance without formal reversal. The project launched at 410 megawatts, expanded to 500 megawatts near-term and 750 megawatts at build-out by May 2025 (claims 4996, 4997). By February 2026, the company withdrew its claim for 606 megawatts of contracted critical IT load across Lake Mariner and Abernathy combined (claim 3623, marked withdrawn), then re-anchored to the same 500/750 megawatt framework without acknowledging the interim commitment. This pattern—escalating then retreating without comment—recurs in February's re-disclosure of Somerset acreage, which widened from 157 to 162.7 acres (claims 5582, 4831), a 3.6% adjustment likely reflecting site-level negotiations rather than expansion.

Power supply presents the clearest near-term constraint. The February 2026 New York Power Authority agreement delivers only 90 megawatts (claim 3627), less than one-fifth of the announced 500 megawatt near-term target and one-twelfth of the 750 megawatt peak. No second utility agreement has been disclosed, leaving a 410 megawatt gap between stated capacity and secured supply. Whether that gap closes through transmission upgrades or customer deferral remains unresolved.

Timeline claims have grown vaguer as the project portfolio has expanded. Lake Mariner's development is flagged with generic "construction and delivery timeline" language in August and October 2025 (claims 4506, 4188) without stated milestones. Concurrently, TeraWulf announced 12 additional facilities—Abernathy in Texas, Cayuga in New York, Justified Data in Kentucky, and others—most without disclosed capacity or cost. This diffusion of project focus, the withdrawal of a near-term load commitment, and the absence of power infrastructure visibility suggest capital is being distributed across regions rather than concentrated on Lake Mariner's completion.

Monitor CB-5's December 2026 deadline (claim 4456, 160 megawatts), any update to interconnection queue positions that would indicate transmission upgrade timing, and whether TeraWulf secures additional utility supply agreements. A second power contract outside NYPA's 90 megawatts would be material validation of the 500 megawatt near-term thesis.

TeraWulf's Lake Mariner facility claims have expanded materially without clear delivery milestones. Initial disclosure of 410 MW has evolved into a 750 MW campus target, but intermediate steps reveal fragmentation. May 2025 filings separated near-term 500 MW expansion (claim 4996) from full 750 MW potential contingent on transmission upgrades (claim 4997), with a distinct 200 MW HPC allocation (claim 4507). By February 2026, these remained consistent (claims 3649, 3650), but a simultaneous withdrawal is telling: claim 3623 retracted a 606 MW contracted critical IT load statement, indicating announced capacity exceeds actually committed power. This gap undermines capacity-expansion credibility.

Timeline slippage is evident. The CB-5 expansion promised completion by December 2026 (claim 4456); as of September 2026, no delivery notification has emerged. Earlier, a December 2024 S-3 registration commitment (claim 5586) was marked delayed, and the Hawesville Kentucky acquisition (claim 3877) missed its February 2026 close date. The portfolio has proliferated without execution visibility: Abernathy's 168 MW shares Fluidstack joint venture timelines (claim 3656) without specific dates, and Cayuga claims 320–400 MW contingent on permitting (claim 3658) with no regulatory progress disclosed. Nine additional facilities carry zero capacity or timeline specifics, functioning as land optionality rather than development pipeline.

Partnership arrangements lack transparency. The Fluidstack 50.1% co-venture at Abernathy (claim 3654), Google power backstops (claim 4509), and New York Power Authority commitments (90 MW, claim 3627) are announced without binding agreement publication or independent verification. The Beowulf transition services extend through May 2027 (claim 4834), suggesting incomplete separation from the acquired platform operator.

Over ninety days, watch whether CB-5 achieves its December 2026 deadline—slippage would mark the second generation of missed timelines. Monitor whether Lake Mariner's stated 500 MW near-term expansion actually commences drawing power and whether NYPA's 90 MW commitment is operationally active. Track Kentucky site acquisitions (Hawesville, Muskie, Hancock County) for proof of site control and development permits. Any further capacity claims without contracted power commitments should be discounted.

TeraWulf's Lake Mariner complex reveals a shifting expansion narrative across its Anthropic AI Factory project. The announced capacity has expanded from an initial 410 megawatts to a stated potential of 750 megawatts, though construction phasing remains unclear across successive disclosures. Claims 4996 and 4997 in May 2025 introduced the 500-megawatt near-term and 750-megawatt potential capacity, but subsequent restatements in February 2026 (claims 3649, 3650) repeat these targets without clarifying sequencing. The CB-5 expansion meanwhile remains bounded at 160 megawatts with a construction completion deadline of December 31, 2026 (claim 4456)—less than four months away.

Several timeline commitments have slipped without public revision. The Form S-3 registration filing missed its promised December 2024 deadline (claim 5586). More significantly, the Morgantown Generating Station acquisition, promised for completion by December 1, 2025 (claim 2116), remains unresolved as of September 2026, nearly ten months overdue. The Hawesville brownfield acquisition likewise passed its promised February 2 date (claim 3877).

A major withdrawal signals internal reassessment. Claim 3623 asserted 606 megawatts of contracted critical IT load across Lake Mariner and Abernathy combined but was withdrawn in February 2026 without public explanation. This suggests either renegotiated customer commitments or delayed deployments that the company chose not to restate openly.

The New York Power Authority supply agreement for 90 megawatts (claims 3627, 2114) provides partial independent verification of at least one component, though delivery timing for that supply remains unstated. The July 2026 announcement of a $3.5 billion financing target for the Anthropic facility indicates continued commitment but also implies higher capex requirements than the original narrative suggested.

Over the next 90 days, monitor CB-5 construction status against its year-end deadline, any public disclosure on the Morgantown acquisition, and whether Lake Mariner's 750-megawatt potential is reaffirmed or quietly revised. The withdrawal of claim 3623 sets a baseline for scrutinizing any new contracted capacity announcements.

TeraWulf's Lake Mariner expansion narrative reveals escalating targets paired with unexplained retreat. Initial 410 MW capacity evolved to 500 MW near-term and 750 MW maximum-with-upgrades by mid-2025 (claims 4996–4997), language consistent through February 2026 (claims 3649–3650). Yet the company withdrew a headline claim citing 606 MW of contracted critical load across Lake Mariner and Abernathy in that same February filing (claim 3623), a material pullback suggesting either unsecured customer commitments or demand forecasts that failed to hold. Actual power secured remains thin: the New York Power Authority backstop provides only 90 MW (claim 3627), orders of magnitude below the 750 MW ceiling, implying transmission constraints or financing gaps real enough to constrain expansion velocity.

Timeline slippage clusters on acquisitions rather than organic builds. Morgantown Generating Station missed its December 2025 close by nine months and counting (claim 2116), now past deadline by nearly a year as of mid-September 2026. Hawesville slipped one day in February (claim 3877). The CB-5 Lake Mariner expansion carries a hard 2026 year-end completion target (claim 4456), three months forward, making it the nearest test of execution credibility. No other major projects carry specific delivery dates in the record.

