Marathon Digital MARA NASDAQ

Marathon's record shows a company that moved through three distinct financing regimes, each roughly an order of magnitude larger than the last. Through 2019 and mid-2020 it was a micro-cap relying on small at-the-market capacity ($7.47M, per claims #15203 and #14915), a $2.3M all-stock placement (claim #15151), and repricing of a $999K convertible (claim #14913). The $100M S-3 in August 2020 (claim #14874) and the $200M ATM in May 2021 (claim #14651) marked the shift to institutional-scale equity, followed by the $747.5M aggregate 2026 convertible in November 2021 (claims #14387, #14388, #14279). The third regime arrived in 2024, with $489.3M raised through the ATM in a single quarter (claim #6207) and roughly $1.85B of new convertibles across November and December (claims #5398, #5342). Recurring equity-plan share increases (claims #14758, #14555, #6539) add a steady layer of dilution beneath these larger events. The mix reads as a deliberate equity-plus-convert structure, although the bank-debt layer was not durable. The Silvergate revolver and term loan (claims #14440, #13836) were partly drawn at $35M (claim #13817) and then terminated in March 2023 with notice and prepayment terms waived (claims #13180–13182). Deployment has diverged from announced purposes at least once. The $67M Compute North bridge loan and $14M in expedite fees (claims #14282, #14283) later surface as an unsecured bankruptcy claim (claim #12985). That outcome is consistent with expansion capital that did not convert into the intended capacity. The 2024 converts state two uses, bitcoin purchases and retirement of the 2026 notes (claims #5401, #5311). The repurchase figures are revised across filings, from $199M to $212M for the November deal (claims #5425, #5400) and among $48M, $50M and $51M for December (claims #5345, #5349, #5310). An August 2024 convertible issuance appears only in later filings (claims #5254, #5231) and has no announcement claim of its own, which requires explanation. Several 2022–2024 claims also mix units, recording millions as raw dollars. No insider transactions are on record, and the dataset stops in December 2024. A careful reader should therefore look for any residual 2026 notes approaching their December 1, 2026 maturity (claim #14387), the holder repurchase dates on the 2024 notes (claim #5405), disclosure that reconciles bitcoin purchases with the stated use of proceeds, and whether the $200M credit line (claim #5574) proves more durable than its predecessor.

Marathon's disclosed financing history falls into three phases, and the scale changes sharply between them. Through 2019 and mid-2020 the company relied on small, repeated equity raises. These included ATM programs sized near $7.47M (claims #15203, #14915), an all-stock private placement (claim #15151), and a stockholder approval to issue more than 19.99% of shares (claim #15048). A convertible note amendment that lowered the conversion price (claim #14913) also falls in this period. Taken together, this looks more like keeping the company funded than executing a plan. The August 2020 $100M S-3 (claim #14874) and the $23.2M Antminer purchase (claim #14843) mark the start of a more coherent build-out. The $200M ATM (claim #14651), the $100M Silvergate revolver (claim #14440) and the $747.5M of 1.00% 2026 convertibles (claims #14279, #14386) then established a layered structure of equity, bank debt and converts.

The bank-debt layer did not last. Silvergate added a $100M term loan in August 2022 (claim #13836), and the facilities were terminated by March 2023 with prepayment and notice terms waived (claims #13180, #13181, #13182). The $67M bridge loan and $14M of expedite fees advanced to Compute North (claims #14282, #14283) later appear as an unsecured bankruptcy claim (claim #12985). The stated amount of that claim requires reconciliation against the original exposure.

The 2024 phase is larger again. It includes a $489.3M first-quarter ATM (claim #6207), a $1B convertible with $980M of stated net proceeds (claims #5398, #5399), and an $850M convertible with $835M of net proceeds (claims #5307, #5308). The announced use of proceeds shifted from mining infrastructure to bitcoin acquisition (claims #5401, #5311) and to partial repurchases of the 2026 notes, about $212M and $51M (claims #5400, #5310). The record does not show actual bitcoin purchases or completed repurchases, so deployment against the stated plans cannot be confirmed here.

Several figures conflict in ways that require explanation. The December offering is recorded as both $700M and $850M (claims #5348, #5342). The Abu Dhabi contributions appear as $81.2M and $406M (claims #12974, #13085). No insider transactions are on record, and the data ends in December 2024.

The items to watch follow from this record. One is the residual 2026 note balance as that maturity arrives. Another is the dilution path implied by the roughly $26 conversion price on the new notes (claim #5402) and the holder repurchase rights (claim #5405). A third is whether ATM issuance continues alongside the convertibles.

Marathon's record shows three distinct financing regimes, and the dilution trajectory shifts with each. Through 2019 and 2020 the company was raising in single-digit millions through small ATM capacity (the $7.47M program, per claims #15203 and #14915), an all-stock private placement (claim #15151), share-settled equipment purchases (claim #15066), and a reduced conversion price on an existing note (claim #14913). The stockholder approval for issuance above 19.99% (claim #15048) shows how much the equity base was being stretched relative to its size. That period reads as opportunistic patching rather than a plan. The 2021 step-up to a $200M ATM (claim #14651) and the $747.5M of 1% convertibles (claims #14279, #14386) marked a move toward structured capital. The Silvergate revolver and term loan, however, carried liquidity and net-worth covenants ($25M and $350M, per claims #14444 and #13839) and needed a waiver just to accommodate the converts (claim #14399). Those facilities were terminated in March 2023 with fees waived (claims #13180, #13181), so the debt leg of that strategy did not last. Deployment against stated purpose requires explanation in two places. The Compute North bridge loan and expedite fees ($67M and $14M, per claims #14282 and #14283) are later reflected only as an unsecured bankruptcy claim (claim #12985). The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085), and the record does not reconcile the two. By late 2024 the stated use of proceeds had shifted from mining capacity to bitcoin accumulation and refinancing the 2026 notes (claims #5401, #5400, #5310). The $489.3M Q1 2024 ATM (claim #6207) and roughly $1.85B of new converts (claims #5398, #5307) made this the heaviest issuance period on record, and actual bitcoin purchases are not documented here. Insider activity on record is limited to RSU grants (claims #13964, #13965), with no reported sales. The record ends in December 2024, so the reader should treat 2025 onward as undisclosed in this dataset. The nearest item to watch is the residual 2026 notes maturing December 1, 2026 (claim #14387): repurchases of roughly $212M and $48-51M (claims #5400, #5345) leave a balance that the record does not state. Beyond that, the conversion terms on the 2024 notes (claim #5402) set the next layer of potential dilution.

Marathon's record divides into three phases, and the shift between them matters more than any single transaction. Through 2020 the company financed itself in small increments: a roughly $7.5M ATM capacity that recurs across several filings (claims 15203, 14915, 14848), a $6.9M underwritten offering (claim 14850), and amendments that lowered the conversion price on a $999,106 note (claims 14913, 14892). That is incremental equity funding of a company that did not yet have scale. The 2021 phase is a step change. A $200M ATM (claim 14651) was followed by a $100M Silvergate revolver (claim 14440) and $747.5M of 2026 convertibles (claim 14279). The mix moved from dilution toward leverage once equity values could support it. The period after 2022 requires explanation. Silvergate's facilities, including the $100M term loan added in August 2022 (claim 13836), were terminated in March 2023 with notice and prepayment waivers (claims 13180–13182). Equity issuance then carries much of the load again: $489.3M through the ATM in Q1 2024 alone (claim 6207), followed by a new $200M line of credit (claim 5574) and a late-2024 run of convertibles. The record reads as coherent in direction, using equity when it is cheap and converts when volatility is priced in, but it is reactive in timing. The late-2024 convertible record is internally inconsistent. Principal appears as $850M (claims 5342, 9275), $1B (claim 5398) and $700M (claim 5348). The planned repurchase of 2026 notes appears variously as $199M, $212M, roughly $50M and $48M (claims 5425, 5400, 5349, 5345). These may be separate tranches, but the record does not reconcile them. The stated use of proceeds, buying additional bitcoin and retiring 2026 notes (claims 5401, 5311), has no matching deployment figures on record, so tracking cannot be confirmed. Earlier commitments show weaker follow-through. The $67M bridge loan to Compute North (claim 14282) is later accompanied by an unsecured bankruptcy claim (claim 12985), and the Abu Dhabi contribution appears at inconsistent magnitudes (claims 13085, 12953, 12974). No insider sales are on record. A reader should watch for reconciled totals on the 2024 notes, the remaining 2026 maturity, and the conversion terms (claim 5402) relative to the share price. Further ATM usage would also show whether equity is again the default funding source.

Marathon's financing record falls into three phases. Through 2020 the company raised capital in small increments sized to its shelf: ATM capacity of roughly $7.5M (claims #15203, #14915), a $6.9M underwritten deal (claim #14850), and share-settled payments for equipment and facility services (claims #15066, #14733). The post-2019 share count of about 8.7M (claim #15152, recorded in dollar units but apparently a share figure) shows how small the base was. In 2021 the scale changed: a $200M ATM (claim #14651, completed per claim #14325), then $747.5M of 2026 convertibles including the $97.5M option (claims #14387, #14279). The 2024 phase moved to larger blocks again: $489.3M of ATM issuance in Q1 alone (claim #6207), followed by back-to-back convertible offerings of $1B gross (claim #5398) and $850M (claim #5342, verified). Dilution has therefore come both directly through the ATM and through conversion overhang that has compounded across note vintages.

The mix reads as a deliberate pairing of ATM equity for steady funding with convertibles for step-changes. The bank-debt layer is the exception. Silvergate facilities were expanded with a $100M term loan in August 2022 (claim #13836), drawn $35M (claim #13817), and terminated by March 2023 (claim #13180). That bank-lending channel was effectively no longer discussed until a $200M line of credit appeared in October 2024 (claim #5574), with no counterparty disclosed here.

On use of proceeds, the November 2024 notes were tied to repurchasing 2026 notes ($199M per claim #5425, revised to $212M per claim #5400) and acquiring bitcoin (claim #5401). The December notes repeat both uses at a smaller repurchase size ($48M to $51M across claims #5345, #5349 and #5310). The record contains no claim confirming the amounts actually repurchased or bitcoin actually bought, so whether deployment tracked the stated uses cannot be assessed from this data. Earlier commitments also have unclear outcomes: the $67M Compute North bridge loan (claim #14282) is followed only by an unsecured bankruptcy claim recorded at $40,000 (claim #12985), and the Abu Dhabi figures appear as both $406M and $81.2M (claims #13085, #12974). Both require explanation, as does the December offering recorded at $700M (claim #5348) against $850M (claim #5342).

The data contains no open-market insider sales, only RSU grants (claims #13964, #13965). Readers should watch for disclosed repurchase completions, the residual 2026 note balance, further ATM usage, and the terms of the $200M credit line.

Marathon's financing record moves through three distinct scales. Through 2020 the company raised capital in single-digit millions, with ATM capacity of roughly $7.47M (claim #15203) and a $6.9M underwritten offering (claim #14850). In 2021 it stepped up to a $200M ATM (claim #14651) and $747.5M of convertible notes due 2026 (claim #14279). By 2024 it was issuing at a still larger scale: $489.3M sold through the ATM in a single quarter (claim #6207), a roughly $1B convertible in November (claim #5398) and a further $850M in December (claim #5342). Dilution therefore accelerated rather than plateaued, and it shifted from direct share issuance toward convertible instruments whose equity effect is deferred. The recurring increases to the equity incentive plan (claims #14758, #14555 and #6539) add a quieter, steady layer on top.

The instrument mix reads as partly deliberate and partly reactive. The 2021–2022 debt layer was a $100M Silvergate revolver (claim #14440) plus a $100M term facility (claim #13836). Both were terminated in March 2023 with prepayment charges and notice periods waived (claims #13180 and #13181), and a $200M line of credit from an unnamed lender appears only in October 2024 (claim #5574). The small $14.25M Series A preferred placement in 2023 (claim #12532) fits no visible pattern. The repeated ATM-then-convertible sequencing is the most coherent thread.

Use of proceeds requires some explanation. The Compute North bridge loan of $67M (claim #14282) and expedite fees of $14M (claim #14283) were followed by a bankruptcy claim (claim #12985) whose recorded figure is not reconcilable as stated. Abu Dhabi joint venture contributions appear as $81.2M (claim #12974) and as $406M (claim #13085), which may reflect the company's share versus total project cost but is not clarified. The planned repurchase of 2026 notes was sized at about $199–212M for the November 2024 deal (claims #5425 and #5400) and only about $48–51M for the December deal (claims #5345 and #5310). The bitcoin acquisitions stated as a use of proceeds (claims #5401 and #5311) have no deployment figures on record.

No insider sales appear in the data; the only insider-linked items are RSU grants (claims #13964 and #13965). A careful reader should watch three things. First, the December 1, 2026 maturity of the 2026 notes (claim #14387), now weeks away, and how much principal remains, since claim #5230's figure is ambiguous in units. Second, the holder repurchase dates on the 2024 notes (claim #5405). Third, whether the $700M figure in claim #5348 is reconciled with the $850M recorded elsewhere.

Marathon's record shows three distinct financing regimes rather than one continuous strategy. Through 2020 the company relied on small ATM programs and share-settled transactions, such as the $7.47M ATM capacity per claim 15203, the 2,335,000-share private placement per claim 15151, and equity paid for facility services and milestone bonuses per claims 14796 through 14799. This is typical of a microcap paying with stock because cash was scarce. The 2021 pivot was abrupt: a $200M ATM per claim 14651, then $747.5M of 2026 convertibles per claim 14279, alongside a $100M Silvergate revolver per claim 14440. That combination of equity, low-coupon converts and bank debt looks deliberate for that period. The bank leg did not last. Silvergate added a term loan in August 2022 per claim 13836, and both facilities were terminated by March 2023 per claim 13180, with waivers of notice and prepayment charges per claims 13182 and 13181. After that, the record shows no replacement bank facility until the $200M line of credit per claim 14588 is not present; the next one is per claim 5574 in October 2024. Equity issuance in the interim was heavy, with $489.3M raised via ATM in Q1 2024 alone per claim 6207. In late 2024 the company issued two convertible tranches in close succession: $1B completed per claim 5398 and $850M per claim 5307. Both state dual uses, acquiring bitcoin and repurchasing the 2026 notes, per claims 5401, 5400, 5311 and 5310. The stated repurchase amounts vary across filings ($199M per claim 5425, $212M per claim 5400, roughly $50M per claims 5349 and 5310), and the record does not show how much was actually retired or how much bitcoin was bought. That reconciliation requires explanation. Similarly, the Abu Dhabi contribution appears as $81.2M per claim 12974 and $406M per claim 13085, and the $67M Compute North bridge loan per claim 14282 resurfaces only as an unsecured bankruptcy claim per claim 12985. Many figures in this record are logged in inconsistent units, which limits precision. No insider transactions are on record, so that dimension cannot be assessed. A careful reader should watch for actual 2026-note repurchase volumes, the conversion terms on the 2024 notes per claim 5402, and whether the ATM resumes alongside the convertibles.

Marathon's record shows two distinct financing regimes. Through 2020 the company relied on small, repeated equity raises, including a $7.47M shelf and ATM capacity (claims #15228, #15203), an all-stock private placement of 2.335M shares (claim #15151), and stockholder approval to issue beyond 19.99% (claim #15048). It also amended a convertible note to reduce its conversion price (claims #14913, #14892). This was incremental patching, typical of a sub-scale issuer funding equipment such as the $23.2M Antminer purchase (claim #14843) as each need arose. From 2021 the scale changes. The company used a $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440), and $747.5M of 2026 convertibles (claim #14279), which required a covenant waiver from the same lender (claim #14399). That sequence suggests the debt and convertible layers were not fully coordinated at inception. The Silvergate facilities, including a later $100M term loan (claim #13836), were terminated in March 2023 with prepayment charges waived (claims #13180, #13181), and lender capacity was later rebuilt through a $200M line (claim #5574). In 2024 the company returned to heavy equity issuance, with $489.3M of ATM sales in a single quarter (claim #6207), followed by roughly $1B (claim #5398) and $850M (claim #5307) of new convertibles. The coherent thread is refinancing the 2026 notes while funding bitcoin accumulation. The record contains no insider sales, only RSU grants (claims #13964, #13965) and repeated incentive-plan increases (claims #14758, #14555, #6539). On deployment, the stated repurchase amounts were revised between filings: $199M became $212M for the November deal (claims #5425, #5400), and roughly $48–51M was cited for December (claims #5345, #5310). A $700M figure for the December offering (claim #5348) does not match the $850M reported elsewhere (claim #5342) and requires explanation. The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085), and the Compute North bridge loan of $67M (claim #14282) later surfaces only as an unsecured bankruptcy claim (claim #12985). The disclosures do not show how the stated bitcoin purchases were actually executed (claims #5401, #5311). The next items to watch are the treatment of any residual 2026 notes against the December 1, 2026 maturity (claim #14387), and the conversion terms on the 2024 notes (claim #5402), which set the next leg of potential dilution.

Marathon's record divides into three phases, each with a different dominant instrument. Through 2020 the company relied on small equity pieces: an ATM sized at roughly $7.5M (claims #15203, #14848), a $6.9M underwritten offering (claim #14850), and shares issued directly for equipment and facility services (claims #14796, #14733). That pattern looks like survival financing rather than a planned capital structure. The $100M S-3 (claim #14874) and the $200M ATM (claim #14651) mark a step change in scale. Late 2021 adds leverage: the $100M Silvergate revolver (claim #14440) and $747.5M of 2026 convertibles (claims #14279, #14387, #14388). The convertibles needed a covenant waiver from the same lender (claims #14399, #14309), which suggests the instruments were layered as they came rather than sequenced in advance. The Silvergate facilities were then terminated in March 2023 with notice and prepayment terms waived (claims #13180, #13181). After that, bank debt is no longer part of the record until a $200M line of credit appears in October 2024 (claim #5574).

Dilution comes mainly through the ATM. The record shows $161M issued by mid-2022 (claim #13818) and $489.3M in Q1 2024 alone (claim #6207), so the pace of equity issuance accelerated sharply. The late-2024 convertibles, at least $1B (claim #5398) and $850M (claim #5342), shift the mix back toward deferred dilution, with conversion priced near $26 per claim #5402.

On use of proceeds, both 2024 offerings named two uses: buying bitcoin (claims #5401, #5311) and repurchasing the 2026 notes. The repurchase figures move within each offering. The November figure is $199M in claim #5425 and $212M in claim #5400, and the December figure appears as $48M, $50M and $51M (claims #5345, #5349, #5310). Claim #5348 also lists a $700M December offering, which conflicts with the $850M recorded elsewhere and needs explaining. Earlier stated deployments are harder to trace. The $67M Compute North bridge loan (claim #14282) is followed by a bankruptcy unsecured claim (claim #12985), and the Abu Dhabi contributions appear as both $81.2M and $406M (claims #12974, #13085).

The record contains no insider transactions, so nothing can be said about insider selling. A sophisticated reader should watch three things: how much of the 2026 notes remains after the repurchases, how ATM issuance develops beyond the Q1 2024 disclosure, and whether bitcoin purchases funded by the convertibles are reported in amounts that reconcile with the net proceeds disclosed in claims #5399 and #5308.

Marathon's financing record falls into three phases, each with a different logic. Through 2020 the company was a micro-cap held together by small fixes. It ran shelf and ATM capacity sized at about $7.47M (claims #15228, #15203). It placed 2,335,000 shares privately against roughly 8.72M shares outstanding afterward (claims #15151, #15152), which is dilution of about a quarter of the float in one step. It needed stockholder approval to exceed the 19.99% threshold (claim #15048). It lowered the conversion price on the Bi-Coastal note twice (claims #14913, #14892). Paying for mining equipment and facility services in shares (claims #15066, #14796, #14733) points to scarce cash rather than a capital plan.

From 2020 to 2021 the scale changed. The company filed a $100M S-3 (claim #14874) and a $200M ATM (claim #14651), then turned to leverage: a $100M Silvergate revolver (claim #14440) and $747.5M of 2026 convertibles including the option (claim #14279). That reads as a deliberate move toward non-dilutive capital at a time of high equity valuations. In 2022 the ATM was still running ($161M, claim #13818) while the revolver was drawn ($35M, claim #13817), so the shift was incomplete.

The record of how proceeds were used is weakest in the middle period. The $67M bridge loan and $14M in expedite fees paid to Compute North (claims #14282, #14283) are followed in the record by an unsecured bankruptcy claim (claim #12985), and that sequence needs explaining. The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085). The Silvergate facilities ended early (claim #13180), and the record does not show a replacement until the $200M line in 2024 (claim #5574).

The 2024 issuance is the largest, and its stated purpose differs from earlier rounds. A $489.3M ATM in a single quarter (claim #6207) was followed by two convertible deals: about $1B with net proceeds of roughly $980M (claims #5398, #5399), and $850M with roughly $835M net (claims #5342, #5344). Both name bitcoin purchases as the main use (claims #5401, #5311), with comparatively small repurchases of the 2026 notes, about $212M and $48–51M (claims #5400, #5345, #5310). The capital is now funding a balance-sheet asset rather than mining capacity. One claim gives the December deal as $700M (claim #5348), which does not match the others.

The record contains no insider sales, only RSU grants (claims #13964, #13965). A careful reader should look for three things: how the remaining 2026 convertibles, due December 2026, are settled; whether ATM use continues alongside the convertibles; and whether disclosed bitcoin holdings account for the stated proceeds. No financing claims are on record after December 2024.

Marathon's record describes a company whose financing scale grew far faster than its share count history alone would suggest. The early period is small and equity-heavy: a shelf and ATM program sized at roughly $7.5M (claims 15228, 15203), an all-stock private placement (claim 15151), and repeated share issuance for equipment and facility services (claims 15066, 14796, 14733). Funding moved to a $100M S-3 (claim 14874) and a $200M ATM (claim 14651), then shifted toward debt-like instruments with the $747.5M of 2026 convertibles (claim 14279), alongside Silvergate revolving and term facilities (claims 14440, 13836). The 2024 activity is larger again: an ATM raising $489.3M in a single quarter (claim 6207), a $200M credit line (claim 5574), and two convertible offerings described as $1 billion (claim 5398) and $850M (claim 9275). Dilution has therefore moved from direct issuance toward deferred, conversion-contingent dilution, with Equity Incentive Plan reserves expanded several times (claims 14758, 14555, 6539).

The mix reads less as a fixed plan than as a series of responses to available windows. The Silvergate relationship required a covenant waiver to permit the 2021 convertibles (claim 14309) and ended in termination in March 2023 (claim 13180). The $67M bridge loan to Compute North (claim 14282) later appears only as an unsecured bankruptcy claim (claim 12985), and the recovery is no longer discussed in the record.

On use of proceeds, the 2024 convertibles were tied to bitcoin purchases and to repurchasing 2026 notes. The stated repurchase amounts were revised from about $199M to $212M (claims 5425, 5400) for the November deal and stated at about $48M to $51M for December (claims 5345, 5310). The record contains no disclosed repurchase totals, so deployment cannot be reconciled. Several figures also require explanation: the December offering appears at both $700M (claim 5348) and $850M (claim 5307), and the Abu Dhabi contribution appears as $81.2M (claim 12974) and $406M (claim 13085). No insider sales are on record, so that dimension cannot be assessed.

A careful reader should watch for how the remaining 2026 notes due December 1, 2026 (claim 14387) are settled, for the holder repurchase dates on the 2024 notes (claim 5405), for any further ATM capacity, and for disclosure of how much of the 2024 proceeds actually retired older debt versus funded bitcoin purchases.

Marathon's record shows a company whose financing grew far faster than its own early instruments would suggest, with an equity-led phase giving way to convertible debt as the main funding tool. Through 2019 and early 2020 the company relied on small at-the-market programs sized around $7.5M (claims #15203, #14915), alongside share issuances tied to equipment and stockholder approval to exceed the 19.99% threshold (claim #15048). The step to a $100M S-3 (claim #14874) and then a $200M ATM (claim #14651) marks the point where dilution moved from incremental to structural. The ATM habit continued well after that, with $161M issued per claim #13818 and $489.3M in a single quarter of 2024 per claim #6207, so equity dilution never paused while the convertibles were being added on top. The mix reads as coherent in direction but improvised in execution. The $747.5M of 2026 convertibles (claim #14279) set the template, and the late-2024 notes repeated it at larger scale: $1B completed with $980M net (claims #5398, #5399), followed within weeks by $850M with $835M net (claims #5307, #5308). Bank debt was the less durable part of the mix. The Silvergate revolver and term loan (claims #14440, #13836) were terminated with waived fees in March 2023 (claims #13180, #13181), and a separate $200M line appeared later (claim #5574). Use of proceeds requires closer reading. The $67M bridge loan and $14M of expedite fees to Compute North (claims #14282, #14283) were followed by an unsecured bankruptcy claim (claim #12985), and that capital is no longer discussed as productive deployment. The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085), a gap that requires explanation. The late-2024 offerings named bitcoin purchases and repurchases of the 2026 notes ($212M and $51M per claims #5400, #5310), but the record does not confirm the repurchases were completed, and one claim gives the December raise as $700M rather than $850M (claim #5348). The record contains no insider sales, only RSU grants (claims #13964, #13965), and it shows no events after December 2024. A careful reader should watch for how the remaining 2026 notes due December 1, 2026 (claim #14387) are settled, whether by cash, refinancing or conversion, and whether ATM issuance resumes to fund that settlement.

Marathon's record describes a company whose capital-raising scaled by roughly two orders of magnitude while the instruments became more sophisticated. The 2019–2020 period consists of small, repeated H.C. Wainwright at-the-market programs sized near $7.5M (claims #15203, #14915, #14848), a small underwritten offering (claim #14850), and an amendment reducing a convertible note's conversion price (claim #14913). That mix reads as patching rather than plan. The pivot comes with the $100M S-3 (claim #14874) and the $200M ATM (claim #14651), followed by $747.5M of 2026 convertibles (claim #14279). Dilution then shifted from frequent small equity draws to large episodic events: $161M of ATM issuance by mid-2022 (claim #13818), $489.3M in Q1 2024 alone (claim #6207), and successive convertible raises in late 2024 of $1B (claim #5398) and $850M (claim #5307).

The later mix looks more coherent. Equity funds operations, convertibles fund balance-sheet positioning, and secured credit is used sparingly. The Silvergate revolver and term loan (claims #14440, #13836) carried liquidity and net-worth covenants (claims #14444, #13839) and were terminated with waived prepayment and notice terms (claims #13180, #13181). A $200M line of credit later replaced that capacity (claim #5574).

The record does not let a reader trace how announced proceeds were actually deployed. The 2024 convertibles were earmarked for repurchasing 2026 notes and acquiring bitcoin (claims #5400, #5401, #5310, #5311), but no claim confirms the repurchase amounts executed or the bitcoin bought. Earlier, the $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) were followed by that counterparty's bankruptcy and an unsecured claim (claim #12985). The recovery is not discussed further. The Abu Dhabi joint venture contribution appears as $81.2M (claim #12974) and as $406M (claim #13085); that difference requires explanation. The record contains RSU grants (claims #13964, #13965) but no insider sales, so it says nothing about insider selling.

Several amounts are recorded in inconsistent units, and the record ends in December 2024, so this read cannot assess anything since. A reader should next watch four things. The first is how much of the 2026 notes remain outstanding. The second is the share price relative to the roughly $26 conversion price (claim #5402) and the holder repurchase date (claim #5405). The third is whether ATM issuance continues at the 2024 pace. The fourth is the cumulative effect of repeated equity-plan reserve increases (claims #14758, #14555, #6539).

Marathon's record divides into two financing regimes, and the transition between them is the most informative feature of the history. Through 2019 and most of 2020 the company raised capital in small, repeated increments. ATM capacity was sized at roughly $7.47M (claims #15203, #14915), alongside a $2.3M all-stock placement (#15151) and a convertible note whose conversion price was cut by amendment (#14913). That is the cadence of a company patching runway, and the stockholder approval to issue above 19.99% (#15048) fits that reading. The $100M S-3 (#14874) and the $200M ATM (#14651) mark the shift to programmatic equity issuance. Dilution then moved into convertible debt: the $650M 2026 notes plus the $97.5M option (#14387, #14388), followed by renewed ATM use of $161M (#13818) and $489.3M in Q1 2024 alone (#6207).

The debt side reads as less settled than the equity side. The Silvergate revolver (#14440) needed a waiver to permit the 2021 convertible (#14399), was expanded with a $100M term loan in August 2022 (#13836), and was terminated in March 2023 with notice and prepayment terms waived (#13180, #13181). The $200M line of credit in October 2024 (#5574) is the only subsequent bank facility on record. On balance the mix looks coherent in direction, with equity and converts as the core tools, but bank debt has been revised repeatedly.

Deployment has not always matched the stated plan. The $67M Compute North bridge loan (#14282) later appears as an unsecured bankruptcy claim (#12985). Abu Dhabi figures appear both as $406M (#13085) and $81.2M (#12974); this likely reflects project total versus Marathon's share, but that requires explanation. The late-2024 convertibles state two uses, bitcoin purchases (#5401, #5311) and repurchase of 2026 notes. The repurchase figures range from about $199M–$212M (#5425, #5400) down to roughly $48M–$51M (#5345, #5310, #5349), so the amount actually retired is not resolved by this record. Recorded issuance totals also conflict, at $850M (#5422), $1B (#5398) and $700M (#5348).

No insider sales are on record; insider activity is limited to equity-plan increases and RSU grants (#14758, #14555, #13964). The data ends in December 2024. Sophisticated readers should check the remaining principal of the 2026 notes against their December 1, 2026 maturity (#14387), track whether ATM issuance continued after Q1 2024, and look for reconciliation of the 2024 convertible totals and of the $9.01M reverse termination fee (#6404), which is not otherwise discussed.

Marathon's record divides into two distinct financing regimes, and reading them as one continuous strategy would overstate its coherence. Through 2020 the company relied on small, frequent equity raises: shelf and ATM capacity of roughly $7.5M (claims 15228, 14915), a $6.9M underwritten offering (claim 14850), conversion-price amendments to a roughly $1M convertible note (claims 14913, 14892), and stock issued directly to vendors and facility operators (claims 14796, 14733). That pattern reads as patching against near-term needs rather than a capital plan. In 2021 the scale changed abruptly. A $200M ATM (claim 14651), a $100M Silvergate revolver (claim 14440), and $747.5M of 2026 convertibles (claim 14279) arrived within roughly seven months, and the mix shifted from equity toward convertible debt layered with bank credit. The bank leg did not last. The Silvergate facilities, expanded with a $100M term loan in August 2022 (claim 13836), were terminated in March 2023 with notice and prepayment terms waived (claims 13180, 13181), which suggests the exit was negotiated. Equity then resumed as the primary lever, with $489.3M raised through the ATM in Q1 2024 alone (claim 6207), before the company returned to converts in late 2024. On use of proceeds, the late-2024 convertible disclosures name bitcoin purchases and repurchase of the 2026 notes (claims 5401, 5400). The repurchase figure is inconsistent across filings, ranging from about $48M to $212M (claims 5345, 5349, 5400), and the issuance size appears as $700M, $850M, and $1B (claims 5348, 5342, 5398). These figures likely reflect separate tranches, but the record requires reconciliation. The $67M Compute North bridge loan (claim 14282) later surfaces only as an unsecured bankruptcy claim (claim 12985), and the Abu Dhabi contribution is stated as both $406M and $81.2M (claims 13085, 12974). The data shows no insider sales, only RSU grants and plan-share increases (claims 13964, 14555). A careful reader should watch for how the remaining 2026 notes are settled, the cumulative ATM share count against the 2024 conversion price (claim 5402), and whether actual bitcoin purchases are disclosed in amounts that track the stated proceeds.

Marathon's financing record shows a company that moved through three distinct capital regimes rather than following one plan. Through 2019 and mid-2020 it raised small amounts: a $7.47M shelf and ATM capacity, per claims #15228 and #15203; a 2,335,000-share private placement, per claim #15151; and a $6.9M underwritten offering, per claim #14850. During this period it also amended a convertible note to lower its conversion price, per claims #14913 and #14892. That pattern reads as patching to stay funded. The August 2020 $100M S-3, per claim #14874, and the $200M ATM, per claim #14651, mark the move to scaled equity. The $747.5M of 2026 convertibles, per claim #14279, then added leverage on top. Issuing those notes required a covenant waiver from Silvergate, per claims #14399 and #14309, which suggests the bank facility and the convertible were not designed together. The Silvergate revolver and term loan were later terminated with fees waived, per claims #13180 and #13181, so the record does not show a stable bank-debt strategy. In 2024 the mix shifts to heavy ATM usage, $489.3M in Q1 alone per claim #6207, and two convertible issues, $1B per claim #5398 and $850M per claim #5342. The stated purposes of these converts were refinancing and bitcoin accumulation, which is a more coherent pattern. Deployment has not always tracked intent. The $67M Compute North bridge loan and $14M in expedite fees, per claims #14282 and #14283, resolved into an unsecured bankruptcy claim, per claim #12985. Abu Dhabi figures appear as $406M and as $81.2M, per claims #13085 and #12974, and that difference requires explanation. Planned 2026-note repurchases also vary across filings, from $199M per claim #5425 to $212M per claim #5400, and later $48–51M per claims #5345 and #5310. The record contains equity grants, per claims #13964 and #13965, but no insider sales, so it says nothing about insider selling behavior. A sophisticated reader should watch how much of the 2026 notes remains outstanding, since the maturity is December 1, 2026, per claim #14387. They should also check whether the net proceeds stated in claims #5399 and #5344 can be reconciled to actual bitcoin purchases and repurchases. Finally, the record has no claims after December 2024, so current ATM usage and any newer issuance are not visible here.

Marathon's record shows three funding phases. Each phase came from a different source, and each was larger than the one before. Through 2020 the company relied on small ATM capacity and share-settled payments for equipment and facility services. The ATM program sized at roughly $7.5 million (claims #15203, #14848) and the 3,000,000-share issuance for facility services (claim #14733) are typical of a microcap paying vendors in stock. The step-up came in 2021. A $200 million ATM (claim #14651) was followed by the $650 million 2026 convertible (claim #14387) and its $97.5 million option (claim #14388), for $747.5 million in total (claim #14279). That moved the main source of dilution from immediate equity to deferred conversion. Equity issuance did not stop. The record shows $161 million of ATM stock by August 2022 (claim #13818) and $489.3 million in Q1 2024 alone (claim #6207). Those figures suggest the convertible added to the ATM rather than replacing it.