Independent verification is sparse. No interconnection queue filings, construction permits, or third-party on-site reporting appear in the data set to anchor the 750 MW roadmap. The withdrawn contracted load claim is telling—internal modeling likely diverges from customer reality. The 90 MW NYPA lock creates a hard ceiling unless new transmission infrastructure is already funded and queued; no evidence of either emerges here.

In the next 90 days, track CB-5 completion credibility before year-end. Monitor silence or announcements on Morgantown—continued delay signals deeper friction. Watch for fresh contracted customer capacity claims for Lake Mariner; absence of new bookings would confirm power availability as binding constraint. Any capacity withdrawals for Cayuga, Abernathy, or other announced sites warrant close scrutiny as signal of demand softening.

TeraWulf has pursued an expansive growth strategy centered on Lake Mariner, which has seen its capacity target scale from an initial 410 megawatts through multiple tranches: 500 megawatts near-term expansion (claims 4996, 3624) and 750 megawatts as a theoretical maximum with transmission upgrades (claims 4997, 3650). The scaling trajectory suggests either evolving customer demand or a shifting baseline, but the firm has not articulated why the upper bound requires conditional infrastructure upgrades unavailable at announcement. Claims 1051, 1054, and 1053 describe specific partnerships—G42 at 70 megawatts, Fluidstack at 200 megawatts, and CB-5 at 160 megawatts—that ostensibly stack toward those targets, yet claim 3623, which asserted 606 megawatts of contracted load across Lake Mariner and Abernathy, was withdrawn in February 2026, suggesting either contractual strain or reconciliation of overstatement.

Timeline slippage is pronounced. Acquisition timelines have drifted significantly: Morgantown Generating Station (claim 2116) missed its December 2025 delivery by nine months as of this writing, and the Hawesville brownfield acquisition (claim 3877) fell seven months past its February 2026 target. A Form S-3 filing commitment (claim 5586) dated December 2024 remains outstanding twenty-one months later. CB-5 construction completion is promised by year-end 2026 (claim 4456), now within ninety days; this is the nearest tangible milestone warranting tracking.

Independent evidence of pace is sparse. No permits, grid interconnection queue filings, or on-site workforce reports have surfaced to corroborate the rate of expansion. The Cayuga site claims have migrated from 350 megawatts announced capacity (claim 1633) to 400 megawatts potential (claims 3644, 3658), but capacity is conditional on permitting completion, a phase not yet publicly documented. The New York Power Authority supply agreement (claim 3627) of 90 megawatts is fixed, but the bulk of Lake Mariner's growth hinges on speculative transmission work and customer contracting.

Watch whether CB-5 closes on schedule by year-end and whether Morgantown and Hawesville acquisitions complete within the next quarter. Monitor whether any withdrawn claims are re-disclosed with different metrics, and track whether new partnerships are announced in advance of prior ones reaching operational status. The ratio of announced capacity to commissioned megawatts will indicate whether growth claims are maturing into realized infrastructure or remaining aspirational.

TeraWulf's capacity announcements reveal substantial gaps between targets and current delivery. Lake Mariner Anthropic AI Factory began with 410 MW initially disclosed, then scaled to 750 MW maximum capacity with transmission upgrades (claim 4997, May 2025) and 500 MW near-term expansion (claim 4996, reaffirmed claims 3649, 3624, Feb 2026). However, only 60 MW remained operational as of April 2026 (claim 2106), supplemented by a 90 MW NYPA power supply agreement (claim 3627). The CB-5 lease expansion of 160 MW (claim 1053) carries a construction deadline of December 31, 2026 (claim 4456). This represents roughly 10–30 percent of headline targets depending on the metric—far below the near-term 500 MW goal.

Cayuga Site capacity claims shifted significantly: 350 MW initially stated (claim 1633, Sept 2025), revised to 400 MW development potential (claim 3644, Feb 2026), then to 320 MW future capacity (claim 3625, Feb 2026). These rescopings occurred within six months with no public acknowledgment. The 80-year lease covers 183 acres (claims 4501, 4500).

Abernathy HPC Campus in Texas stabilized at 168 MW through a 50.1 percent Fluidstack joint venture (claims 3654, 3641, Feb 2026), but construction timelines lack specific completion dates. A note redemption was conditionally tied to campus completion (claim 3954, Dec 2025), signaling deadline pressure without public commitment.

Justified Data at Hawesville shows a slippage: the brownfield acquisition promised by February 2, 2026 (claim 3877) was marked delayed. The site carries 480 MW stated power availability (claim 2109, Apr 2026), though no operational milestone has been achieved.

Fluidstack partnerships aggregate 200–368 MW across Lake Mariner and Abernathy (claims 1052, 1053, 1054), though these overlap with core facility capacity rather than additive growth.

No independent corroboration—permits, interconnection queue positions, construction photographs—appears in available claims. The December 2026 CB-5 deadline at Lake Mariner and any Abernathy campus update merit closest monitoring in the next 90 days. Watch for revised timelines without explicit announcement, a pattern evident in Cayuga's downward capacity adjustments.

TeraWulf's capacity announcements have scaled dramatically without corresponding operational progress, creating a widening gap between stated targets and delivered infrastructure. Lake Mariner, the company's flagship Anthropic-leased facility, illustrates this pattern: initially disclosed at 410 MW, then escalated to 750 MW potential (4997, 3650), then re-emphasized at 500 MW near-term expansion (4996, 3624). Actual operational capacity as of April 2026 reached only 60 MW (2106)—roughly 8 percent of the initial claim and 12 percent of near-term targets. The CB-5 lease expansion, carrying 160 MW of committed capacity (1053), faces a December 2026 completion deadline (4456), now fewer than ninety days away and serving as an immediate test of TeraWulf's execution capability. Timeline commitments have demonstrably slipped. The Form S-3 registration promised by December 2024 remains unfiled (5586), and the Hawesville brownfield acquisition missed its February 2, 2026 close date (3877). More striking is the February 2026 withdrawal of claim 3623, which had asserted 606 MW of contracted critical IT load across Lake Mariner and Abernathy—retracted without public disclosure of rationale. The announced portfolio compounds the execution gap through redundant and conflicting capacity figures. Cayuga Site capacity claims shifted from 350 MW proposed (1633) to 400 MW development potential (3644) to 320 MW future capacity (3625) within eight months, none accompanied by binding timeline. Abernathy HPC Campus, a Fluidstack joint venture pegged at 168 MW (3655), generated construction timeline language throughout late 2025 (3952, 3967) but shows no operational status in claims after February 2026. Six announced sites lack capacity specifications, suggesting announcement precedes technical definition. The absence of claims data after April 2026 is notable. For the next ninety days, evidence of progress hinges on whether CB-5 achieves its December close, whether Abernathy demonstrates operational advance, whether Cayuga permitting yields hardened timelines, and whether TeraWulf issues updated capacity or timeline guidance or continues silence. Interconnection queue position, permit status, and third-party construction documentation provide external verification of pace.