The debt layer reads more as a reaction to events than a plan. The Silvergate revolver of $100 million (claim #14440) was later paired with a $100 million term loan (claim #13836) and drawn by $35 million (claim #13817). It was then terminated in March 2023 with notice and prepayment charges waived (claims #13180, #13181). A $200 million line of credit appears in October 2024 (claim #5574) with no stated terms. The $67 million bridge loan to Compute North (claim #14282) is later followed only by a bankruptcy unsecured claim (claim #12985). The record does not reconcile how much was recovered.

Use of proceeds requires explanation. The late-2024 convertible is described at $850 million (claim #5342), $700 million (claim #5348) and $1 billion (claim #5398). The planned repurchase of 2026 notes is stated as $199 million (claim #5425), $212 million (claim #5400) and about $50 million (claims #5349, #5310). Bitcoin acquisition is the other stated use (claims #5401, #5311). No claim confirms how either use was ultimately carried out. Likewise, the Abu Dhabi contribution appears as both $406 million (claim #13085) and $81.2 million (claim #12974). That may be a total-versus-share distinction, but the record does not say so.

The record contains no insider sales, only RSU grants (claims #13964, #13965). It ends in December 2024. A careful reader should watch for the 2026 notes reaching maturity on December 1, 2026 (claim #14387) and for how much principal remains outstanding after the repurchases. They should also watch whether ATM issuance resumes to settle that balance, and whether later filings reconcile the conflicting figures for the 2024 offering.

Marathon's record shows three distinct financing regimes. Through 2020 the company relied on small, repeated at-the-market programs sized around $7.47M (claims 15203, 14915), supplemented by all-stock private placements (claim 15151) and share-settled equipment and facility arrangements (claims 14796, 14733). That cadence is consistent with a company funding itself in increments sized to its market value at the time. The shift came in 2021, with the $100M S-3 (claim 14874) giving way to a $200M ATM (claim 14651) and then the $650M 2026 convertible plus its $97.5M option (claims 14387, 14388), together $747.5M per claim 14279. Equity dilution then accelerated: $161M of ATM issuance is recorded in 2022 (claim 13818) and $489.3M in Q1 2024 alone (claim 6207). The third regime, late 2024, layered new convertibles on top of the existing ones, including the $1B offering (claim 5398) and the $850M offering (claim 5422).

The mix looks deliberate in its broad direction, with equity funding growth and convertibles adding scale at low coupons, but the bank-debt leg was not durable. The Silvergate revolver and term loan (claims 14440, 13836), drawn by $35M per claim 13817, were terminated in March 2023 with notice and prepayment waivers (claims 13180, 13181). A $200M credit line later appears without a named counterparty (claim 5574).

Use of proceeds tracks only partially. The Compute North bridge loan and expedite fees ($67M and $14M, claims 14282, 14283) are later followed by an unsecured bankruptcy claim (claim 12985), and the record does not show how much was recovered. The Abu Dhabi commitment is stated as $406M at the project level (claim 13085) against an $81.2M company contribution (claim 12974), a distinction readers should keep in view. The 2024 offerings consistently cite bitcoin purchases and repurchases of 2026 notes (claims 5401, 5311), but the stated repurchase amounts range from roughly $48M to $212M across offerings (claims 5345, 5400). That range requires explanation in subsequent filings.

The data has gaps. No insider sales are recorded, several figures carry inconsistent units, and the record contains no completed bitcoin acquisition amounts. The 2026 notes mature December 1, 2026 (claim 14387), so how the remaining balance is settled, whether in cash, shares, or refinancing, is the nearest item to watch. After that, readers should watch the holder repurchase dates on the 2024 notes (claim 5405) and whether ATM issuance resumes after Q1 2024.

Marathon's financing record moves through three distinct regimes. Through 2020, capital came in small, repeated increments: shelf and ATM capacity of roughly $7.47M (claims #15228, #14915), a 2020 conversion-price reduction on an existing note (claim #14913), and equity paid directly to vendors for equipment and facility services (claims #14796, #14733). That reads as survival financing rather than design. The 2021 step-up to a $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440) and the $747.5M 2026 convertible (claims #14387, #14388) marks a deliberate shift toward scale, with debt and convertibles layered over continuing equity issuance. The 2024 regime is larger again: $489.3M raised through the ATM in a single quarter (claim #6207), a $200M credit line (claim #5574), and two convertible deals described as $1B and $850M (claims #5398, #5342). Dilution therefore did not plateau. It shifted from frequent common issuance to a heavier reliance on convertibles whose dilution is deferred.

The bank-debt leg did not hold. The Silvergate revolver was expanded with a $100M term loan in August 2022 (claim #13836), with $35M drawn (claim #13817), and was terminated in March 2023 with notice and prepayment terms waived (claims #13180, #13181). Silvergate was not replaced by comparable bank financing until the 2024 credit line. The $67M bridge loan to Compute North (claim #14282) later appears as a bankruptcy unsecured claim (claim #12985), and the reported amounts do not reconcile on their face.

On use of proceeds, the 2024 convertibles were tied to bitcoin purchases and to repurchasing the 2026 notes (claims #5401, #5311). The stated repurchase figures vary across filings, from $199M and $212M (claims #5425, #5400) to roughly $48–51M (claims #5345, #5310, #5349). That variation requires explanation, and the record does not show actual deployment. The Abu Dhabi commitment is also stated two ways, as $406M (claim #13085) and as an $81.2M contribution (claim #12974). The $9.0M reverse termination fee (claim #6404) has no surrounding context.

The insider data contains only RSU grants (claims #13964, #13965) and no recorded sales, so no inference about insider selling is possible.

The immediate item to watch is the residual 2026 notes maturing December 1, 2026 (claim #14387) and how much remains after the 2024 repurchases. After that come the holder repurchase and redemption dates on the 2024 notes (claims #5404, #5405) and the conversion price of about $26 (claim #5402) relative to the share price.

Marathon's record divides into two financing regimes. Through 2020 the company funded itself in small, repeated increments: at-the-market capacity of roughly $7.5 million appears in several filings (claims #15203, #14915), alongside an all-stock private placement (claim #15151), a convertible note amendment that lowered the conversion price (claim #14913), and equity paid to vendors for hosting and construction milestones (claims #14796 through #14799). That is patching rather than strategy, and the repricing amendment shows a lender being accommodated on the company's side. The pivot comes with the $100 million S-3 (claim #14874) and the $200 million ATM (claim #14651), after which the instruments scale by an order of magnitude. The $747.5 million of 2026 convertibles (claim #14279), the Silvergate revolver (claim #14440) and the later term loan (claim #13836) show a deliberate move toward low-coupon debt and secured credit layered over continued ATM issuance, $161 million of which is recorded by mid-2022 (claim #13818). The termination of all Silvergate facilities in March 2023 (claim #13180) ends the bank-credit leg. The $489.3 million ATM tally for Q1 2024 alone (claim #6207) shows equity issuance accelerating rather than moderating. The November and December 2024 convertibles, $1 billion (claim #5398) and $850 million (claim #5342), form a coherent refinancing-plus-accumulation program. Stated uses pair repurchases of the 2026 notes with bitcoin purchases (claims #5401, #5311). Deployment is harder to confirm. The repurchase figures are revised across filings, $199 million versus $212 million for the first deal (claims #5425, #5400) and $48 million to $51 million for the second (claims #5345, #5310), and a separate $700 million figure (claim #5348) does not reconcile with either offering. Elsewhere, the $67 million Compute North bridge loan (claim #14282) later surfaces as an unsecured bankruptcy claim (claim #12985), and the Abu Dhabi contribution appears as both $81.2 million and $406 million (claims #12974, #13085). Both require explanation. Insider activity on record is limited to RSU grants (claims #13964, #13965), with no disclosed sales. A careful reader should watch for the residual 2026 note balance after repurchases, which the record does not state, conversion overhang near the roughly $26 price on the 2024 notes (claim #5402), and whether ATM issuance continues alongside the convertibles. The record also ends in December 2024, so anything since then is unobserved here.

Marathon's record shows a company that moved from small, frequent equity patches to large convertible financings, with convertibles becoming the main funding tool by the end of the period. Between 2019 and mid-2020 the company relied on at-the-market programs sized near $7.5M (claims #15203, #14915, #14848), equity issued for mining equipment (claim #15066), and a reduced conversion price on an existing note (claims #14913, #14892). That period reads as opportunistic patching. Scale changed in 2021 with a $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440), and $747.5M of 1% notes due 2026 (claims #14279, #14386). Equity issuance then continued alongside the debt rather than replacing it, with $161M of ATM issuance by August 2022 (claim #13818) and $489.3M in Q1 2024 alone (claim #6207). Dilution therefore did not slow as the capital base grew; it compounded. Deployment has not always matched the stated purpose. The $67M bridge loan and $14M of expedite fees to Compute North (claims #14282, #14283) were later followed by an unsecured bankruptcy claim (claim #12985). The Silvergate term and revolving facilities, which carried a $350M net-worth covenant (claim #13839), were terminated with notice and prepayment waivers (claims #13180, #13181, #13182). After that, they are no longer discussed as a funding source. The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085), which requires explanation. The December 2024 offering is likewise recorded as $700M in one claim and $850M in others (claims #5348, #5307). In late 2024 the company issued $1B and $850M of new convertibles (claims #5398, #5307). The stated uses were buying bitcoin and repurchasing about $212M and $51M of the 2026 notes (claims #5400, #5310). The record does not show whether those repurchases were completed. Insider activity on record is limited to RSU grants (claims #13964, #13965), and no insider sales are disclosed. A careful reader should watch three things. First, how the remaining 2026 notes are settled, given that the record shows only partial repurchase plans against the original $747.5M. Second, whether holders exercise the put rights on the 2024 notes (claim #5405). Third, whether ATM issuance continues to fund the gap between refinancing and bitcoin accumulation. The record ends in December 2024, so later events cannot be assessed from it.

Marathon's financing record moves through three distinct regimes. Through 2020 it relied on small, frequent equity raises: an at-the-market capacity of roughly $7.5M (claim 15203), a $6.9M underwritten offering (claim 14850), and share-settled equipment and facility deals (claims 14796, 14733). In 2021 the scale changed, with a $200M ATM program (claim 14651) followed by $747.5M of 2026 convertible notes (claim 14279). By 2024 the company was raising on a much larger base, with $489.3M sold through the ATM in a single quarter (claim 6207) and several convertible offerings in quick succession, including the completed $1B issue (claim 5398) and the $850M issue (claim 5307). Dilution therefore did not ease over time. It moved from frequent small issuance toward large deferred issuance through conversion features, and three equity plan share increases (claims 14758, 14555, 6539) add a steady compensation layer on top.

The mix looks more opportunistic than planned. The Silvergate revolver and term loan (claims 14440, 13836) came with liquidity and net-worth covenants of $25M and $350M (claims 13840, 13839), needed a waiver within weeks of the 2026 notes (claim 14399), and were terminated in March 2023 (claim 13180). The $67M bridge loan and $14M of expedite fees to Compute North (claims 14282, 14283) ended in an unsecured bankruptcy claim (claim 12985). Neither of these use-of-proceeds paths appears to have performed as originally presented.

For the 2024 convertibles, the stated uses were bitcoin purchases and repurchases of the 2026 notes. The planned repurchase amounts vary across the record: $199M (claim 5425), $212M (claim 5400), and roughly $48M to $51M (claims 5345, 5349, 5310). No claim confirms the amounts actually retired. The Abu Dhabi contribution is recorded as both $81.2M (claim 12974) and $406M (claim 13085), and a $700M December offering (claim 5348) sits beside the $850M figure. These discrepancies require explanation. No discretionary insider sales appear in the record, only RSU grants (claims 13964, 13965).

Going forward, the key date is the December 1, 2026 maturity of the remaining 2026 notes (claim 14387), which is now close. Readers should watch whether that balance is settled in cash, shares, or new paper. They should also track the holder repurchase dates on the 2024 notes (claim 5405), whether the $200M credit line (claim 5574) is drawn, and whether ATM activity resumes at the scale seen in early 2024.

Marathon's financing record falls into three distinct phases, and the shift between them is more telling than any single raise. Through 2019 and 2020 the company funded itself in small, overlapping increments: repeated ATM capacity of roughly $7.5M (claims 15203, 14915, 14848), a $6.9M underwritten offering (claim 14850), shares issued directly for mining equipment (claim 15066), and an amendment reducing a convertible note's conversion price (claim 14913). That pattern reads as opportunistic patching rather than a capital plan. The $200M ATM (claim 14651) and the $747.5M of 2026 convertibles (claim 14279) mark a move to scale, and the 2024 sequence, a $489.3M first-quarter ATM (claim 6207), a $1B convertible (claim 5398) and an $850M convertible (claim 5307), shows the company now relying on equity-linked paper at institutional size. Share-count dilution cannot be traced from this record. The only outstanding-share figure is 8,720,405 from 2019 (claim 15152), alongside three incentive-plan reserve increases (claims 14758, 14555, 6539). The bank-debt leg did not hold. The Silvergate revolver and term loan (claims 14440, 13836), which carried a $350M adjusted net worth covenant (claim 13839), were terminated in March 2023 with notice and prepayment terms waived (claims 13180–13182). The $67M bridge loan and $14M expedite fees paid to Compute North (claims 14282, 14283) later appear only as an unsecured bankruptcy claim (claim 12985), and that deployment requires explanation. The Abu Dhabi joint venture contribution is recorded as both $81.2M and $406M (claims 12974, 13085), which also requires reconciliation. On use of proceeds, the 2024 convertibles were tied to bitcoin purchases and to repurchasing 2026 notes. The stated repurchase amounts are roughly consistent across filings: $199M versus $212M (claims 5425, 5400) and $48M–$51M (claims 5345, 5349, 5310). However, the record contains no confirmation of actual repurchases or bitcoin deployed. One claim also lists the December offering at $700M (claim 5348) against $850M elsewhere. No insider sales appear, only RSU grants (claims 13964, 13965). A sophisticated reader should watch for the residual balance of the 2026 notes due December 1, 2026 (claim 14387), whether the $200M credit line (claim 5574) has been drawn, and why the disclosed record stops in December 2024.

Marathon's record describes a company that moved from micro-cap patching to large-scale, convertible-led funding in roughly three years. Before mid-2020 the instruments were small and improvised: an ATM sized at $7,472,417 (claims #15203, #14915), an all-stock placement of 2,335,000 shares (claim #15151), stock issued for equipment and services, and an amendment cutting a convertible note's conversion price (claim #14913). These are consistent with a company funding itself in whatever form was available. The step to a $100,000,000 S-3 (claim #14874) and then a $200,000,000 ATM (claim #14651) marks a change in scale. The 2021 Silvergate revolver of $100,000,000 (claim #14440) and the $747,500,000 in 2026 convertibles (claim #14279) give the mix more structure. Even so, the converts required a covenant waiver from the lender almost immediately (claim #14399), and the Silvergate facilities were terminated in March 2023 with fees waived (claims #13180, #13181). On the debt side, the structure looks more reactive than planned. The record contains no insider sales, only RSU grants (claims #13964, #13965) and repeated plan-share increases (claims #14758, #14555, #6539), so the main source of dilution is corporate issuance, not insider selling.

Stated uses of proceeds and actual deployment diverge in places. The $67,000,000 bridge loan and $14,000,000 in expedite fees to Compute North (claims #14282, #14283) were later followed by an unsecured bankruptcy claim (claim #12985), and that outcome is not reconciled in the record. The Abu Dhabi joint venture appears as an $81,200,000 contribution (claim #12974) and separately as $406,000,000 (claim #13085). That gap requires explanation. The 2024 convertibles shift the stated purpose from infrastructure toward bitcoin acquisition (claims #5401, #5311), alongside buybacks of 2026 notes.

The late-2024 figures also need reconciling. One offering is cited at $850 (read as millions; claim #5422), completed at $1,000 with $980 net (claims #5398, #5399), while a separate $850 deal shows $835 net (claim #5344) and a $700,000,000 figure also appears (claim #5348). Planned 2026-note repurchases are cited as $199 and $212 in one deal (claims #5425, #5400) and around $48–51 in the other (claims #5345, #5310, #5349). Read together with the $489,300,000 Q1 2024 ATM (claim #6207), the balance has tilted decisively toward equity-linked funding.

A sophisticated reader should establish how much of the 2026 notes remains outstanding as the December 1, 2026 maturity (claim #14387) approaches. They should also confirm the conversion terms near $26 (claim #5402) and track whether purchased bitcoin is held or sold to meet obligations. The record ends in December 2024, so anything after that date is not covered here.

Marathon's financing record moves through three distinct regimes. Through 2020 the company relied on small, repeated equity raises sized to a shelf of roughly $7.5M (claims 15228, 14915), a 2019 stockholder approval to issue above 19.99% (claim 15048), and shares paid to counterparties for equipment and facility services (claims 15066, 14796). That pattern reads as patching rather than strategy. From 2021 the scale changed: a $200M ATM (claim 14651), $747.5M of 2026 convertibles including the $97.5M option (claim 14279), and a $100M Silvergate revolver (claim 14440). ATM issuance followed, at $161M by August 2022 (claim 13818) and $489.3M in Q1 2024 alone (claim 6207). Late 2024 added a $1B convertible (claim 5398) and an $850M convertible (claim 9275) within about two weeks, with a $200M credit line shortly before (claim 5574). Dilution has shifted from frequent small issuance to large ATM draws plus a convertible overhang that defers dilution rather than avoiding it.

After 2021 the mix looks more coherent, with convertibles partly refinancing earlier convertibles and the ATM funding operations. The bank leg did not hold. The Silvergate facilities were terminated, with the bank waiving its notice period and prepayment charge (claims 13180-13182). That exit was forced rather than planned.

Use of proceeds is only partly traceable. The late-2024 notes cited bitcoin purchases and repurchases of the 2026 notes (claims 5401, 5311). The stated repurchase amounts vary: $199M and $212M for the November deal (claims 5425, 5400), roughly $48-51M for December (claims 5345, 5310), and a separate $50M entry (claim 5349). The record does not confirm completed amounts. An earlier commitment, the $67M bridge loan to Compute North (claim 14282), ended in an unsecured bankruptcy claim (claim 12985). The Abu Dhabi contributions appear in inconsistent units (claims 13085, 12953). A $700M convertible entry (claim 5348) does not reconcile with the other offerings and requires explanation.

On the insider side, the record shows equity incentive plan increases (claims 14758, 14555, 6539) and executive RSU grants (claims 13964, 13965), but no insider sales. A sophisticated reader should watch for four things: confirmation of completed 2026 note repurchases; the conversion terms on the 2024 notes (claim 5402); the ATM pace relative to Q1 2024; and whether bitcoin-purchase disclosures reconcile with stated net proceeds of $980M and $835M (claims 5399, 5308).

Marathon's record divides into two regimes. Through 2020 the company financed itself in small increments: ATM capacity of roughly $7.5M was put in place more than once (claims #15203, #14915, #14848), alongside a $6.9M underwritten offering (claim #14850), an all-stock private placement of 2,335,000 shares (claim #15151), and share-settled equipment purchases (claim #15066). That is the cadence of a company covering near-term needs rather than executing a plan. From 2021 the scale changes. A $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440), and $747.5M of 2026 convertibles (claim #14279) arrived within roughly six months. Further ATM issuance followed, $161M per claim #14818 and $489.3M in Q1 2024 per claim #6207, and then a $1B convertible in November 2024 (claim #5398) and an $850M convertible in December (claim #5342). Dilution now arrives through both direct issuance and embedded conversion optionality, and its absolute size has grown with each cycle.

The mix looks more deliberate after 2021, but its shape was partly forced. Bank debt was a stated pillar, including a revolver and a $100M term loan with covenants (claims #13835, #13836, #13839). It ended with the Silvergate termination in March 2023 (claim #13180), and a $200M line of credit appeared only in October 2024 (claim #5574). In the interval the company relied on ATM equity and a $14.25M preferred placement (claim #12486). The record shows no insider sales. Insider activity is limited to RSU grants (claims #13964, #13965) and repeated expansions of the incentive plan (claims #14758, #14555, #6539).

Use-of-proceeds tracking is incomplete. The 2024 convertibles were earmarked for bitcoin purchases and for repurchasing 2026 notes, roughly $212M and $51M per claims #5400 and #5310. Only a $50M repurchase is recorded as executed (claim #5349). The $67M Compute North bridge loan (claim #14282) is later reflected only as an unsecured bankruptcy claim (claim #12985), and that outcome requires explanation. The Abu Dhabi contribution is recorded inconsistently, as $406M, $406,000, and $81.2M across claims #13085, #12953, and #12974. Several other amounts in the record also show apparent unit-scaling errors.

The next thing to watch is the December 1, 2026 maturity of the original convertibles (claim #14387). The record does not establish how much principal remains outstanding after the announced repurchases. It also does not show whether the converts-refinancing-converts pattern will continue or give way to equity.

Marathon's record shows dilution that grew by orders of magnitude rather than steadily. Equity programs sized near $7.5M in 2019 (claim 15203) gave way to a $200M ATM in 2021 (claim 14651), $161M of ATM issuance reported by August 2022 (claim 13818), and $489.3M raised through the ATM in Q1 2024 alone (claim 6207). Running alongside this are repeated equity incentive plan increases (claims 14758, 14555, 6539). Taken together, the company leaned on its own stock as the default funding source, and each new step was larger than the last.

The financing mix is more opportunistic than planned. The 2021 cycle stacked a $100M Silvergate revolver (claim 14440), a $747.5M convertible package (claim 14279), and a term loan added in 2022 (claim 13836). Within roughly four months of that term loan, a covenant waiver (claim 14309) and an amendment (claim 14399) were needed. The Silvergate facilities were then terminated in March 2023 with notice and prepayment charges waived (claims 13180, 13181). A small preferred placement (claim 12486) and a $200M credit line in 2024 (claim 5574) look like gap-fillers rather than parts of a design. The 2024 convertibles do show a deliberate pivot: proceeds were used to refinance the 2026 notes and to buy bitcoin (claims 5400, 5401, 5311).

The link between announced use of proceeds and actual deployment is weakest around Compute North. A $67M bridge loan and $14M in expedite fees (claims 14282, 14283) are later followed only by an unsecured bankruptcy claim (claim 12985), and the recovery amount is not discussed. The Abu Dhabi contribution appears as both $406M and $81.2M (claims 13085, 12974), which is consistent with a project-level total versus Marathon's share, but the record does not reconcile the two. The 2024 offering figures also require explanation. The same deal is recorded at $850 (claim 5422), $700,000,000 (claim 5348), and as a completed $1 billion raise (claim 5398), and the planned 2026-note repurchases range from $48 to $212 (claims 5345, 5400).

The record ends in December 2024, so nearly two years of activity is absent. A careful reader should watch for the outstanding balance of the 2026 notes at their December 1, 2026 maturity (claim 14387), and for whether bitcoin bought with convertible proceeds is held or sold. Draws on the $200M line deserve attention, as does whether ATM issuance continues to outpace the share counts the company reports. The record shows no insider sales, only grants, so that question cannot be assessed here.

Marathon's financing record traces a company that moved from small-scale patching to large, structured capital raising, though the sequence was shaped by events as often as by plan. Through 2019 and 2020 the company raised in single-digit millions, mainly through H.C. Wainwright at-the-market programs of roughly $7.5M (claims 15203, 14848), an all-stock private placement (claim 15151), and repeated repricing of a small convertible note (claims 14913, 14892). The step-change came in 2021. A $200M ATM (claim 14651), a $100M Silvergate revolver (claim 14440) and $747.5M of 2026 convertible notes (claim 14279) arrived within about seven months. Dilution then accelerated again: $161M of ATM issuance is recorded by mid-2022 (claim 13818) and $489.3M in Q1 2024 alone (claim 6207). Alongside this, the equity incentive plan reserve was enlarged three times (claims 14758, 14555, 6539).

The mix reads as coherent in intent, pairing low-coupon converts with ATM equity and bank liquidity, but less coherent in execution. The Silvergate relationship was expanded with a $100M term loan in August 2022 (claim 13836), then terminated in March 2023 with prepayment and notice requirements waived (claims 13180, 13181, 13182). That ended the debt leg rather than refinancing it. The $14.25M Series A preferred placement in mid-2023 (claim 12486) looks like a gap-filler between larger instruments.

Use of proceeds tracks only partially. The $67M bridge loan and $14M expedite fees to Compute North (claims 14282, 14283) were later followed by an unsecured bankruptcy claim (claim 12985), and the recorded amount of that claim requires explanation. The Abu Dhabi contribution appears as both $81.2M and $406M (claims 12974, 13085). The 2024 convertible offerings stated two purposes, acquiring bitcoin and repurchasing 2026 notes (claims 5401, 5311). The stated repurchase amounts were revised across filings: $199M, then $212M for the November deal (claims 5425, 5400), and $48M to $51M for December (claims 5345, 5310), with $50M recorded as executed (claim 5349). The record does not show the bitcoin deployment. No insider sales appear on record, only RSU grants (claims 13964, 13965).

The nearest item to watch is the December 1, 2026 maturity of the original converts (claim 14387), now weeks away. The record does not establish how much principal remains after repurchases, and that requires explanation. Readers should also note that the record contains no financing events after December 2024.

Marathon's record describes two companies. Through 2020 it was a micro-cap financing itself in small increments: an ATM sized at roughly $7.47M (claim 15203, restated in claim 14915), an all-stock placement of 2,335,000 shares that left 8,720,405 shares outstanding (claims 15151, 15152), stockholder approval to exceed the 19.99% issuance threshold (claim 15048), and a convertible note amendment that lowered the conversion price (claim 14913). That sequence reads as opportunistic patching rather than strategy. The shift came with the $100M S-3 (claim 14874) and the $200M ATM (claim 14651), after which the company moved to large instruments: $747.5M of 2026 convertibles including the exercised option (claims 14387, 14369, 14279), Silvergate revolving and term facilities (claims 14440, 13836), $161M of further ATM issuance by August 2022 (claim 13818), and $489.3M of ATM sales in Q1 2024 alone (claim 6207). Dilution was continuous throughout, but the instrument mix grew more deliberate over time, with converts used for large raises and the ATM used to cover the periods between them.

The link between stated and actual use of proceeds is uneven. The $67M bridge loan and $14M of expedite fees paid to Compute North (claims 14282, 14283) were later followed by an unsecured claim in Compute North's bankruptcy (claim 12985), and the record does not show what was recovered. The Silvergate facilities, whose covenants included a $350M minimum adjusted net worth (claim 13839), were terminated early with notice and prepayment waivers (claims 13180, 13181). The Abu Dhabi contribution appears as both $81.2M and $406M (claims 12974, 13085), and those figures require explanation. The late-2024 convertibles moved the stated use of proceeds from infrastructure toward buying bitcoin and repurchasing 2026 notes (claims 5401, 5311). The planned repurchase amounts were revised several times, from $199M to $212M for the first deal (claims 5425, 5400) and from $48M to $51M for the second (claims 5345, 5310). The record states these intentions but does not document that the purchases were carried out.

The data is incomplete. The final claim is truncated, nothing is recorded after December 2024, and no insider transactions appear at all, so no view on insider selling can be drawn. A careful reader should first establish how much of the 2026 notes remains after the stated repurchases, because that maturity falls in December 2026 (claim 14387). They should also check whether the bitcoin purchases funded by the 2024 offerings are reported as completed, whether the conversion terms of the new notes (claim 5402) and the holder put date (claim 5405) create refinancing pressure, and whether ATM usage has continued at the pace recorded in 2024.

Marathon's financing record falls into three distinct phases, and the scale changes sharply between them. Through 2019 and mid-2020 the company relied on small, repeated equity measures. These included a shelf of roughly $7.47M (claim #15228), a 2,335,000-share private placement (claim #15151), stockholder approval to issue above the 19.99% threshold (claim #15048), and two amendments lowering the conversion price on the Bi-Coastal note (claims #14913, #14892). Repricing an existing instrument to keep it serviceable reads as patching rather than a plan. The $100M S-3 (claim #14874) and the $200M ATM (claim #14651) moved the company to an equity-funded growth posture. The $650M convertible due 2026 (claim #14387), with its $97.5M option exercised (claim #14369), then made convertible debt the primary instrument. That pattern held through the November 2024 $1B offering (claim #5398) and the December $850M offering (claim #5307). Over the same period, the equity incentive plan reserve was raised repeatedly (claims #14758, #14555, #6539).

Bank debt has a shorter and less settled history. The Silvergate revolver (claim #14440) was amended with a waiver within about five weeks to accommodate the convertible (claim #14399). It was expanded with a term loan (claim #13836) and then terminated in March 2023 with prepayment and notice waivers (claims #13180, #13181, #13182). The $200M line in October 2024 (claim #5574) is the only subsequent bank facility on record.

Stated uses of proceeds can only partly be traced to deployment. Equipment purchases are documented (claims #14843, #14795). The $67M bridge loan to Compute North (claim #14282) ends in an unsecured bankruptcy claim (claim #12985), whose recorded figure requires explanation. The Abu Dhabi contribution appears as both $81.2M (claim #12974) and $406M (claim #13085), and the record does not reconcile the two. For the 2024 notes, the planned repurchase of 2026 notes was revised from $199M (claim #5425) to $212M (claim #5400). Bitcoin acquisition is stated as intent (claims #5401, #5311), but no claim confirms the purchases. The record also lists a $700M figure (claim #5348) against the $850M completed offering, which requires explanation.

The record contains no insider transactions, and it truncates in December 2024. A careful reader should therefore look for the residual 2026 principal after repurchases, since those notes mature December 1, 2026 (claim #14387). They should also look for whether the $26 conversion terms (claim #5402) come into play, how the Abu Dhabi figures reconcile, and any post-2024 ATM activity on the scale of the $489.3M Q1 2024 issuance (claim #6207).

The record shows Marathon moving from micro-cap survival financing to large-scale convertible issuance, with equity sales running alongside throughout. The early years are small and reactive. Shelf and ATM capacity sat near $7.47M (claims #15228, #15203, #14915), a convertible note was amended to lower its conversion price (claim #14913), and stockholders approved issuance above the 19.99% threshold (claim #15048). These read as patches rather than a plan. The scale changes in 2021. A $200M ATM (claim #14651) was followed by $650M of 2026 convertibles plus a $97.5M option exercise (claims #14387, #14369), and a $100M Silvergate revolver (claim #14440) added bank leverage. By 2024 the pattern looks deliberate. Equity was issued through the ATM, $489.3M in Q1 2024 alone (claim #6207), while successive convertible deals refinanced the 2026 notes: a $1B offering with $212M earmarked for repurchase (claims #5398, #5400), then $850M with roughly $51M directed the same way (claims #5307, #5310). The stated use of proceeds also shifted, from infrastructure toward acquiring bitcoin (claims #5401, #5311). Deployment does not always track announcements. The $67M bridge loan and $14M of expedite fees paid to Compute North (claims #14282, #14283) were later followed by an unsecured claim in that hosting provider's bankruptcy (claim #12985), and that outcome requires explanation relative to the original rationale. The Silvergate facilities were expanded with a $100M term loan (claim #13836) and drawn by $35M (claim #13817), then terminated within months with fees waived (claims #13180, #13181). The December 2024 deal is recorded as both $700M and $850M (claims #5348, #5307), which should be reconciled. Several figures appear in inconsistent units across filings. The record holds no insider sale data, only RSU grants (claims #13964, #13965), so insider behavior cannot be assessed, and the data ends in December 2024. A careful reader should watch four things: how much of the 2026 notes remains after the partial repurchases, the holder put and earliest redemption dates on the new notes (claims #5404, #5405), whether ATM issuance continues at the 2024 pace, and whether the Abu Dhabi commitment, recorded as $406M at the facility level and $81.2M in contributions (claims #13085, #12974), shows up in later deployment disclosures.

Marathon's record divides into two regimes. Through 2020 the company funded itself in small, reactive increments: ATM capacity of roughly $7.5M (claims 15203, 14915), a private placement (claim 15151), a stockholder vote to exceed the 19.99% issuance cap (claim 15048), and a convertible note amendment that lowered the conversion price (claim 14913). That is patching rather than strategy, and lowering a conversion price on existing paper is the kind of concession that tends to accompany limited negotiating leverage. The August 2020 S-3 for $100M (claim 14874) marks the transition. From 2021 onward the instruments scale sharply and become more deliberate: a $200M ATM (claim 14651), $747.5M of 2026 convertibles including the exercised option (claims 14279, 14369), and Silvergate revolving and term facilities (claims 14440, 13836) carrying a $25M liquidity covenant and a $350M adjusted net worth covenant (claims 14444, 13839). Those facilities were terminated in March 2023 with notice and prepayment charges waived (claims 13180, 13181), so the debt layer was unwound rather than rolled. Equity dilution continued at scale through the ATM, with $161M issued by mid-2022 (claim 13818) and $489.3M in Q1 2024 alone (claim 6207). On deployment, the 2021 period requires explanation. A $67M bridge loan and $14M in expedite fees to Compute North (claims 14282, 14283) were followed by an unsecured claim in that hosting provider's bankruptcy (claim 12985), and the record does not show what was recovered. The Abu Dhabi venture appears as both an $81.2M contribution and a $406M facility figure (claims 12974, 13085), a gap the record leaves unreconciled. The November and December 2024 convertibles, $1B and $850M (claims 5398, 5307), state two uses: repurchasing 2026 notes ($212M and roughly $51M per claims 5400, 5310) and acquiring bitcoin (claims 5401, 5311). Part of the proceeds therefore extends maturities, and part funds a balance-sheet asset rather than operations. Several principal figures remain not_yet_verifiable (claims 5342, 5422), and claim 5348's $700M figure does not match the $850M elsewhere. The record contains no insider sales, and it is truncated after December 2024. A reader should next watch how much of the 2026 notes remain outstanding, the holder repurchase dates on the new paper (claim 5405), whether bitcoin purchases are disclosed at amounts matching the stated use of proceeds, and whether the ATM continues to run alongside convertible issuance.