TeraWulf's portfolio exhibits classical patterns of capacity inflation without corresponding timeline clarity. The Lake Mariner facility exemplifies this: initial disclosures (October 2024) cited 410 MW, escalating within months to claims of 500 MW near-term and 750 MW maximum (claims 4996, 4997 in May 2025; restated February 2026 as claims 3649, 3650). Yet actual operational capacity remains 60 MW as of April 2026 (claim 2106), more than 18 months after the initial facility acquisition. The company has not acknowledged this gap between announced and delivered capacity as a sustained delay; instead, it repackages targets across overlapping entities—Lake Mariner, Akela Data, La Lupa—blurring accountability for individual facility progress.

Dates have slipped without explicit revision. The Form S-3 filing promised by December 8, 2024 (claim 5586) arrived late; the Hawesville site acquisition, due February 2, 2026 (claim 3877), was delayed. More significantly, a February 2026 withdrawal of the 606 MW contracted critical IT load claim (claim 3623)—previously stated as secured—suggests either a major customer cancellation or a fundamental restatement, neither addressed in subsequent filings. The CB-5 construction deadline (December 31, 2026, claim 4456) remains eight months away; no mid-project updates are visible in the record.

Capacity claims increasingly depend on conditional partnerships rather than autonomous execution. The Fluidstack joint venture dominates near-term growth (168 MW Abernathy, claim 3655; 200 MW Lake Mariner, claim 1054; 50.1% ownership, claims 3633, 3641), with Google backstop guarantees (claims 4509, 4458) providing financial cushion. The Morgantown Generating Station acquisition (February 2026, 210 MW operational capacity, claim 3881) adds supply but not demand certainty. Across the portfolio, acreage and lease terms are heavily documented (Cayuga's 183 acres, 80-year term; Lake Mariner's 162.7 acres), yet power-delivery timelines remain vague or future-dated.

The next 90 days will reveal whether CB-5 meets its December 2026 deadline and whether Abernathy campus construction has begun. Any update to Fluidstack JV terms, Morgantown commissioning status, or NYPA power delivery at Lake Mariner (90 MW promised, claim 3627) would signal actual facility progression versus continued announcement accumulation.

TeraWulf's Lake Mariner campus tells a story of ambitious projection meeting operational reality. Through May 2025, the company announced a near-term path to 500 megawatts and ultimate 750-megawatt capacity at Lake Mariner (claims 4996, 4997), building on the original 410-megawatt facility disclosed for the Anthropic lease. Yet by April 2026, TeraWulf reported only 60 megawatts of current operational capacity at the campus (claim 2106), while reiterating a 500-megawatt expansion target (claim 2107). The gap between stated near-term plans and actual deployment widened rather than narrowed over nine months. This contraction mirrors a February 2026 withdrawal of a 606-megawatt contracted load claim (claim 3623), suggesting either customer delays, power availability constraints, or revised internal projections that went unannounced. The CB-5 lease expansion of 160 megawatts was promised by December 31, 2026 (claim 4456), now just over three months away with no recent capacity updates. Elsewhere, Abernathy campus construction remains described in vague timeline language (claims 3635, 3654, 3656) with no specific delivery date since late February. Cayuga received 350-megawatt capacity claims (claim 1633, September 2025) and potential 400-megawatt capacity pending permits (claim 3658, February 2026), yet an 80-year lease term (claim 4501) and property option extending to 2075 (claim 4502) signal long-lead development. Morgantown generating station acquisition added 210 megawatts of generation capacity (claim 3881, February 2026), providing power rather than IT load. No independent evidence—utility interconnection queue positions, construction permits filed, or third-party inspections—is cited to validate deployment pace. Watch the December 2026 CB-5 completion date, any public updates on Lake Mariner's path from 60 to 500 megawatts, and whether Abernathy announces specific timelines or capacity figures in the next quarter.

TeraWulf's Lake Mariner campus illustrates a widening chasm between announced capacity and delivery. Initial disclosures in October 2024 targeted 410 MW; by May 2025 this had scaled to 750 MW potential through infrastructure upgrades (claim 4997), with 500 MW cited as near-term (claim 4996). As of April 2026—eighteen months after initial announcement—only 60 MW was operationally deployed (claim 2106). The company attributed this gap to phased buildout, with CB-5 expansion (160 MW) promised for delivery by December 31, 2026 (claim 4456). If CB-5 holds to schedule, Lake Mariner would reach approximately 220 MW by year-end, positioning the facility far behind the 500 MW near-term target articulated in February (claim 3649).

Capacity claims have also shifted materially for secondary sites. Cayuga Site capacity was initially proposed at 350 MW (claim 1633), revised upward to 400 MW in development potential (claim 3644), then inexplicably downshifted to 320 MW future capacity (claim 3625)—all within eight months. This trajectory suggests permitting or grid-interconnection constraints that management has not publicly acknowledged.

Acquisition timelines have slipped without explicit revision. The Morgantown Generating Station (210 MW) was promised by December 1, 2025, but the transaction was not completed until after February 1, 2026 (claim 2116). Hawesville Site acquisition similarly missed its February 2, 2026 target, with status unclear as of April (claim 3877). These delays compound: both sites carry long lead times for environmental permitting and grid upgrades.

The Abernathy HPC Campus partnership with Fluidstack (168 MW, 50.1% ownership) remains on the announcement phase with note amortization conditional on campus completion (claim 3954)—suggesting no hard completion guarantee in the financing structure. Independent verification of actual permitting progress, grid interconnection queue positions, or on-the-ground construction activity has not been referenced in TeraWulf disclosures.

In the next ninety days, monitor whether CB-5 construction remains on track for December completion and whether TeraWulf resets capacity guidance downward for 2026. Watch for any revision to Cayuga Site capacity without acknowledgment of prior targets, and track whether Abernathy has achieved environmental clearances and begun ground-level work.

TeraWulf's Lake Mariner facility illustrates a widening gap between capacity promises and operational delivery. Initial October 2024 disclosures outlined a 410 MW facility on 157 acres in Somerset, New York under a 35-year lease (5581, 5582). By May 2025, the narrative shifted to dual targets: 500 MW near-term expansion and 750 MW maximum capacity with transmission upgrades (4996, 4997). These framings remained consistent through February 2026 (3649, 3650). Yet Lake Mariner's actual operational footprint as of April 2026 stood at 60 MW (2106)—roughly one-eighth of the initial commitment—despite continuous project runtime since late 2024.