Marathon's disclosed financing history falls into three regimes. Through 2020 the company funded itself in small increments: an ATM sized at roughly $7.5M (claims #15203, #14915), a 2,335,000-share all-stock placement (claim #15151), a reduced conversion price on the Bi-Coastal note (claims #14913, #14892), and share-settled equipment and services arrangements (claims #15066, #14796). That pattern reads as opportunistic patching rather than a capital plan. The $100M S-3 (claim #14874) and the $200M ATM (claim #14651), reported completed by January 2021 per claim #14296, mark the shift to scale. They were followed by the $650M 1.00% convertible due 2026 and its $97.5M option (claims #14387, #14369), and the $100M Silvergate revolver (claim #14440). The 2022 term loan layer (claim #13836) carried a $350M adjusted net worth covenant (claim #13839). Those facilities were terminated in March 2023 with the notice and prepayment charges waived (claims #13180, #13181), a change driven by the counterparty rather than by plan.

Equity issuance stayed constant in the background and grew larger over time. The $161M of ATM issuance cited in August 2022 (claim #13818) was exceeded by $489.3M in Q1 2024 alone (claim #6207), so dilution accelerated as the company scaled. Equity plan reserves were also raised repeatedly (claims #14758, #14555, #6539). The record contains no insider sale disclosures, so that dimension cannot be assessed.

The late-2024 convertibles move the structure toward debt plus conversion optionality. The $1B offering (claim #5398) and the $850M offering (claim #5307) both cite repurchases of the 2026 notes, but those allocations are modest at roughly $212M and $51M (claims #5400, #5310). The bulk of proceeds was designated for bitcoin purchases (claims #5401, #5311), a balance-sheet strategy rather than operating capex.

Earlier stated uses of proceeds match actual outcomes less cleanly. The $67M Compute North bridge loan and the $14M expedite fees (claims #14282, #14283) were followed by an unsecured bankruptcy claim (claim #12985). The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085), a discrepancy that requires explanation. Several figures in the record are reported in inconsistent units, and the dataset truncates in December 2024.

Going forward, readers should watch four things: whether the remaining 2026 notes are retired before maturity, how the 2024 notes' conversion terms (claim #5402) compare with the share price, whether ATM issuance continues alongside convertible issuance, and whether the $200M credit line (claim #5574) is drawn.

The record traces a company whose financing scale grew by several orders of magnitude, from a $7.47M shelf and ATM capacity in 2019 (claims #15228, #15203) to a $200M ATM in 2021 (claim #14651) and an at-the-market raise of $489.3M in a single quarter of 2024 (claim #6207). Dilution was continuous rather than episodic. ATM programs recur in nearly every year on record, supplemented by equity incentive plan increases in 2020, 2021 and 2023 (claims #14758, #14555, #6539) and by share-settled equipment and facility payments (claims #15066, #14733).

The mix shifted coherently in one direction: from small equity and ATM raises toward large convertible issuance. The 1.00% notes due 2026 totaled $747.5M (claim #14279). The Silvergate relationship was less durable. A $100M revolver (claim #14440) was followed by a $100M term facility (claim #13836), and both were terminated in March 2023 with prepayment and notice waivers (claims #13194, #13181). The facilities therefore ran less than eighteen months. The late-2024 cycle used new convertibles partly to retire the 2026 notes, which reads as liability management rather than patching. The volume and overlap of the late-2024 figures, however, $1B (claim #5398), $850M (claim #5307) and $700M (claim #5348), make it hard to reconcile from the record alone how many distinct tranches were issued.

Announced uses of proceeds track actual deployment only partially. The planned repurchase of 2026 notes was stated at $199M (claim #5425) and $212M (claim #5400), then at $48M–$51M (claims #5345, #5310), with $50M executed (claim #5349). That sequence is a revision, not a contradiction, but it requires explanation. Bitcoin acquisition was stated as a use (claims #5401, #5311), and the record contains no corresponding deployment figure. The $67M bridge loan to Compute North (claim #14282) later appears alongside an unsecured bankruptcy claim (claim #12985). The Abu Dhabi contribution is recorded as $81.2M (claim #12974) and separately as $406M (claim #13085), and these figures do not reconcile.

No insider sales appear in the record, so that dimension cannot be assessed. A reader should watch for three things. The first is the holder repurchase date and earliest redemption date on the 2024 notes (claims #5404, #5405). The second is whether residual 2026 notes are retired or refinanced again. The third is whether ATM usage continues at 2024 scale. The record provided ends mid-entry, so later events may not be reflected here.

Marathon's financing record moves through three distinct phases, and the shift between them matters more than any single transaction. Through 2019 and into 2020, the company funded itself in small increments: at-the-market capacity of roughly $7.5 million (claims 15203 and 14915), an all-stock private placement (claim 15151), and shares issued directly for mining equipment (claim 15066). It also cut the conversion price on a consultant's note (claim 14913). That pattern reflects a company managing liquidity rather than executing a capital plan. The $100 million S-3 in August 2020 (claim 14874) and the $200 million ATM in 2021 (claim 14651) mark the move to scale, and the $650 million convertible plus the $97.5 million option (claims 14387 and 14369) shifted the funding base from equity to convertible debt.

The bank-debt chapter is the least durable part of the record. The Silvergate revolver (claim 14440) was expanded with a term loan (claim 13836) and drawn for $35 million (claim 13817), then terminated in March 2023 with notice and prepayment charges waived (claims 13180 through 13182). The $67 million bridge loan to Compute North (claim 14282) is later followed only by an unsecured bankruptcy claim (claim 12985), and its ultimate recovery is no longer discussed in the record. The Series A preferred placement (claim 12486) is marked delayed and requires explanation.

The 2024 activity looks like a deliberate return to the earlier template at much larger size: a $489.3 million ATM in a single quarter (claim 6207), then a $1 billion convertible (claim 5398) and an $850 million convertible (claim 5307), with stated uses of buying bitcoin and repurchasing 2026 notes. The planned repurchase amounts were revised between pricing and closing ($199 million to $212 million, claims 5425 and 5400; $48 million to $51 million, claims 5345 and 5310). The record does not confirm that either repurchase was executed. One claim gives the December offering as $700 million (claim 5348), which conflicts with the $850 million stated elsewhere.

The record contains equity-plan increases (claims 14758, 14555 and 6539) and RSU grants (claims 13964 and 13965) but no insider sales, so it cannot support conclusions about insider behavior. The data also ends abruptly in December 2024. A reader should look for evidence of how much of the 2026 notes remains ahead of the December 2026 maturity (claim 14387). Other open items are how the holder put rights on the 2024 notes are structured (claim 5405) and whether ATM issuance has resumed alongside the convertible stack.

Marathon's record traces a company that moved from micro-cap patching to programmatic capital raising, with dilution concentrated in equity programs until 2021 and in convertibles from 2021 onward. The early years show small, overlapping instruments: a $7.47M ATM capacity (claim #15203), an all-stock private placement (claim #15151), a convertible note whose conversion price was cut (claim #14913), and repeated equity issuance to vendors and facility operators (claims #14796, #14733). That reads as opportunistic financing driven by liquidity needs rather than a defined capital plan. The equity plan reserve was also raised repeatedly, by 10M, 5M and 7.5M shares (claims #15460, #14758, #14555), plus a further increase (claim #6539), so compensation dilution is a recurring layer.

From 2021 the scale changed. The $200M ATM (claim #14651) was followed by $650M of 1.00% notes plus a $97.5M option (claims #14387, #14369) and Silvergate revolving and term facilities of $100M each (claims #14440, #13836). The bank leg did not hold up: $35M was drawn (claim #13817) and the facilities were terminated in March 2023, with prepayment and notice waivers (claims #13180–#13182). ATM issuance then resumed at size, with $489.3M in Q1 2024 alone (claim #6207). The resulting mix is equity at the market, convertibles for size, and secured bank debt abandoned. That pattern looks coherent in hindsight, although it emerged by substitution rather than design.

On deployment, two items require explanation. The $67M bridge loan and $14M of expedite fees to Compute North (claims #14282, #14283) were followed by an unsecured bankruptcy claim (claim #12985), and the record does not quantify recovery. The 2024 notes, $1B and $850M (claims #5398, #5307), cited repurchase of the 2026 notes and bitcoin purchases (claims #5401, #5311). The stated repurchase amounts were about $199–212M for the first offering (claims #5425, #5400) and about $48–51M for the second (claims #5345, #5310, #5349), but the data does not confirm the repurchases or the bitcoin purchases as executed. The Series A preferred placement (claim #12532) is marked delayed and is no longer discussed.

The disclosed data contains no insider sales, and the record ends in December 2024, so the analysis cannot extend past that point. A careful reader should next look for confirmation of the 2026 note repurchases, for the holder repurchase and redemption dates on the 2024 notes (claims #5404, #5405), and for whether ATM issuance resumes alongside the convertible stack.

Marathon's record describes a financing posture that scaled by roughly two orders of magnitude in five years. The 2019-2020 period consists of small, repeated equity patches: an ATM program sized at $7,472,417 (claim 15203, restated in claim 14915), a $5,293,246 equity offering (claim 14895), and convertible note amendments that reduced conversion prices (claims 14913, 14892). The $100,000,000 S-3 (claim 14874) and the $200,000,000 ATM (claim 14651) mark the transition to institutional-scale issuance. Each later step was larger: the $650,000,000 convertible due 2026 with its $97,500,000 option exercised (claims 14387, 14369), an ATM draw of $161,042,274 (claim 13818), a Q1 2024 ATM of $489,300,000 (claim 6207), and two convertibles in late 2024, $1,000 million (claim 5398) and $850 million (claim 5307).

The mix reads as more coherent after 2021 than before it. Bank debt with Silvergate, a $100,000,000 revolver and term loan with liquidity and net-worth covenants (claims 14440, 13836, 13839), was terminated in March 2023 with fees waived (claims 13180, 13181). That ended the secured-lending approach. Since then the company has relied on ATM equity and unsecured convertibles, with a $200,000,000 line of credit (claim 5574) as the only new facility on record. The $14,250,000 Series A preferred placement (claim 12532) is the exception and is marked delayed.

The record only partly shows how proceeds were deployed. The $67,000,000 bridge loan and $14,000,000 in expedite fees to Compute North (claims 14282, 14283) were later followed by an unsecured bankruptcy claim (claim 12985), which needs explanation alongside the original rationale. The Abu Dhabi contribution appears as both $81,200,000 (claim 12974) and $406,000,000 (claim 13085). The record does not reconcile the company share against the project total. For the 2024 convertibles, the stated uses were repurchasing 2026 notes, $212 million per claim 5400 and $51 million per claim 5310, and buying bitcoin (claims 5401, 5311). Only a $50,000,000 repurchase is confirmed (claim 5349). The record contains no insider sales, only RSU grants (claims 13964, 13965).

Several things are worth watching. The first is whether the larger November 2024 repurchase was completed. The second is how much of the 2026 notes remains ahead of the December 1, 2026 maturity (claim 14387), which is now close. The third is whether bitcoin purchases are disclosed in amounts that match the stated net proceeds. The record is truncated after December 2024, so later events are not covered.

Marathon's record divides into two regimes. Through 2020 the company raised capital in small increments: a shelf of $7,472,417 (claim #15228), an ATM program of the same size with H.C. Wainwright (claim #15203), and an all-stock private placement of 2,335,000 shares that left 8,720,405 shares outstanding (claims #15151, #15152). The $100,000,000 S-3 (claim #14874) marks the transition, and 2021 brings a different order of magnitude: a $200,000,000 ATM (claim #14651), $650,000,000 of 1.00% convertible notes plus the $97,500,000 option exercise (claims #14387, #14369), and a $100,000,000 Silvergate revolver (claim #14440). ATM issuance reached $489,300,000 in Q1 2024 alone (claim #6207), and repeated equity incentive plan increases (claims #14758, #14555, #6539) add a steady layer of compensation dilution beneath the larger offerings. Dilution has been continuous and has accelerated.

The mix looks more deliberate from 2021 onward. Equity funds operations, converts provide large tranches of term capital, and bank credit adds liquidity. The bank leg did not hold. The Silvergate revolver and term loan (claim #13836) were terminated in March 2023 with fees waived (claims #13180, #13181), and a $200,000,000 line of credit appears only later (claim #5574). The 2024 converts partly refinance the 2026 notes, with planned repurchases of roughly $199M to $212M (claims #5425, #5400) and about $50M (claims #5349, #5310). Even so, most proceeds are new money rather than refinancing.

Deployment tracks the stated use of proceeds unevenly. The $67,000,000 bridge loan and $14,000,000 expedite fees to Compute North (claims #14282, #14283) were later followed by an unsecured claim in that counterparty's bankruptcy (claim #12985). Abu Dhabi contributions appear as $406,000,000, $81,200,000 and $406,000 across filings on the same date (claims #13085, #12974, #12953), which requires explanation. The 2024 notes name bitcoin purchases as a use of proceeds (claims #5401, #5311), but the record shows no corresponding deployment.

Several things are worth watching next. One is the remaining 2026 convertible balance and the holder repurchase date on the new notes (claim #5405). Another is whether the conversion price of roughly $26 (claim #5402) stays in range, and whether ATM usage resumes at 2024 scale. No insider transactions appear in this record, several amounts are recorded in inconsistent units, and the dataset is truncated after claim #5255, so conclusions about 2025 onward are not supported.

Marathon's record traces a company that moved from survival-scale financing to balance-sheet engineering in about three years. The 2019-2020 period relied on small, repeated shelf and ATM capacity near $7.5M (claims #15228, #14915), supplemented by stock paid directly for equipment and services (claims #15066, #14843, #14733). That is the profile of a micro-cap paying with paper. The step change came with the $200M ATM (claim #14651) and then the $650M convertible plus its $97.5M option (claims #14387, #14369), which together moved the capital stack from equity dribbles to institutional debt. Dilution did not stop when convertibles arrived. ATM issuance of $161M by mid-2022 (claim #13818) and $489.3M in Q1 2024 alone (claim #6207) indicate that equity remained the primary lever, with convertibles layered on top rather than replacing it.

The mix reads as directionally coherent but tactically reactive. The Silvergate revolver and term loan (claims #14440, #13836), carrying a $350M net-worth covenant (claim #13839), were terminated in March 2023 with fee and notice waivers (claims #13180, #13181, #13182), so that debt channel closed under external pressure rather than by plan. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) later surface only as a bankruptcy unsecured claim (claim #12985). The record does not show a recovery, and that outcome requires explanation.

Use-of-proceeds tracking is incomplete. The 2024 convertibles state bitcoin acquisition and repurchase of the 2026 notes (claims #5401, #5311). The planned repurchase figure was revised from $199M (claim #5425) to $212M (claim #5400), and the December deal targets roughly $51M (claim #5310). Actual deployment into bitcoin is not disclosed in this record. December sizing also requires reconciliation: $700M (claim #5348) appears against $850M completed (claim #5307). The $81.2M Abu Dhabi contribution sits against a $406M facility figure (claim #13085), and the record does not show the remainder being funded.

The record contains equity-plan increases (claims #14758, #14555, #6539) and RSU grants (claims #13964, #13965), but no insider sales, so that dimension cannot be assessed. The data appears to end in December 2024. A careful reader should watch four things: how much of the 2026 notes remains outstanding, whether the stock trades through the roughly $26 conversion price (claim #5402), whether ATM usage resumes alongside the convertibles, and whether the bitcoin purchases from note proceeds are ever reported separately.

Marathon's record shows three distinct financing regimes rather than one steady cadence. Through 2020 the company relied on small, repeated equity raises: at-the-market capacity of roughly $7.5 million (claims #15203, #14915), an all-stock placement of 2,335,000 shares (claim #15151), stockholder approval to issue above the 19.99% threshold (claim #15048), and a convertible note amended to a lower conversion price (claim #14913). That pattern reads as funding to survive rather than funding to a plan. The 2021 step-up was abrupt. A $200 million ATM (claim #14651) was followed within months by $650 million of 1.00% convertible notes due 2026 (claim #14387) plus the $97.5 million option (claim #14369). A Silvergate revolver (claim #14440) required a covenant waiver almost immediately to accommodate those notes (claims #14399, #14309). The 2022 term loan (claim #13836) and $35 million revolver draw (claim #13817) were unwound in March 2023, with the prepayment charge waived (claims #13180, #13181). That makes bank debt look like a bridge instrument rather than a durable part of the structure. By 2024 the mix had consolidated around ATM equity, with $489.3 million in Q1 alone (claim #6207), and back-to-back convertibles of $1 billion (claim #5398) and $850 million (claim #5307). That is more coherent than the early period, but it depends heavily on equity-linked paper. Use of proceeds is only partly traceable. The $67 million Compute North bridge loan and $14 million expedite fees (claims #14282, #14283) were later followed by an unsecured claim in Compute North's bankruptcy (claim #12985), a deployment outcome that differs from the original purpose. The Abu Dhabi contribution appears as both $81.2 million and $406 million (claims #12974, #13085), which requires explanation. The 2024 notes named 2026-note repurchases and bitcoin purchases as uses (claims #5400, #5401, #5311). The stated repurchase amounts differ across filings (claims #5345, #5349, #5310), and the record contains no confirmation of how much bitcoin was actually acquired. The record supplied ends mid-December 2024 and contains no insider transactions, so no read on insider selling is possible. Going forward, a reader should watch how much of the 2026 notes remains after the repurchases, whether conversion prices near $26 (claim #5402) turn the convertible stack into equity dilution, and whether the $200 million credit line (claim #5574) is drawn.

Marathon's financing record traces a company that moved from micro-cap patching to capital-markets scale in roughly two years, then repeated that escalation at a larger magnitude in 2024. The 2019–2020 period consists of small, overlapping instruments: an ATM capacity of about $7.47M (claims #15203 and #14915), an all-stock private placement of 2,335,000 shares (claim #15151), and a repriced convertible note (claim #14913). The shift comes with the $100M S-3 (claim #14874) and the $200M ATM (claim #14651). That ATM is recorded as completed in January 2021 (claim #14296) but the agreement claim is dated May 2021, a sequencing gap that requires explanation. From there the mix leans toward convertibles: $650M of 2026 notes plus the $97.5M option (claims #14387, #14369), and later a $1B offering with $980M net (claims #5398, #5399) followed by an $850M issue with $835M net (claims #5307, #5308). Equity dilution continued alongside this, including the $161M ATM tranche (claim #13818) and $489.3M of ATM sales in Q1 2024 alone (claim #6207). The pattern reads as deliberate layering rather than distress financing, though the debt leg has been revised repeatedly. The Silvergate revolver and term loan (claims #14440, #13836), with $35M drawn (claim #13817), were terminated in March 2023 with fees waived (claims #13180, #13181), and a $200M credit line appeared later (claim #5574). Use-of-proceeds tracking is uneven. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) were followed by an unsecured claim in that counterparty's bankruptcy (claim #12985). The Abu Dhabi joint venture contribution of $81.2M (claim #12974) is also marked delayed in one entry (claim #13066). The 2024 converts state two uses, repurchasing 2026 notes and buying bitcoin (claims #5400, #5401, #5311). The stated repurchase amounts are small relative to gross proceeds, roughly $212M and $51M (claims #5400, #5310). One claim lists the December offering at $700M (claim #5348) against $850M elsewhere, which requires reconciliation. No insider sales appear in this record; insider activity is limited to RSU grants (claims #13964, #13965). A careful reader should watch four things: the residual 2026 note balance, conversion behavior around the stated conversion price (claim #5402), exercise of the delayed greenshoe (claim #5309), and the holder repurchase date (claim #5405). The record is truncated after December 2024, so later events are not assessed here.

Marathon's record describes a company that moved from micro-cap patching to large-scale balance-sheet engineering, and the dilution cadence follows that arc. Before 2021, capital came in small, reactive pieces: at-the-market capacity of roughly $7.47M (claim #15203, #14915), an all-stock private placement (claim #15151), stockholder approval to issue above 19.99% (claim #15048), and a convertible note amendment that lowered the conversion price (claim #14913). Equipment and facility services were repeatedly paid for in shares (claims #15066, #14796, #14733), and the equity plan reserve was enlarged three times (claims #14758, #14555, #6539). Taken together, the early mix reads as opportunistic rather than planned.

From 2021 the structure becomes more deliberate. A $200M ATM (claim #14651, completion per #14325) was followed by the $650M convertible plus the $97.5M option, $747.5M in total (claims #14387, #14369, #14279), with a Silvergate revolver and term loan layered alongside (claims #14440, #13836). That bank relationship was then unwound in March 2023 with prepayment and notice waivers (claims #13180, #13181, #13182). The 2024 cycle repeats the pattern at larger scale: $489.3M of ATM issuance in one quarter (claim #6207), then a $1B convertible (claim #5398) and an $850M convertible (claim #5307), each partly earmarked to retire 2026 notes (claims #5400, #5310).

Use of proceeds has shifted materially. Early disclosures tie capital to mining hardware, such as the Antminer purchase (claim #14843). The Compute North bridge loan and expedite fees (claims #14282, #14283) were followed by a bankruptcy unsecured claim (claim #12985), a deployment the record does not revisit. By late 2024 the stated uses are bitcoin acquisition and refinancing (claims #5401, #5311) rather than operating capacity. The Abu Dhabi contributions appear in inconsistent units across claims #13085, #12974 and #12953, so their scale requires explanation. The Series A preferred placement (claim #12532) is marked delayed and is no longer discussed.

A careful reader should watch four things: how much of the 2026 notes remains after the partial repurchases, whether the $150M greenshoe (claim #5309, delayed) was exercised, how the roughly $26 conversion reference (claim #5402) compares with the share price ahead of the holder put date (claim #5405), and whether ATM usage resumes. The record contains no insider sale data, and the supplied list ends mid-entry in December 2024, so later events are not assessed here.

Marathon's financing record describes a company that changed scale rather than strategy. Through 2020 it funded itself in small increments: shelf and ATM capacity of roughly $7.5M (claims 15228, 15203, 14915), an all-stock private placement of 2,335,000 shares against 8,720,405 outstanding (claims 15151, 15152), a convertible note amendment that lowered the conversion price (claim 14913), and shares issued in place of cash for equipment and facility services (claims 15066, 14796, 14733). The stockholder approval to issue above 19.99% (claim 15048) and repeated increases to the equity plan reserve (claims 14758, 14555, 6539) confirm that dilution was the default funding tool in this period. From 2021 the same pattern repeats at far larger size. A $200M ATM (claim 14651) was followed by $747.5M of 1.00% convertibles due 2026 (claim 14279), a further ATM draw of $161.0M (claim 13818), and first-quarter 2024 ATM sales of $489.3M (claim 6207). The late-2024 convertibles, $1B completed with $980M net (claims 5398, 5399) and $850M completed with $835M net (claims 5307, 5308), extend the approach of pairing equity issuance with long-dated, low-coupon debt. The layer of bank debt is where the plan was revised. The Silvergate revolver and term loan (claims 14440, 13836), which carried a $350M adjusted net worth covenant (claim 13839), were terminated in March 2023 with notice and prepayment waivers (claims 13180, 13181, 13182). A $200M credit line later took their place (claim 5574). Recorded deployment is thin. The $67M bridge loan and $14M expedite fees paid to Compute North (claims 14282, 14283) appear next in the record as an unsecured bankruptcy claim (claim 12985), and that outcome requires explanation. The Abu Dhabi contribution of $81.2M equals 20% of the $406M project figure (claims 12974, 13085), which is internally consistent. The 2024 proceeds were earmarked for repurchasing 2026 notes and buying bitcoin (claims 5400, 5401, 5311), but the only completed repurchase on record is $50M (claim 5349), and the record contains no bitcoin purchases. The Series A preferred placement is still marked delayed (claim 12532). No insider sales appear, so that dimension cannot be assessed. The record stops in December 2024. A reader should watch how the remaining 2026 notes, which mature December 1, 2026 (claim 14387), are retired, and whether the stated bitcoin acquisitions are disclosed. The holder put dates on the 2024 notes (claim 5405) also matter, since a falling share price would push the company back toward ATM issuance.

Marathon's financing record shows three distinct phases, and the scale of dilution changes sharply between them. From 2018 through mid-2020 the company raised small amounts. The shelf and ATM capacity sat near $7.47M (claims #15228, #15203, #14915). Equity went directly to vendors for mining equipment and facility services (claims #15066, #14796, #14733). A convertible note was amended to lower its conversion price (claim #14913). This looks like opportunistic patching by a company with limited market access. The $100M S-3 in August 2020 (claim #14874) and the $200M ATM in 2021 (claim #14651) marked the move to institutional-scale funding. The $650M convertible (claim #14387) and its $97.5M option (claim #14369), $747.5M in total per claim #14279, confirmed it.

The 2021–2023 mix leaned on bank debt beside the convertibles. The Silvergate revolver and later term loan came with a $350M minimum adjusted net worth covenant (claims #14440, #13836, #13839). The company drew $35M against them (claim #13817), and ATM issuance reached $161M (claim #13818). Silvergate waived covenants when the convertibles were issued (claim #14309). The facilities were terminated in March 2023, with the prepayment and notice terms waived (claims #13180, #13181, #13182). Deployment in this period did not always track intent. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) were later followed by a bankruptcy unsecured claim (claim #12985). That claim's recorded amount appears unit-inconsistent and requires explanation. The same is true of the Abu Dhabi contribution figures (claims #13085, #12953). A $9.0M reverse termination fee (claim #6404) points to a transaction that was not completed.

The 2024 phase is more coherent. Q1 ATM issuance of $489.3M (claim #6207) and a $200M credit line (claim #5574) came first. Two convertible offerings followed, the $1B deal (claim #5398) and the $850M deal (claim #9275). Their stated use of proceeds shifted from infrastructure to buying bitcoin (claims #5401, #5311) and partially refinancing the 2026 notes. The disclosed figures for that refinancing vary within a few weeks: planned repurchases are $199 and $212 (claims #5425, #5400), and $48, $50M and $51 (claims #5345, #5349, #5310). Claim #5348 also records the offering as $700M rather than $850M, which requires reconciliation.

The record contains no insider sales. It shows only equity-plan expansions and RSU grants (claims #14758, #14555, #6539, #13964). The data ends in December 2024. A careful reader should watch the remaining 2026 notes due December 1, 2026 (claim #14387). They should also watch the holder put and redemption dates on the new notes (claims #5404, #5405), the $26 conversion reference (claim #5402), and whether ATM usage resumes to service or settle that stack.

Marathon's financing record describes a company that moved from micro-cap equity patching to large, recurring convertible issuance. The early years are small and reactive. There is an at-the-market program sized at roughly $7.47M (claims #15203 and #15228), a 2,335,000-share private placement (claim #15151), and a convertible note amendment that lowered the conversion price (claim #14913). These are the tools of a company funding month to month. The inflection is 2021. A $200M ATM (claim #14651) was followed by $650M of 1.00% convertible notes due 2026 (claim #14387) and a $97.5M option exercise (claim #14369), for $747.5M in total (claim #14279). One dating issue requires explanation: claim #14296 describes the $200M ATM as completed in January 2021, while the agreement itself is dated May 2021 in claim #14651.

The debt layer looks less coherent than the equity layer. The Silvergate revolver of $100M (claim #14440) needed a covenant waiver to accommodate the convertibles (claim #14399). It was expanded with a $100M term loan (claim #13836) and then terminated early, with prepayment and notice waivers, in March 2023 (claims #13180 and #13181). Deployment also diverged from intent. A $67M bridge loan and $14M in expedite fees to Compute North (claims #14282 and #14283) are later followed by an unsecured bankruptcy claim (claim #12985). The Abu Dhabi joint venture contribution appears as both $81.2M (claim #12974) and $406M (claim #13085), a discrepancy that requires explanation. The Series A preferred placement of $14.25M is marked delayed (claim #12486) and is not discussed afterward.

From 2024 the pattern becomes consistent. A $489.3M Q1 ATM (claim #6207) was followed by a $1B convertible with about $980M net (claims #5398 and #5399) and an $850M convertible with about $835M net (claims #5307 and #5308). Both carried stated uses of repurchasing 2026 notes and buying bitcoin (claims #5400, #5401 and #5311). The record states these plans but does not show actual deployment, so whether the proceeds were used as announced cannot be confirmed. The record contains no insider sales. Insider activity is limited to RSU grants (claims #13964 and #13965) and successive plan increases (claims #14758, #14555 and #6539).

A careful reader should look for three things next. The first is the residual balance of the December 2026 notes (claim #14387) after the disclosed repurchases, since maturity is close. The second is the size and conversion terms of any further convertible issuance, given the $26 conversion price on the 2024 notes (claim #5402). The third is whether any ATM activity since 2024 has been disclosed. The record ends in December 2024, so this read does not cover 2025 or 2026.

The trajectory here is one of scale change rather than gradual drift. For roughly three years the company financed itself at shell proportions — a $7,472,417 shelf that recurs verbatim across three separate 2019 filings (claims 15228, 15217, 15203), a $2,335,000 all-stock placement against only 8,720,405 shares then outstanding (claims 15151, 15152). The inflection is 2020-2021: a $100,000,000 S-3 (claim 14874), a $200,000,000 ATM (claim 14651), a $100,000,000 Silvergate revolver (claim 14440), then $747,500,000 of 1.00% converts due 2026 stacked in nine days (claims 14387, 14369, 14279). Within about eighteen months the financing unit moves three orders of magnitude.

The mix reads as coherent within each regime but the regimes differ. Phase one is equipment-linked: the $23,159,174 Antminer purchase (claim 14843) and $14,000,000 Phase I build (claim 14793) map plausibly onto the equity raised alongside them, including share-settled construction milestones (claims 14796-14799). Phase two is treasury-linked and says so — proceeds from the $1,000,000,000 November 2024 offering and the $850,000,000 December follow-on are explicitly earmarked for bitcoin (claims 5401, 5311). That is a stated strategy, not patching.

Where announced use and deployment diverge is the 2026 refinancing. The company described repurchasing $199,000,000 and then $212,000,000 of existing 2026 notes (claims 5425, 5400), and $48,000,000 in the December pricing (claim 5345); the executed figures on record are $50,000,000 and roughly $51,000,000 (claims 5349, 5310). The gap between the larger stated intention and the smaller completed repurchase requires explanation. Separately, the $67,000,000 Compute North bridge (claims 14593, 14282) resolves into an unsecured bankruptcy claim (claim 12985), and the $81,200,000 / $406,000,000 Abu Dhabi contributions (claims 12974, 13085) are carried as delayed and are not revisited.

A reader should watch three things: whether the remaining 2026 principal is retired before the holder repurchase and redemption dates (claims 5405, 5404); whether the three $150,000,000 greenshoes (claims 5423, 5343, 5309), all recorded as delayed, are ever exercised; and whether the $489,300,000 of Q1 2024 ATM issuance (claim 6207) continues alongside convertible issuance, since running both simultaneously is the clearest signal of how binding the capital need is. No insider sales appear on this record — only RSU grants (claims 13964, 13965), both delayed. Several entries carry inconsistent units, so magnitudes should be confirmed against source filings.

The trajectory describes two distinct companies. Through 2019 and 2020 the financing footprint was measured in single-digit millions — an ATM sized at $7,472,417 (claim #15203), a private placement leaving 8,720,405 shares outstanding (claim #15152), an H.C. Wainwright ATM at $79,817 (claim #15061) — and equity functioned as transactional currency rather than capital: $3,000,000 issued for operations and maintenance with milestone bonuses of $150,000, $150,000 and $200,000 (claims #14796, #14797, #14798, #14799), alongside a convertible note repriced downward (claim #14913). That is opportunistic patching, not a program.

The pivot is 2021. The $200M ATM (claim #14651), the $100M Silvergate revolver (claim #14440), and the $747.5M convertible stack — $650M plus the $97.5M option (claims #14387, #14369) — arrived within seven months and constitute deliberate layering: low-coupon converts for the balance sheet, a bank line for working capital, an ATM for the gaps. The stated deployment tracked initially, with $67M bridged to Compute North plus $14M in expedite fees (claims #14593, #14594). The outcome did not: that position resolves as a bankruptcy unsecured claim (claim #12985), and the bank channel closed entirely when Silvergate's facilities were terminated under fee and notice waivers (claims #13180, #13181, #13182).

What follows reads as a coherent second strategy. The Q1 2024 ATM alone printed $489,300,000 (claim #6207), and the November and December convertible offerings — $1B completed at $980M net (claims #5398, #5399) and $850M at $835M net (claims #5307, #5308) — were explicitly dual-purpose: retire the 2026 notes (claims #5400, #5310) and acquire bitcoin (claims #5401, #5311). Announced use and disclosed use align there.

Three items require explanation. The Abu Dhabi contribution appears as both $406,000 and $406,000,000 (claims #12953, #13085) and is not discussed in later filings. The $14.25M Series A preferred at a $15 conversion price (claims #12486, #12487) is anomalous against surrounding scale. Successive incentive-plan reserve increases (claims #15460, #14758, #14555) form a persistent low-grade dilution channel.

The record contains no insider transactions, so no read on selling behavior is possible, and it ends in December 2024. Watch the residual 2026 balance after repurchases, whether the $150M greenshoes were exercised (claims #5309, #5343), and whether ATM issuance resumes between convertible windows — that cadence, more than any single raise, sets the dilution slope.

The shape here is a small-cap that repeatedly re-based its own scale rather than growing into it. The 2019 apparatus — a $7.47M shelf and ATM (claims 15228 and 15203) and a $2.3M all-stock placement that left 8,720,405 shares outstanding (claims 15151 and 15152) — sits roughly three orders of magnitude below the 2024 apparatus, in which a single quarter of ATM issuance reached $489.3M (claim 6207) and two convertible offerings priced within five weeks of each other raised $1B and $850M (claims 5398 and 5307). Dilution did not accrete gradually; it stepped, and each step is legible against a bitcoin price regime rather than against an operating plan.