Facility acquisition timelines have slipped without explicit public acknowledgement. Morgantown generating station promised December 1, 2025 delivery but closed after February 1, 2026 (2116, 1385). Hawesville brownfield acquisition, promised by February 2, 2026, was absorbed into February 1 disclosures (3877, 2115). An S-3 registration commitment due December 8, 2024 similarly delayed (5586).

TeraWulf's strategy has shifted toward risk-sharing partnerships. Fluidstack, introduced August-December 2025, now anchors two major projects: a 200 MW Lake Mariner hosting agreement (1054) and a 50.1% joint venture stake in the 168 MW Abernathy campus (3655, 3641). Google provided a backstop guarantee on certain lease arrangements (4458, 4509). These structured agreements suggest either financing constraints or demand uncertainty requiring external co-anchoring.

Portfolio expansion has accelerated markedly. Twelve sites now appear on TeraWulf's slate, predominantly from May 2025 onward, with most lacking defined capacity targets or timelines. This breadth without proportional Lake Mariner ramp may reflect either genuine multi-region strategy or velocity-focused communications offsetting execution delays.

The CB-5 expansion at Lake Mariner (160 MW) carries a December 2026 completion target (4456). Delivery would bring Lake Mariner to approximately 220 MW, substantially below the 500-750 MW framework. Watch for: Abernathy campus completion timing clarification (construction claims span October-December 2025 without defined end dates); whether Fluidstack deployments accelerate regional buildout; and Cayuga Site interconnection queue visibility, which would signal permitting advancement beyond land control.

TeraWulf's Lake Mariner facility illustrates the gap between announcement and delivery. The Anthropic AI Factory began at 410 MW capacity but expanded to 500 MW near-term and 750 MW ultimate targets by May 2025 (claims 4996, 4997). February 2026 filings reiterated these figures (claims 3649, 3650). Yet April 2026 disclosures show only 60 MW operational (claim 2106), suggesting 15-month progress of roughly 15% of near-term targets. The CB-5 lease expansion promised 160 MW delivery by year-end 2026 (claim 4456), leaving three months to close that gap. A critical data point: TeraWulf withdrew its claimed 606 MW of contracted load across Lake Mariner and Abernathy in February 2026 (claim 3623), signaling that customer commitments may not track announced capacity. At Cayuga, capacity figures have compressed rather than expanded. Initial disclosure of 350 MW (September 2025, claim 1633) grew briefly to 400 MW (February 2026, claims 3644, 3658) before shrinking to 320 MW in the same filing (claim 3625), suggesting permitting constraints are tightening rather than resolving. Acquisitions have slipped: Morgantown generating station, promised for December 1, 2025 (claim 2116), had not closed by April 2026, and the Hawesville brownfield site, promised February 2, 2026 (claim 3877), remained unclosed at the same date. The Form S-3 registration, promised December 8, 2024, was still pending (claim 5586). Abernathy HPC Campus remains announced only, with no independent evidence of ground acquisition or permitting progress as of September 2026. Over the next 90 days, monitor whether CB-5 achieves its December 31 completion deadline, whether Cayuga's permitting resolves or compresses capacity further, and whether actual MW delivered at Lake Mariner breaks 100 MW. Interconnection queue status from regional transmission operators will indicate whether announced megawatt claims reflect actual grid connection pathways.

TeraWulf's facility expansion roadmap shows significant capacity ambition paired with execution risk. The Lake Mariner campus, initially disclosed at 410 megawatts, has since been restated through multiple claims into a tiered structure: 200 megawatts of dedicated HPC capacity with near-term expansion to 500 megawatts and potential long-term ceiling at 750 megawatts conditional on transmission upgrades (claims 4507, 4997, 3650). The progression across different claim dates suggests ongoing internal recalibration, with specificity declining as capacity targets increase.

The withdrawal of a contracted load claim is material. In February 2026, TeraWulf stated 606 megawatts of contracted critical IT load across Lake Mariner and Abernathy (claim 3623), but this figure was subsequently withdrawn. That erasure without explicit acknowledgment of what changed leaves the actual demand backstop unclear. Abernathy's separate 168-megawatt capacity (claim 3655) and its 50.1 percent joint venture with Fluidstack (claim 3633) remain stated, but the vanished aggregate number raises questions about whether customer commitments were renegotiated or restructured downward.

On timelines, Hawesville represents a clear miss. Originally promised by February 1, 2026 (claims 3877, 2115), it remains unexecuted as of September—more than six months overdue without a revised target disclosed. The CB-5 expansion at Lake Mariner still targets end-of-2026 completion (claim 4456), now roughly four months away. Cayuga cycles through overlapping capacity claims, ranging from 320 to 400 megawatts potential, dependent on permitting outcomes not yet evidenced in public records.

The entire portfolio rests on stated claims with no cited permits, interconnection queue positions, or independent validation. Watch whether CB-5 meets its December deadline, whether Hawesville finally closes in Q4, and whether amended customer contracts emerge as new demand anchors to replace the withdrawn 606-megawatt figure. These 90 days will signal whether the campus portfolio represents genuine acquisition and construction progress or further refinement of announced ambitions into smaller realized footprints.

TeraWulf's Lake Mariner Anthropic AI Factory reveals a widening gap between announced capacity and operational delivery. The facility started at 410 MW in initial disclosures but grew to a 750 MW campus target by May 2025 (claim 4997), with a 500 MW near-term expansion flagged as the intermediate milestone (claim 4996). By February 2026, the company continued to cite the same 500 MW near-term and 750 MW potential figures (claims 3643, 3650), but actual operational capacity as of April 2026 stood at only 60 MW (claim 2106). This represents an 88 percent shortfall against the near-term expansion target.

Construction milestones have similarly drifted. The CB-5 expansion, promised at 160 MW with completion by December 31, 2026 (claim 4456, August 2025), remains under construction with fewer than four months remaining. Multiple claims reference vague delivery timelines for Lake Mariner, Akela facilities, and La Lupa expansions without specific dates (claims 3631, 4506, 3629), making it difficult to assess whether the company's expansion roadmap remains credible.

Partnership announcements offer partial visibility. The Beowulf acquisition closed in May 2026 for $52.4 million plus $32 million in earnout milestones (claims 4828, 4829), suggesting operational continuity at Somerset facilities. The Fluidstack joint venture committed 168 MW across multiple sites (claim 1052, October 2025), and Google's backstop guarantee for Fluidstack leases (claims 4509, 4458) provides some demand certainty. However, these partnerships do not directly explain why Lake Mariner's operational capacity remains constrained.