The instrument mix reads as sequenced rather than improvised. The ATM is the constant from 2019 through 2024, scaling from $7.47M to $200M (claim 14651) as the float supported it. Equity-for-services and equity-for-equipment appear precisely in the capital-constrained window — the $3M operations issuance and $750,000 Beowulf equipment purchase (claims 14796 and 14795), the pending $1.85M equipment share issuance (claim 15066, delayed). Secured bank debt arrives only once there was collateral to pledge (the $100M Silvergate revolver, claim 14440, and matching term loan, claim 13836), and converts arrive once the equity was liquid enough to price them (claims 14387 and 14369). The genuine discontinuity is the March 2023 Silvergate termination (claims 13180 and 13194), which removed the secured leg entirely; the $14.25M Series A preferred two months later (claims 12486 and 12532, both delayed) is too small relative to the balance sheet to be anything but gap-filling, and bank capacity was not restored until the $200M line in October 2024 (claim 5574).

Stated use of proceeds tracks actual deployment reasonably well, with one visible pivot: the 2021 vintage funded machines and hosting, while the 2024 vintage is explicitly bitcoin acquisition plus refinancing of the 2026 notes (claims 5401, 5311, 5400, 5310). Two items require explanation. The $67M Compute North bridge and $14M expedite fees (claims 14282 and 14283) resolve into a $40,000 unsecured bankruptcy claim (claim 12985) — that magnitude gap is unreconciled on this record. The Abu Dhabi structure, a $406M project with $81.2M attributed contribution (claims 13085 and 12974), carries a delayed flag (claim 13066) and is not discussed afterward.

Watch three things: the residual 2026 maturity after repurchases tranched at roughly $212M, $50M and $51M (claims 5400, 5349, 5310); whether share count stabilizes now that converts rather than the ATM carry the load, given the $26 conversion reference (claim 5402); and the record itself, which stops in December 2024 and leaves roughly twenty-one months unobserved.

The trajectory reads as three regimes rather than one policy. Through 2019 and mid-2020 the company financed itself in increments that would not cover a single month of later spending — a $7,472,417 shelf and matching ATM capacity, per claims #15228 and #15203, a $2,335,000 all-stock private placement, per claim #15151, and an ATM agreement whose early utility was measured at $79,817, per claim #15061 — alongside a habit of paying for hardware and services in shares rather than cash, visible in the October 2020 Beowulf arrangements (claims #14795 through #14799) and the 3,000,000-share facility-services issuance, per claim #14733. That is opportunistic patching, and it is not a criticism; there was no balance sheet to leverage. The incentive-plan reserve was enlarged in step, from $10,000,000 at inception (claim #15460) through increases in 2020, 2021 and 2023 (claims #14758, #14555, #6539).

The second regime, 2020 through 2022, shows genuine sequencing: a $100,000,000 shelf (claim #14874), a $200,000,000 ATM (claim #14651), secured bank capacity at Silvergate (claim #14440), then $747,500,000 of 1.00% convertibles (claims #14387, #14369). Equity first, cheap converts once equity had run — a conventional ladder. Deployment is where coherence thins. The $67,000,000 bridge loan and $14,000,000 of expedite fees to Compute North (claims #14593, #14594) resolve on this record into a bankruptcy claim recorded at $40,000, per claim #12985, a gap that requires explanation rather than inference. The Abu Dhabi commitment — $406,000,000 at project level, $81,200,000 at the company's share, per claims #13085 and #12974 — carries a delayed marker on claim #13066 and is not discussed thereafter.

The third regime, 2024, is the treasury-company template: $489,300,000 of ATM issuance in one quarter (claim #6207), then $1,000,000,000 and $850,000,000 of converts within three weeks (claims #5398, #5307), with proceeds designated for bitcoin and partial retirement of the 2026 notes (claims #5401, #5400, #5310, #5349). Announced and actual use appear to track here, but the retirements were partial against $747,500,000 outstanding, so gross leverage rose.

Watch the residual 2026 maturity (claim #14387) against the $26 conversion reference on the newer notes (claim #5402), whether ATM cadence resumes at 2024 scale, and whether Abu Dhabi contributions are reaffirmed. This record ends in December 2024. No insider sales appear — only RSU grants (claims #13964, #13965) — so that axis is unreadable from the data.

The trajectory here is one of scale change rather than acceleration in kind. From 2019 through mid-2020 the company financed itself in increments that would not cover a week of later operations: an ATM sized at $7,472,417 (claim 15203), a private placement of $2,335,000 against 8,720,405 shares then outstanding (claims 15151, 15152), a convertible note repriced downward at $999,106 (claim 14913). The $100 million shelf of August 2020 (claim 14874) is the hinge; by May 2021 the ATM capacity is $200 million (claim 14651), and by November 2021 the company places $650 million of 1.00% converts due 2026 with a $97.5 million greenshoe exercised days later (claims 14387, 14369). The 2024 sequence is the same instrument at another order of magnitude: $489.3 million drawn through the ATM in Q1 alone (claim 6207), then $1 billion of converts in November (claim 5398) and $850 million more within two weeks (claim 5307).

The mix is coherent in structure and opportunistic in timing. Equity is raised continuously through the ATM; debt appears when a lender will underwrite bitcoin collateral, as with the Silvergate revolver (claim 14440) and term loan (claim 13836), of which only $35 million was drawn (claim 13817) before the facilities were terminated in March 2023 with the 0.50% prepayment charge waived (claims 13180, 13181). Convertibles are used to refinance convertibles — $212 million of 2026 notes repurchased from the November proceeds (claim 5400), $51 million more from the December tranche (claim 5310).

Use of proceeds has migrated, and the migration is the most analytically significant fact on the record. Early raises attach to identifiable assets: Antminers at $23,159,174 (claim 14843), Phase I construction at $14,000,000 (claim 14793), a $67 million bridge loan to Compute North (claim 14282) that resolves into a $40,000 unsecured bankruptcy claim (claim 12985). The Abu Dhabi commitment of $406,000,000 (claim 13085) is carried with one contribution entry marked delayed (claim 13066) and is not revisited in later filings on this record. By late 2024 the stated purpose is bitcoin acquisition itself (claims 5401, 5311).

Insider activity is effectively absent here — two RSU grants, both delayed (claims 13964, 13965), and no sales — so no read is available on that dimension. Watch whether the 2026 maturity is retired in full or rolled again, and whether the Abu Dhabi contributions resume disclosure.

Marathon's record shows a sharp change in scale. Through 2019 and 2020 the company raised capital in single-digit millions: a shelf and ATM sized at roughly $7.5M (claims #15228, #15203), an all-stock private placement of 2,335,000 shares (claim #15151), and about $7.1M sold through the ATM in the first half of 2020 (claim #14849). The $100M S-3 in August 2020 (claim #14874) and the $200M ATM in 2021 (claim #14651) moved the company to a different order of magnitude. By Q1 2024 a single quarter's ATM activity reached $489.3M (claim #6207), which places dilution in the ongoing, programmatic category rather than the episodic one. The mix changed in phases rather than by improvisation. Equity came first. The $650M 2026 convertible and its $97.5M option (claims #14387, #14369) then added debt that could convert. Bank debt through Silvergate (claims #14440, #13836) followed and was terminated in March 2023 with fees waived (claims #13180, #13195). Late-2024 convertibles, recorded in millions, of $1,000 (claim #5398) and $850 (claim #5307) were used in part to repurchase 2026 notes, with $212 and $51 cited (claims #5400, #5310). This reads as a deliberate extension of maturities using convertibles, with the ATM carrying most of the equity load. The 2023 Series A preferred placement of $14.25M (claim #12486) is the one instrument that does not fit the pattern and remains unexplained in the record. On use of proceeds, the record documents stated intentions more fully than their results. The Compute North bridge loan of $67M and expedite fees of $14M (claims #14282, #14283) were later followed by a Compute North bankruptcy and an unsecured claim (claim #12985). That sequence requires explanation of how much was recovered. The Abu Dhabi figures also require reconciliation, since the record shows an $81.2M contribution (claim #12974) alongside a $406M project figure (claim #13085). The late-2024 offerings name bitcoin purchases as a use of proceeds (claims #5401, #5311), but the record contains no confirmation of what was actually deployed. The insider record is limited to RSU grants (claims #13964, #13965), and no insider sales are recorded, so no conclusion about insider selling can be drawn. The record supplied ends mid-entry in December 2024. Readers should watch for the remaining 2026 note balance, the holder repurchase and redemption dates on the new notes (claims #5405, #5404), and whether ATM issuance continues at the 2024 pace.

The record traces a company that moved from micro-cap patching to large-scale balance-sheet engineering, and the dilution cadence changes character along the way. Through 2020 the financings were small and reactive: an ATM sized at roughly $7.5M (claims #15203, #14915), a 2,335,000-share private placement against 8,720,405 shares outstanding (claims #15151, #15152), and a convertible note whose conversion price was cut by amendment (claim #14913). Equity was also used as currency for equipment and hosting services (claims #14843, #14796, #14733). The August 2020 $100M S-3 (claim #14874) marks the shift. After it, capital arrived in step-changes: a $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440), and $747.5M of 1.00% convertibles due 2026 (claims #14387, #14369, #14279), followed by $161.0M of ATM issuance by mid-2022 (claim #13818) and $489.3M in Q1 2024 alone (claim #6207).

The mix reads as increasingly deliberate rather than opportunistic. Bank debt was layered on in 2021-2022, with the revolver joined by a $100M term loan and a $35M draw (claims #13836, #13817). It was then unwound in March 2023 with fee and notice waivers (claims #13180, #13195, #13196). Since then, funding has rested on ATM equity and convertibles, with a $200M credit line added in October 2024 (claim #5574).

Use of proceeds tracks only partially. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) ended as an unsecured bankruptcy claim (claim #12985), and that recovery is not otherwise discussed. The Abu Dhabi joint venture appears as both $81.2M and $406M contributions (claims #12974, #13085), which requires explanation. The late-2024 convertibles, $1B and $850M (claims #5398, #5307), named 2026-note repurchases of $212M and $51M (claims #5400, #5310) plus bitcoin purchases (claims #5401, #5311). Only $50M of repurchase execution appears on record (claim #5349), and the planned figures were revised between announcement and closing (claims #5425, #5345).

The record has limits. It ends mid-entry in December 2024, discloses no insider transactions, and mixes units across claims. Readers should watch how the residual 2026 notes are settled at their December 2026 maturity (claim #14387), how close the share price sits to the roughly $26 conversion level on the new notes (claim #5402), and when the holder put dates fall (claim #5405). They should also watch whether ATM issuance continues at the 2024 pace.

The record traces a shift from micro-cap equity patching to large-scale convertible financing. Early activity was small and frequent. It included a shelf and ATM sized at about $7.47M (claims #15228, #15203), a 2,335,000-share private placement that left 8,720,405 shares outstanding (claims #15151, #15152), and a convertible note amendment that lowered the conversion price (claim #14913), which requires explanation. Equity was also used as payment for operations and facility services (claims #14796, #14733). The 2018 plan reserve was raised repeatedly (claims #14758, #14555, #6539), so compensation dilution ran alongside financing dilution. The trajectory then steepens. A $200M ATM (claim #14651) was followed by $747.5M of 1.00% 2026 convertibles, including the exercised $97.5M option (claims #14279, #14369), and later a $489.3M ATM in Q1 2024 alone (claim #6207).

The mix points to a repeated pattern rather than a single plan. ATMs fund ongoing needs, and convertibles are used when market access allows. Bank debt did not last. The Silvergate revolver and $100M term loan, with a $350M minimum net worth covenant (claims #13836, #13839), were terminated in March 2023 with prepayment and notice waivers (claims #13194, #13181, #13182). It was replaced by a $200M line of credit in October 2024 (claim #5574). The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) ended in a bankruptcy unsecured claim (claim #12985), and that deployment is no longer discussed in the record.

Use of proceeds tracks only loosely. The stated repurchase of 2026 notes was revised between announcement and close. It moved from $199M to $212M on the $1B deal (claims #5425, #5400) and from $48M to $51M on the $850M deal (claims #5345, #5310). One claim sizes the December offering at $700M (claim #5348) against $850M elsewhere (claims #5307, #9275), which requires explanation. Bitcoin purchases are stated as the intended use (claims #5401, #5311), but the record does not document actual deployment. The Series A preferred placement (claim #12532) and several ATM agreements carry a delayed status.

A careful reader should watch four things. The first is how much of the 2026 convertibles remains after the partial repurchases. The second is the holder put and redemption dates on the 2024 notes (claims #5404, #5405), which are referenced but not dated here. The third is whether the $150M greenshoe (claim #5309) was exercised. The fourth is whether ATM issuance continues at the 2024 pace. The supplied record contains no insider transactions and is truncated after December 2024, so later events cannot be assessed.

Marathon's record shows three distinct financing regimes, each much larger than the one before. Through 2020 the company raised capital in small increments: at-the-market capacity of roughly $7.5M (claims #15203 and #14915), an all-stock private placement of 2,335,000 shares that left 8,720,405 shares outstanding (claims #15151 and #15152), and a convertible note repriced lower (claim #14913). The August 2020 $100M S-3 (claim #14874) and the $200M ATM (claim #14651) mark the shift to a second regime, which culminated in the $650M 1.00% convertible due 2026 plus its $97.5M option (claims #14387 and #14369). In the third regime, equity sales became the main funding source rather than a supplement: $161M through the ATM by mid-2022 (claim #13818) and $489.3M in Q1 2024 alone (claim #6207). That steady ATM usage probably explains more of the dilution than the headline convertibles do.

The mix looks more deliberate after 2021 than before it. Early activity reads as patching, with small ATMs, repriced notes and equity paid to vendors (claim #14796). The 2021–2022 layering of a Silvergate revolver and term loan (claims #14440 and #13836) on top of the convertibles was coherent until the Silvergate facilities were terminated in March 2023 (claim #13180). That termination forced a return to equity and, later, a $200M line of credit (claim #5574). The late-2024 convertibles, $1B completed (claim #5398) and $850M completed (claim #5307), were partly structured to retire the 2026 notes. They therefore extend the maturity profile more than they change the strategy.

The record does not settle whether proceeds went where they were said to go. The $67M bridge loan and $14M in expedite fees to Compute North (claims #14282 and #14283) are later followed by an unsecured bankruptcy claim (claim #12985), and that sequence requires explanation. The planned 2026-note repurchases are revised across filings: $199M then $212M (claims #5425 and #5400), and $48M, $50M and $51M on the second offering (claims #5345, #5349 and #5310). Claim #5348 records a $700M offering that is not reconciled with the $850M figure. The stated plan to buy bitcoin (claims #5401 and #5311) has no deployment claim on record. The record contains RSU grants (claims #13964 and #13965) but no insider sales.

A careful reader should watch for four things. The first is how much of the $747.5M 2026 tranche (claim #14279) remains outstanding after the repurchases. The second is the holder put date on the 2024 notes (claim #5405). The third is whether ATM issuance resumes at the 2024 pace. The fourth is any disclosure confirming the bitcoin purchases. The supplied record is truncated at claim #5254, so later events are not reflected here.

The record traces three distinct capital regimes, and the transitions between them say more than any single raise. Through 2020 the company funded itself in small increments: a roughly $7.5M shelf and ATM capacity (claims #15228, #15203, #14915), shares issued for mining equipment (claim #15066), a stockholder vote to exceed the 19.99% issuance threshold (claim #15048), and a cut to a convertible note's conversion price (claim #14913). That pattern is closer to patching than to strategy. Dilution then scaled sharply. There was a $100M S-3 (claim #14874), a $200M ATM (claim #14651), and repeated increases to the equity incentive plan reserve (claims #14758, #14555).

Late 2021 marks a deliberate turn toward the balance sheet. The $650M convertible (claim #14387), with its $97.5M option exercised (claim #14369), sat alongside a $100M Silvergate revolver (claim #14440) that was later extended into a term loan carrying a $350M net-worth covenant (claim #13839). That bank relationship was terminated in March 2023, with prepayment and notice waivers (claims #13180, #13181, #13182). Equity never stopped, however. It ran from $161M of ATM issuance (claim #13818) to $489.3M in Q1 2024 alone (claim #6207). By 2024 the company was pairing ATM supply with the $1B convertible (claim #5398) and the $850M convertible (claim #5307), which reads as a coherent, if aggressive, reliance on both instruments at once.

Use of proceeds is where the record requires the most explanation. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) were later followed by an unsecured claim in that counterparty's bankruptcy (claim #12985). Abu Dhabi capital contributions appear as both $81.2M and $406M (claims #12974, #13085), likely reflecting different scopes. The 2024 notes named two uses, repurchasing 2026 notes ($212M and $51M, per claims #5400 and #5310) and buying bitcoin (claims #5401, #5311). The record does not show whether either was executed.

The data itself has limits. It ends mid-December 2024, it contains no insider transactions, and it mixes units, for example $650 versus $650,000,000 for the same notes. Several events tagged delayed appear from later claims to have closed. A reader should watch several things next: how much of the 2026 notes remains outstanding, the holder repurchase and redemption dates on the 2024 notes (claims #5404, #5405), whether ATM issuance continued at the 2024 pace, and whether disclosed bitcoin holdings reconcile with the stated deployment.