Outside the flagship Lake Mariner project, the pipeline consists entirely of announced facilities with no disclosed capacity or timelines. Abernathy HPC Campus, pitched at 168 MW (claim 3655, February 2026), remains in announcement phase despite repeated timeline claims across December 2025. Cayuga Site has been positioned with 320–400 MW potential (claims 3625, 3658, 3644, February 2026) but shows no permitting progress or operational metrics.

Watch the CB-5 completion status through year-end. Track whether the promised 500 MW expansion at Lake Mariner reaches operational capacity by Q1 2027, or receives another revision. Monitor Abernathy campus groundbreaking and Cayuga permitting activity—both have been repeatedly stated but lack independent confirmation via queue entries or permits.

TeraWulf's facility announcements reflect a pattern of capacity expansion ambitions that outpace demonstrated execution. Lake Mariner, the flagship Anthropic AI Factory, exemplifies this trajectory. Initial disclosures cited 410 MW capacity; by May 2025, the company articulated three distinct capacity metrics simultaneously—500 MW near-term, 50 MW bitcoin mining expansion, and 750 MW maximum with future transmission upgrades (claims 4996, 4995, 4997). By February 2026, messaging consolidated around 500 MW near-term and 750 MW potential (claims 3649, 3650), though the 200 MW HPC-specific capacity cited in August 2025 (claim 4507) remained disconnected from the broader campus narrative. A $3.5 billion financing target announced in July 2026 signals ongoing capital intensity, yet the withdrawn 606 MW contracted critical IT load claim from February 2026 (claim 3623) signals earlier capacity projections did not materialize.

Timeline adherence is mixed. The CB-5 expansion at Lake Mariner carries a promised completion date of December 31, 2026—approximately four months away (claim 4456). Yet TeraWulf has demonstrated serial slippage on supporting infrastructure: Morgantown Generating Station acquisition slipped from December 2025 to unspecified (claim 2116), Hawesville site acquisition from February 2, 2026 to pending (claim 3877), and a Form S-3 filing from December 8, 2024 forward (claim 5586). Each delay has passed without revised public guidance on pipeline impact.

Cayuga Site illustrates escalating claims without corresponding permitting evidence. The company progressed from 350 MW announced capacity in September 2025 (claim 1633) to 320 MW potential in February 2026 (claim 3625) to 400 MW development potential in the same month (claim 3658)—conflicting metrics across adjacent claims. Abernathy HPC Campus similarly emerged from vapor to 168 MW Fluidstack joint venture with 50.1% TeraWulf ownership, though construction timelines remain unspecified (claims 3641, 3655, 3656).

Somerset-Brookings claims layer without clarification: 750 MW infrastructure (claim 4505), 200 MW Fluidstack hosting (claim 1054), and 438 MW La Lupa and Akela contracted load (claim 2110) lack transparency on overlap. Watch the CB-5 delivery deadline in Q4 2026. Any miss would represent the first missed facility completion date and would likely pressure near-term expansion narratives across Lake Mariner, Cayuga, and Abernathy simultaneously.

TeraWulf's Lake Mariner capacity trajectory has expanded consistently: from 410 MW initial disclosure to 500 MW near-term expansion (claims 4996, 3624, 3649) and 750 MW potential with transmission upgrades (claims 4997, 3650). The consistency of these restatements across multiple disclosure windows suggests genuine infrastructure planning rather than opportunistic inflation, though the persistent gap between near-term and maximum capacity indicates material transmission constraints remain unresolved.

Other facilities show compression or revision without acknowledgement. Cayuga Site's initial 350 MW headline (claim 1633) was subsequently reframed as 400 MW potential (claims 3658, 3644) then 320 MW future capacity (claim 3625)—obscuring whether permitting delays or internal forecasting drove the recalibration. More critically, the contracted critical IT load of 606 MW at Lake Mariner and Abernathy (claim 3623) was explicitly withdrawn in February 2026, signaling actual power commitments fell substantially short of public capacity statements.

Timeline credibility is fragmented. The S-3 registration commitment delayed past December 8, 2024 (claim 5586). The Hawesville brownfield acquisition missed its February 2, 2026 deadline (claim 3877). Most construction timelines lack specificity: Lake Mariner expansion (claim 3631), Abernathy (claims 3635, 3656), La Lupa (claim 3629), and Cayuga (claim 1055) reference only generic delivery windows. CB-5 construction completion by December 31, 2026 (claim 4456) is the sole near-term milestone with a firm date, though it represents only 160 MW of the broader portfolio.

Independent validation is absent. No interconnection queue status, permitting records, or construction phase updates substantiate the announced cadence. The NYPA power agreement (claim 3627) exists on paper, but energization into operational capacity remains uncertain. Watch through November 2026 for CB-5 completion confirmation and revised Lake Mariner near-term guidance. Morgantown's note structure conditioning amortization on campus completion (claim 3954) creates financial transparency, but specific commissioning dates remain undisclosed.

TeraWulf's delivery narrative has grown increasingly complex as the company layered multiple projects and partnership frameworks atop its flagship Lake Mariner facility. The stated capacity trajectory reveals persistent ambiguity rather than straightforward growth. Lake Mariner began with a disclosed 410 MW footprint, but by May 2025 the company was describing three separate capacity figures—50 MW for bitcoin mining (claim 4995), 500 MW for near-term expansion (claim 4996), and 750 MW as a long-term potential with transmission upgrades (claim 4997). By February 2026, these had calcified into formal partnership commitments: 500 MW near-term (claim 3649), 750 MW potential (claim 3650), and 200 MW under a specific Fluidstack lease (claim 4507). The portfolio presents these concurrently rather than as an evolution, suggesting the company is packaging different revenue streams—Anthropic's leased capacity, Fluidstack's contracted load, and speculative expansion—under a single facility roof. Similar elasticity appears at Cayuga, which shifted from a 350 MW proposal (claim 1633) in September 2025 to simultaneous claims of 400 MW (claim 3644, 3658) and 320 MW (claim 3625) by February 2026, tied to permitting status that remains opaque. Abernathy contracted backward, from 200 MW in a Fluidstack agreement (claim 1054) to 168 MW as stated capacity (claim 3655).

On delivery timing, the record shows friction. Three major acquisitions—Hawesville (promised February 2, 2026, now seven months overdue per claim 3877), Morgantown power station (promised December 1, 2025, now nine months late per claim 2116), and an S-3 registration filing (promised December 8, 2024, now twenty-one months overdue per claim 5586)—have not moved from "delayed" status. Lake Mariner CB-5 construction carries a December 31, 2026 deadline (claim 4456); with less than four months remaining, this is the nearest hard deadline on the public record. Most new projects remain classified as "announced" with no disclosed completion dates. Independent validation is absent from this record—no permits, queue positions, or on-site progress reports appear. Over the next ninety days, watch whether the CB-5 completion holds and whether any of the three delayed acquisitions enter the actual pipeline, as these would signal whether the company is managing capital constraints or facing execution headwinds.

TeraWulf's disclosed pipeline shows capacity ambition tempered by acquisition slippages. Lake Mariner scaled from an initial 410 MW to 500 MW near-term with 750 MW potential (claims 4996, 4997, 3649, 3650), anchored by the Anthropic AI Factory lease. The CB-5 expansion targets 160 MW by year-end 2026 (claim 4456), now four months away, while a newly announced 90 MW New York Power Authority agreement (claim 3627) addresses upstream power constraints. Abernathy commits to 168 MW through a 50.1 percent Fluidstack joint venture (claim 3655, 3641), and Cayuga claims 400 MW potential contingent on permitting completion (claim 3658). Total disclosed capacity spans 1.3 GW across these three anchors, yet execution shows friction.

Hawesville was promised closed by February 2026 (claim 2115) but remained unresolved through April (claim 2112), a two-month slip cascading into construction readiness. Morgantown Generating Station, officially acquired February 2026, was originally promised December 2025 (claim 2116), with earnout milestones deferred (claim 4829). Neither delay is catastrophic individually, but they suggest procurement timelines are tightening against aggressive schedules.

Independent evidence remains thin. Google backstop guarantees (claims 4458, 4509) and the NYPA agreement (claim 3627) provide some validation, as do Fluidstack's ownership stakes, signaling customer demand exists. However, no permitting documentation, interconnection queue positions, or environmental compliance timelines appear in the disclosed record. Capacity claims for Cayuga hinge explicitly on permitting completion (claim 3658), an undefined variable. The breadth of announced sites—Muskie, La Lupa, Nautilus, Hancock County—lacks any capacity or timeline disclosure, suggesting either early-stage optionality or maturity beyond tracked claims.

Watch CB-5 delivery completion by December 31, 2026 (claim 4456) as a bellwether for Lake Mariner execution. Monitor Hawesville's acquisition closure timing; a second slip would indicate supply-chain stress. Cayuga permitting progress will determine whether 400 MW becomes real or retreats to contingency status. Any revisions to the 750 MW Lake Mariner ceiling without acknowledgement of constraint relief would signal demand absorption rather than delivery confidence.

TeraWulf's capacity claims for Lake Mariner have undergone substantial revision between announcement and current status. The facility was initially disclosed at 410 MW but expanded to a stated target of 750 MW with transmission upgrades by May 2025 (claims 4996–4997). Yet as of April 2026, the site's actual operational capacity stands at only 60 MW—a thirteen-fold gap between stated targets and delivered power. The company has hedged this expansion with multiple timelines: 500 MW near-term and 750 MW maximum (claims 3624, 3649–3650), though these restatements from February 2026 merely repeat earlier thresholds without advancing delivery dates.

Promised timelines have consistently slipped. The Form S-3 registration filing commitment from October 2024 was delayed (claim 5586). Two significant acquisitions—Morgantown Generating Station (promised by December 1, 2025) and Hawesville Site purchase agreement (promised by February 1, 2026)—remain unresolved as of the April 2026 disclosure (claims 2116, 2115), indicating execution challenges in the build-or-buy strategy. The CB-5 construction completion is scheduled for December 31, 2026 (claim 4456), which provides a measurable near-term checkpoint.

Independent evidence is sparse. TeraWulf has announced thirteen data center projects across seven states, yet most carry announced status with no disclosed capacity metrics or specific timelines. The company's withdrawal of a "Contracted critical IT load" claim of 606 MW at Lake Mariner and Abernathy (claim 3623, February 26, 2026) suggests either customer defections or overclaiming. The persistent delta between Fluidstack partnership commitments (200 MW announced August 2025, 168 MW October 2025) and operational delivery raises questions about customer readiness or supply chain constraints.

Watch for: (1) whether CB-5 reaches 160 MW capacity by December 31, 2026 (claim 4456); (2) materialization of the Morgantown and Hawesville acquisitions that have missed prior dates; (3) any restatement of the 500 MW near-term target or new customer announcements for the load that was withdrawn; (4) whether other announced sites—Cayuga, Abernathy, La Lupa—progress beyond announcement status with measurable timelines and capacity disclosure.

TeraWulf's Lake Mariner facility exemplifies a pattern of capacity expansion announcements outpacing construction pace. Initial disclosures targeted 410 MW, which expanded to 750 MW maximum capacity by May 2025 (claim 4997), with a 500 MW near-term expansion promised (claim 4996). Yet April 2026 filings report only 60 MW operational (claim 2106)—suggesting that three-quarters of promised near-term capacity remains incomplete eighteen months after the expansion announcement. The Anthropic partnership, valued at $3.5 billion, includes a 90 MW supply commitment from New York Power Authority (claim 3627), though this backstop itself suggests earlier transmission assumptions proved insufficient. Cayuga Site illustrates how capacity ranges have widened without permitting clarity. Between September and February 2026, announced potential capacity shifted from 350 MW (claim 1633) to 400 MW (claim 3658) to 320 MW (claim 3625), all contingent on undefined permitting timelines. Most facility delivery claims contain no specific dates, only references to construction or timeline (claims 4506, 4174, 3952, 3656, 3635). Core acquisitions have slipped: Morgantown, promised December 2025, remained incomplete as of April 2026 (claim 2116), and Hawesville, promised February 1, 2026, crossed its threshold without public update (claim 3877). The S-3 registration filing promised by December 2024 (claim 5586) also missed its deadline. Fluidstack partnership announcements recycled across portfolio assets from October 2025 through February 2026 (claims 1052, 1054, 3641, 3654, 3966, 3970, 3956), suggesting incomplete prior deployment rather than sequential progress. Abernathy HPC Campus, promised at 168 MW (claim 3655), offers no specified completion date despite repeated construction timeline mentions throughout late 2025. Independent validation is absent. No utility interconnection queue positions, building permits, or equipment delivery documentation have appeared in public filings. The 60 MW versus 500 MW gap at Lake Mariner and missing delivery dates for Abernathy suggest material execution constraints not yet addressed. In the next 90 days, watch for Morgantown acquisition closure and operational status toward its 210 MW target (claim 3881), Cayuga's air and water permit decisions, and whether Lake Mariner construction generates permit activity or third-party interconnection filings indicating progress toward 500 MW. Partnership announcements unsupported by hard asset or construction data warrant careful weighting.

TeraWulf's stated capacity ambitions for Lake Mariner have expanded dramatically from an initial 410 MW to a 750 MW potential ceiling by 2025, though intermediate 500 MW targets have been layered across claims 4996, 3624, and 3649. What complicates this arc is that as of April 2026, only 60 MW remains operationally available at Lake Mariner's core facility (claim 2106), suggesting execution velocity lags announcement velocity by an order of magnitude. Separately, Cayuga has cycled through three distinct capacity framings: 350 MW proposed (claim 1633), revised to 320 MW (claim 3625), then expanded again to 400 MW (claims 3644, 3658)—a pattern that reads less like emerging clarity than recalibration without firm anchoring.

Timeline adherence shows strain. The Form S-3 registration promised for December 8, 2024 carries a delayed tag (claim 5586); the Hawesville acquisition promised for February 2, 2026 similarly did not materialize (claim 3877). The CB-5 expansion at Lake Mariner carries a hard 2026-12-31 deadline (claim 4456) now only 117 days distant, making this the near-term credibility test.

The dataset lacks independent markers—no permit issuance dates, interconnection queue positions, or third-party infrastructure records. The Morgantown acquisition (claims 3880, 3881) brings 210 MW of existing generation, yet integration pathways and operational timelines remain opaque. The 90 MW New York Power Authority agreement for Lake Mariner (claim 3627) represents less than one-fifth of even the near-term 500 MW target, widening the gap between transmission commitments and announced capacity.

The next 90 days will clarify whether 2026's revised timelines hold. CB-5 must deliver or slip; Abernathy's Fluidstack joint venture (claims 3641, 3654) should show material construction progress; and Lake Mariner's expansion from 60 MW toward 500 MW by year-end warrants close tracking. Until actual megawatt-hours flow at scale, the pattern remains one of ascending targets outpacing deployment.

TeraWulf's facility announcements have expanded dramatically in scope while concrete delivery has lagged substantially. The company opened with a 410 MW target for Lake Mariner in late 2024, but by May 2025 revised expectations upward to 500 MW near-term and 750 MW ultimate capacity (claims 4996–4997), messaging that remained unchanged through February 2026 (claims 3649–3650). Yet by April 2026, only 60 MW of Lake Mariner's Data Campus was operationally deployed (claim 2106)—a narrow fraction of the disclosed capacity and evidence of multi-year slippage in expansion timelines.

Partnerships central to claimed capacity show internal inconsistencies. The Fluidstack AI compute arrangement initially promised 200 MW (claim 1054, August 2025) but was rescaled to 168 MW by October 2025 (claim 1052), a 16 percent reduction. More significantly, TeraWulf disclosed 606 MW of contracted critical IT load across Lake Mariner and Abernathy in February 2026 (claim 3623) but subsequently withdrew this claim entirely, signaling post-announcement concerns about capacity commitment status.

Timeline specificity remains sparse. CB-5 carries a concrete completion date of December 31, 2026 (claim 4456), while most expansions lack defined milestones. Two prior commitments slipped measurably: a Form S-3 filing promised by December 8, 2024 was marked delayed (claim 5586), and the Hawesville acquisition initially due February 2, 2026 was similarly delayed (claim 3877).

Capacity targets themselves fluctuated across sites. Cayuga's potential dropped from 350 MW (claim 1633, September 2025) to 320 MW (claim 3625, February 2026) before rising to 400 MW in subsequent messaging (claims 3644, 3658), indicating shifting assumptions about permitting or transmission constraints.

In the next 90 days, watch for CB-5 progress toward its year-end deadline, any clarification on the withdrawn 606 MW contracted load claim, and public interconnection queue filings that would confirm transmission access for Cayuga, La Lupa, and Abernathy. Independent construction evidence beyond Lake Mariner—permits filed, interconnection queue entries updated, on-ground reporting from announced sites—would clarify whether the expansion pipeline is advancing or facing headwinds.

TeraWulf's Lake Mariner campus capacity discourse reveals a widening gap between announced targets and deployed infrastructure. Initial disclosure set 410 MW baseline, escalating to 500 MW near-term expansion (claims 4996, 3649) and 750 MW with transmission upgrades (claim 4997). Yet April 2026 filings show only 60 MW operational at Lake Mariner Data Campus (claim 2106)—an 8-to-12x shortfall six months prior to today. This gap remains publicly unreconciled with revised timelines or acknowledged constraints.

Capacity commitments elsewhere show revision without clear explanation. Cayuga Site capacity swung from 350 MW proposed (claim 1633, Sept 2025) to 320 MW (claim 3625, Feb 2026) back to 400 MW pending permitting (claims 3644, 3658), with no public rationale for the contraction or return. Abernathy HPC Campus committed to 168 MW through Fluidstack at 50.1% ownership (claim 3654, Feb 2026), yet construction timelines remain undated across multiple references (claims 3635, 3656, 3967). The NYPA power supply agreement for Lake Mariner (claim 3627, Feb 2026) pledges 90 MW but does not specify data-center delivery timing.

Promised timelines are sparse and mostly unmonitored. The CB-5 Lake Mariner expansion targets December 31, 2026 for 160 MW (claim 4456)—the only concrete near-term milestone in the portfolio. Hawesville acquisition slipped one day (claim 3877, promised Feb 2). The S-3 registration promised by December 8, 2024 (claim 5586) is flagged as delayed without disclosed revised filing date. Most facility delivery timelines lack published deadlines, making tracking impossible.

Independent evidence of deployment pace does not appear in the record. Lake Mariner remains under construction; NYPA power delivery to operational infrastructure is unconfirmed. No interconnection queue entries, construction permits, or third-party site verification are publicly disclosed. The 170 million-plus in acquisition commitments (Nautilus, Beowulf, Hawesville, Morgantown) reflect consolidation strategy but lack integration milestones.

Over the next 90 days, watch CB-5 completion as a test of near-term execution. Cayuga permitting status—currently blocking 400 MW deployment—requires transparent update. Abernathy construction indicators and NYPA power flow confirmation would anchor capacity rhetoric. Public disclosure of interconnection queue positions would establish credibility.

TeraWulf's Lake Mariner Anthropic AI Factory has undergone significant capacity recalibration during its construction phase. Initial disclosure positioned the site at 410 MW; by mid-2025, the company had announced a 750 MW campus target (claim 4997, May 8, 2025). This threefold expansion, however, brackets a more complex reality: 750 MW requires transmission infrastructure upgrades, while a 500 MW near-term expansion represents the intermediate target (claim 4996, May 8, 2025). By February 2026, TeraWulf emphasized 606 MW of contracted IT load capacity (claim 3623), though this characterization was subsequently withdrawn, suggesting internal recalibration between committed and aspirational capacity. The 410 MW initial facility remains under construction, with no public completion date disclosed.

TeraWulf identified CB-5, a 160 MW lease expansion (claim 4455, Aug 17, 2025), as the only facility component with a stated deadline: December 31, 2026 (claim 4456). No timeline has been provided for the 500 MW near-term or 750 MW ultimate buildout, leaving the delivery path for the majority of announced capacity materially undefined.

Execution has proven uneven. The company missed its October 2024 Form S-3 filing commitment by December 8, 2024 (claim 5586). The Morgantown Generating Station acquisition, targeted for December 1, 2025, and the Hawesville property purchase, promised for February 1, 2026, both remained unresolved by April 2026 (claims 2116, 2115). These are not administrative delays but acquisitions critical to claimed expansion pathways.

TeraWulf's announced portfolio spans thirteen sites, but most lack capacity specifications. Cayuga Site and Abernathy HPC Campus carry stated targets (400 MW and 168 MW respectively), with no delivery dates. The portfolio suggests a strategy of acquiring power and land assets while anchoring demand through customer partnerships: Anthropic at Lake Mariner, Fluidstack across multiple campuses, and Google backstop guarantees (claim 4509). The core facility remains under construction against its 410 MW specification.

Over the next 90 days, watch for: CB-5 construction milestones toward December completion; resolution of delayed Morgantown and Hawesville acquisitions; any update on 500 MW near-term expansion timing; and NYPA power allocation deployment against the 90 MW agreement (claim 3627) underlying claimed utilization growth.

TeraWulf's Lake Mariner Anthropic AI Factory has undergone significant revisions to both capacity targets and delivery pace over twenty months. Initial disclosure pegged the facility at 410 MW; by May 2025, the company announced a near-term expansion to 500 MW and a potential 750 MW campus build (claims 4996, 4997). This two-figure jump suggested aggressive platform acceleration. However, April 2026 filings reveal only 60 MW currently operational at Lake Mariner (claim 2106), a stark discrepancy from the earlier rhetoric and indicating either conservative disclosure discipline or execution friction.

Promised dates have been sparse and vague. The CB-5 construction phase carries a deadline of December 2026 (claim 4456), but most capacity milestones lack firm dates—merely referencing "construction and delivery timeline" without specifics (claims 3635, 4506). The Hawesville acquisition, promised by February 2, 2026 (claim 3877), is marked delayed. Abernathy HPC Campus appears repeatedly in filings starting October 2025 as a 168 MW joint venture with Fluidstack (claims 3654, 3655), yet no ground-truth construction updates appear in the record.

Independent validation exists but is thin. The New York Power Authority committed 90 MW of supply (claim 3627), and Fluidstack inked a 10-year, 200 MW AI hosting agreement (claim 1054) with Google backstop protection (claim 4509). These suggest genuine anchor demand. However, these partnerships do not confirm facility readiness—only contractual willingness. Permitting records and utility interconnection queue positions are absent from the disclosure set.

The gap between 410 MW announced and 60 MW delivered over eighteen months raises questions about timeline credibility. The 500 MW near-term and 750 MW long-term figures have persisted unchanged from May 2025 through February 2026 (claims 4996, 3649), suggesting either stable ambition or frozen projections. Watch the CB-5 completion by year-end 2026 and any revised operational capacity guidance from TeraWulf's next quarterly filing. If the 60 MW figure remains static into Q4 2026, the entire expansion narrative may require recalibration.

TeraWulf's Lake Mariner capacity narrative has expanded from an initial 410 MW disclosure to a stated potential of 750 MW, but the composition and timing of this growth remains fragmented across overlapping partnership claims. Near-term expansion to 500 MW is repeatedly cited (claims 4996, 3649, 3643), while the 750 MW ceiling appears contingent on transmission infrastructure upgrades (claim 4997) and appears to conflate Lake Mariner proper—allocated 200 MW of dedicated HPC capacity (claim 4507)—with the broader Somerset-Brookings leasehold footprint (claim 4505). Intermediate capacity tranches are scattered: Akela's G42 partnership pledges 70 MW (claim 1051), Fluidstack secures 200 MW under a 10-year hosting agreement (claim 1054), and the CB-5 lease expansion represents 160 MW (claim 1053). This vertical stacking of commitments without clearly sequenced delivery gates raises questions about timing and availability sequencing.

The only commitment with a hard deadline is CB-5 construction completion by end of 2026 (claim 4456), now four months away. The Abernathy HPC Campus, positioned as a secondary anchor with 168 MW (claims 3655, 1052), carries no specific delivery date in the record, only vague "construction and delivery timeline" language from February 2026 (claim 3656). Cayuga Site capacity grew from an announced 350 MW (claim 1633, September 2025) to 400 MW potential (claim 3644, February 2026) without explanation, though both figures are explicitly contingent on future permitting. Two timeline commitments have already slipped: Hawesville brownfield acquisition was promised by February 2, 2026 and is marked delayed (claim 3877), and the Form S-3 registration missed its December 2024 target (claim 5586).

The February 2026 withdrawal of a 606 MW "contracted critical IT load" claim (claim 3623) is notable—it suggests either overcounting or strategic recalibration, and no replacement metric was offered. Partnership dependencies amplify execution risk: Google backstop guarantees underpin Fluidstack leases (claim 4509), the Abernathy campus is 50.1 percent Fluidstack-controlled (claim 3633), and NYPA supplies only 90 MW of committed power to Lake Mariner (claim 3627). Watch whether the CB-5 completion meets its December 2026 gate and whether Abernathy's construction timeline emerges with specificity by November. Any extension of CB-5 or absence of Abernathy timeline detail would signal execution slippage beginning to compound.

TeraWulf's Lake Mariner facility announcement shows notable capacity expansion within a compressed timeframe. The 410 MW initial capacity disclosed on July 15 was already labeled a revision, suggesting prior guidance existed—but the historical baseline remains unclear from available records. By August 3, barely three weeks later, TeraWulf introduced a 750 MW campus target, representing an 83 percent increase in stated power capacity. This rapid upward revision raises questions about whether the initial 410 MW represented a conservative opening position or a genuine rescaling based on Anthropic's demand signals.

The $3.5 billion financing target anchors the investment case, though the relationship between that figure and the expanded 750 MW target has not been articulated. At $4.67 per watt, the economics appear consistent with hyperscale AI infrastructure builds, but no formal cost-per-watt breakdown has been disclosed for either phase.

Notably absent from the public record are permitting milestones, utility interconnection queue status, or construction progress metrics that would independently verify the stated under construction status. New York's Article VII power plant certification process, if applicable, typically involves multi-year lead times; no filings or hearing dates are evident. Similarly, the grid interconnection queue position—critical for a 750 MW facility—has not been publicly disclosed. Without these data points, the trajectory depends entirely on company assertion.

Anthropic's leasing commitment likely drives the rapid upward revision, and securing that anchor tenant validates the business case. However, the 3.8x leverage between initial and expanded capacity warrants scrutiny of technical readiness and financing certainty.

Over the next 90 days, watch for specific operational milestones from TeraWulf or Anthropic confirming construction phasing, any disclosure of state regulatory filings or queue positions from NYISO, revised timelines or cost updates if financing encounters obstacles, and communication about which phases map to which capacity tranches.