Marathon's financing record falls into three phases. Each is larger than the last, and the instruments change as the company grows. Through 2019 and 2020 the company funded itself as a micro-cap. It used a $7,472,417 at-the-market program with H.C. Wainwright (claims #15203, #14915), a 2,335,000-share private placement (claim #15151), stockholder approval to issue above the 19.99% threshold (claim #15048), and a convertible note amended to a lower conversion price (claim #14913). It also paid in stock for mining equipment and facility services (claims #15066, #14796, #14733). That was dilution through many small issuances rather than a planned program. The $100M S-3 (claim #14874) and the $200M ATM (claim #14651) marked the move to scale. The $650M 1.00% convertible and its $97.5M option (claims #14387, #14369) then made convertible debt the main source of capital. The Silvergate revolver and term loan (claims #14440, #13836) needed a covenant waiver to fit alongside the notes (claim #14309). Both were terminated in March 2023, with the prepayment charge waived (claims #13180, #13181), so bank debt was a short-lived part of the structure. Deployment in that period requires explanation. The $67M bridge loan and $14M expedite fees to Compute North (claims #14282, #14283) are followed by an unsecured bankruptcy claim (claim #12985). The Abu Dhabi joint-venture contribution appears as both $81.2M, marked delayed (claim #13066), and $406M (claim #13085), and the record does not reconcile the two. A $14.25M Series A preferred placement is also marked delayed (claim #12532), and a $9,012,500 reverse termination fee (claim #6404) has no stated context. The 2024 phase is more coherent. The $489.3M Q1 ATM (claim #6207) and a $200M credit line (claim #5574) came first. Two convertibles followed: $1B with $980M net (claims #5398, #5399) and $850M with $835M net (claims #5307, #5308). In both, announced repurchases of the 2026 notes ($212M and $51M, per claims #5400 and #5310) are small next to the stated plan to buy bitcoin (claims #5401, #5311). Stated use of proceeds has therefore shifted from hosting and equipment to balance-sheet bitcoin. The record shows RSU grants and plan increases (claims #13964, #14758, #14555) but no insider sales. It ends in December 2024 and contains nothing for 2025 or 2026. Readers should watch for disclosure of whether the $150M greenshoe was exercised (claim #5309), for how the remaining 2026 notes are handled, and for how the $26 conversion price (claim #5402) compares with trading levels ahead of the holder repurchase date (claim #5405).

Marathon's financing record shows a clear change in scale more than a change in method. The 2019–2020 period relied on small H.C. Wainwright at-the-market programs, such as the $7.5M agreement (claim 14848) and the $6.9M underwritten deal (claim 14850). It also paid for equipment and facility services in shares, including the $23.2M Antminer purchase (claim 14843) and the $3M operations issuance (claim 14796). From 2021 the same tools grew by orders of magnitude. The ATM went from $200M capacity (claim 14651) to $489.3M raised in a single quarter of 2024 (claim 6207). Convertible debt went from the $650M 1.00% notes due 2026 (claim 14387) to the $1B November 2024 offering (claim 5398) and the $850M December 2024 offering (claim 5307). Repeated increases to the 2018 plan's share reserve (claims 14758, 14555, 6539, 5026) add a steady compensation layer to that dilution.

The mix looks less like patching than like a funding approach that has been revised several times. Bank credit was tried and dropped. The $100M Silvergate revolver (claim 14440) and term loan (claim 13836) were terminated in March 2023, with prepayment and notice waivers (claims 13180, 13181, 13182). A $200M line of credit reappeared in late 2024 (claim 5574). Since then, convertibles have been the main instrument, with some proceeds used to retire earlier convertibles.

The stated use of proceeds needs closer checking. The late-2024 offerings named bitcoin purchases and repurchases of the 2026 notes (claims 5401, 5311). The repurchase amounts changed between announcement and closing: $199M became $212M (claims 5425, 5400), and $48M became $51M (claims 5345, 5310). The later treasury disclosure (claim 5023) fits bitcoin accumulation but does not tie specific dollars to specific purchases. Earlier deployments did not go as described. The $67M Compute North bridge loan (claim 14282) ended as an unsecured bankruptcy claim (claim 12985). The Abu Dhabi joint venture contribution (claim 13066) and the $14.25M Series A preferred placement (claim 12486) are both marked delayed.

The most immediate item to watch is the December 2026 maturity of the original 1.00% notes. The record shows partial repurchases (claims 5349, 4677) but no complete figure for what remains outstanding. Readers should also watch whether the 2025 convertible offering (claim 4675) repeats the pattern of refinancing plus bitcoin purchases, and how the roughly $26 conversion terms on the 2024 notes (claim 5402) compare with the share price. The data provided includes no insider transactions and ends partway through the July 2025 entries, so no conclusions about insider behavior or later capital events can be drawn from it. Several figures are also recorded in inconsistent units across duplicate claims and should be read with that in mind.

Marathon's financing record falls into three phases. In 2020 and early 2021 the company was a micro-cap relying on small equity raises. It used a $7.5M at-the-market facility (claim #14848), a $6.9M underwritten offering with H.C. Wainwright (claim #14850) and a reduced conversion price on a legacy note (claim #14913). It also paid vendors in stock, including the $3M equity issuance for facility services (claim #14733). The second phase added leverage on top of equity: a $200M ATM (claim #14651), a $100M Silvergate revolver (claim #14440) and $747.5M of 1.00% 2026 convertibles (claim #14279). By the Silvergate term loan, the lender required a $350M adjusted net worth covenant (claim #13839). That bank structure ended in March 2023, when Silvergate waived its notice period and prepayment charge (claims #13180, #13181). The third phase, from 2024 onward, relies on convertibles, preceded by a $489.3M ATM draw in a single quarter (claim #6207). The company completed a $1B offering in November 2024 (claim #5398), $850M in December (claim #5307) and another $850M in July 2025 (claim #4675), the last with $36.9M of capped calls (claim #4673). Share reserve increases under the 2018 plan recur across the period (claims #14758, #14555, #6539, #5026), which adds a steady layer of dilution beneath the headline raises. Taken as a whole, the mix looks less like patching than a deliberate move from issuing equity to deferring it through convertibles, with bitcoin named as the intended use (claims #5401, #5311). Earlier stated uses of proceeds held up less well. The $67M bridge loan to Compute North (claim #14282) later appears only as an unsecured bankruptcy claim (claim #12985). Abu Dhabi joint venture contributions of $81.2M are marked delayed (claim #13066). The Series A preferred placement is also marked delayed (claim #12486). Repurchases of the 2026 notes also call for reconciliation. The planned amount was revised from $199M (claim #5425) to $212M (claim #5400), and December figures of $48M, $50M and $51M appear across claims #5345, #5349 and #5310. In July 2025 a planned $50M repurchase (claim #4677) is followed by a reported $19M (claim #4672). No open-market insider transactions appear in the record beyond RSU grants (claims #13964, #13965), and the record is truncated at claim #4671. Readers should watch whether the recent notes settle in shares or cash, whether the Abu Dhabi commitment is completed or restructured, and whether reported repurchase and bitcoin purchase amounts match what was announced.

Marathon's financing record falls into three phases. Each is internally coherent, but the phases do not add up to a single durable strategy. From 2021 through 2022 the company layered bank credit onto its first large convertible: the Silvergate revolver of $100,000,000 (claim #14440), later joined by a $100,000,000 term loan (claim #13836), alongside $747,500,000 of 1.00% notes due 2026 (claim #14279). That bank relationship ended in March 2023 through waivers of notice and prepayment terms (claims #13180 and #13181), not through scheduled maturity. The $67,000,000 bridge loan to Compute North (claim #14282) later appears only as an unsecured bankruptcy claim (claim #12985), and its recovery is not discussed further in the record. The Abu Dhabi contribution of $81,200,000 is marked delayed (claim #13066), and a parallel figure of $406,000,000 (claim #13085) is recorded inconsistently elsewhere (claim #12953), which requires explanation. Equity dilution was steady rather than episodic, running through at-the-market sales of $161,042,274 (claim #13818), $489,300,000 in a single quarter of 2024 (claim #6207), and repeated incentive-plan reserve increases (claims #14555, #6539, #5026). The 2024 to 2025 phase leaned heavily on converts: $1,000 million (claim #5398), $850 million (claim #5307), and $950 million (claim #4660). The recorded conversion price fell from $26 (claim #5402) to $20 (claim #4667), which implies more shares per dollar of potential conversion. On use of proceeds, the stated bitcoin purchases (claims #5401 and #5311) are broadly consistent with the treasury figure recorded as 44,893 (claim #5023). The 2026 repurchases track less closely. Planned buybacks of the 2026 notes shrank from $212 million (claim #5400) to $51 million (claim #5310) to $19 million (claim #4672), against $50 million announced (claim #4677). The 2026 record then reverses direction: bitcoin sales (claim #9092), roughly $1.0 billion of convertible repurchases (claim #9265), a $785 million bridge-financed acquisition (claims #1871 and #928), an AI-oriented restructuring (claim #2952), new bitcoin-backed debt of $600 million (claim #11583), and a Q2 2026 net loss of $610 million (claim #613). The insider data contains only RSU grants (claims #13964 and #13965), so no conclusion on insider selling is possible, and the record supplied is truncated. A careful reader should watch four things: how the Barclays bridge is refinanced, the conflicting pledged-collateral figures (claims #11143 and #11575), whether further bitcoin is sold to service debt, and whether the AI pivot is funded by equity or by more convertibles.

Marathon's financing record falls into three phases, and each one changes the character of the balance sheet. The first, in 2021 and 2022, combined bank credit with a single large convertible: the Silvergate revolver (claim 14440) and the $650 million 2026 notes plus a $97.5 million option exercise (claims 14387, 14369). Equity drawn through the at-the-market program, $161 million per claim 13818, patched gaps as the Compute North exposure developed; that exposure included the $67 million bridge loan (claim 14282) and later an unsecured bankruptcy claim (claim 12985). The Silvergate facilities were terminated in March 2023 with fee and notice waivers (claims 13180, 13181). Smaller commitments from that period, the Abu Dhabi contributions (claim 13066) and the Series A preferred placement (claim 12532), are marked delayed.

The second phase, from 2024 through mid-2025, looks like a deliberate strategy rather than patching. A $489.3 million ATM quarter (claim 6207) was followed by three convertibles: $1 billion (claim 5398), $850 million (claim 5307) and $950 million (claim 4660). The stated conversion price fell from roughly $26 (claim 5402) to $20 (claim 4667), which implies more shares per dollar raised. The plan share-reserve increases (claims 6539, 5026) add to that dilution. The disclosures for these deals are also internally inconsistent in ways that require explanation. One claim puts the December offering at $700 million (claim 5348) against $850 million elsewhere, and the July 2025 repurchase of 2026 notes appears as both $50 million (claim 4677) and $19 million (claim 4672).

The third phase, in 2026, shows actual deployment diverging from the announced uses of proceeds. The 2024-25 raises cited bitcoin accumulation (claims 5401, 5311). Yet 2026 brings bitcoin sales (claims 2156, 9092), roughly $1 billion of 2030 and 2031 note repurchases (claims 2154, 2155, 9265), and bitcoin pledged as collateral (claims 11143, 11575). New bitcoin-backed debt of $600 million (claim 11583) and the $785 million Long Ridge debt assumption on a Barclays bridge (claims 1871, 928) point toward power and AI infrastructure, alongside a $48 million restructuring (claim 2952). The accumulation rationale is no longer discussed in the later disclosures.

The record contains only RSU grants for insiders (claims 13964, 13965) and no insider sales, so insider behavior cannot be assessed. From here, a careful reader should watch four things. The first is how the Barclays bridge is taken out. The second is whether the pledged-collateral figures, which differ between claims 11143 and 11575, are reconciled. The third is what remains outstanding across the 2030 and 2031 notes after the repurchases. The fourth is whether the Q2 2026 net loss of $610 million (claim 613) prompts a return to the ATM.

Marathon's financing record falls into three phases. First came equity-heavy issuance. Then came a concentrated run of convertibles. Most recently the company has turned to secured and bitcoin-collateralized debt. The 2022 base included roughly $730 million of convertible notes and $161 million raised through the at-the-market program (claims #13816, #13818). The ATM pace then rose sharply, with $489.3 million issued in Q1 2024 alone (claim #6207). Between November 2024 and July 2025 the company completed three convertible deals: $1.0 billion, $850 million and $950 million (claims #5398, #5307, #4660). The conversion price on the latest deal was about $20, below the roughly $26 set in November 2024 (claims #4667, #5402). Each dollar of that later paper therefore carries more potential share count. The $37 million capped call attached to the July 2025 deal suggests the company recognized this dilution cost (claim #4673).

The stated use of proceeds was consistent across these deals: acquire bitcoin and retire the 2026 notes (claims #5401, #5311). The bitcoin leg tracks. Holdings reached 44,893 by April 2025 (claim #5023). The refinancing leg narrowed over time. Planned 2026-note repurchases fell from about $212 million to $199 million, then to around $50 million, and finally to $19 million (claims #5400, #5425, #5310, #4672).

By March 2026 the direction had reversed. The company repurchased $1.0 billion of the 2030 and 2031 convertibles, and those repurchases are marked delayed (claims #9265, #2154, #2155). This coincided with bitcoin sales (claim #9092). Holdings then declined to 35,577 (claim #612). Borrowing shifted to secured and collateralized debt. Examples include the $785 million Long Ridge debt assumption backed by a Barclays bridge (claims #928, #1871), $600 million of bitcoin-backed facilities (claim #11583) and a $350 million collateralized line (claim #2498).

Whether this amounts to a coherent strategy or opportunistic patching depends on how the bridge is refinanced. Two figures require explanation on that point. Pledged bitcoin rose from 9,270 to 18,750 (claims #11575, #11143). Over the same period the company reported a $610 million quarterly net loss, negative $238 million free cash flow and a current ratio below 1.0 (claims #613, #1765, #10116). Several records are internally inconsistent. These include the Abu Dhabi contribution figures (claims #13085, #12953), the December 2024 note amounts (claims #5256, #5233) and a 50,000-bitcoin holdings figure (claim #2931). The insider record is sparse. It contains only 2022 RSU grants (claims #13964, #13965) and no disclosed sales.

A careful reader should watch four things. The first is the terms of the Barclays bridge takeout. The second is further growth in pledged bitcoin relative to holdings. The third is whether the $48 million AI restructuring plan (claim #2952) is followed by capital raised explicitly for that purpose. The fourth is whether the delayed Abu Dhabi contributions are resolved or no longer discussed.

Marathon's financing record describes three distinct phases rather than one continuous strategy. Through 2022 and early 2024 the company leaned on bank facilities and at-the-market equity, including the $161M ATM issuance per claim 13818 and the $489.3M Q1 2024 ATM per claim 6207, with the Silvergate facilities terminated in March 2023 under waived notice and prepayment terms (claims 13180, 13181). From August 2024 through July 2025 the mix shifted decisively to convertible notes: the $1B offering per claim 5398, the $850M offering per claim 5307, and the $950M offering per claim 4660. Since early 2026 the posture has moved again, toward retiring convertibles (the $1.0B repurchase per claim 9265), selling bitcoin (claims 9092, 6101), and borrowing against bitcoin (the $600M facilities per claim 11583 and the $350M line per claim 2498). Dilution pressure is therefore now mostly contingent rather than realized, though the conversion price fell from $26 per claim 5402 to $20 per claim 4667, which lowers the threshold at which the 2025 notes convert into equity. Each convertible was paired with partial repurchases of the 2026 notes, and the sizes vary across filings ($199 per claim 5425, $212 per claim 5400, roughly $50M per claims 5349 and 4677). That pattern reads as rolling maturities forward more than as a single plan. The stated use of proceeds requires explanation. The 2024 offerings were announced to fund additional bitcoin purchases (claims 5401, 5311), yet holdings are recorded at 44,893 per claim 5023 and 35,577 per claim 7612's companion claim 612, alongside a stated opportunistic sales strategy per claim 7971. Several items are marked delayed, including the Abu Dhabi contributions (claim 13066), the Series A preferred placement (claim 12532), and the special interest payment start date (claim 4668). Some inputs conflict and should be treated cautiously: claim 5348 cites a $700M offering against $850M elsewhere, claim 2931 cites 50,000 bitcoin against 35,577, and several figures lack consistent units. The record also contains no insider transactions, so no read on insider behavior is possible. A careful reader should watch four things: how the $785M Barclays bridge for Long Ridge (claim 1871) is refinanced; whether pledged bitcoin, 9,270 per claim 11575 and 18,750 per claim 11143, keeps rising as a share of holdings; whether negative free cash flow of $238M (claim 1765) and a current ratio under 1.0 (claim 10116) push the company back to the ATM; and how the 2026 restructuring charge (claim 2952) squares with the AI pivot.

Marathon's financing record describes a company that moved from equity-funded expansion to convertible-funded bitcoin accumulation, and more recently to asset-backed borrowing and balance-sheet repair. The early record is modest: the Silvergate facilities were terminated on waived terms (claims #13180, #13181), and a $14.25M Series A preferred placement followed (claim #12532). The at-the-market program then became the primary equity source, at $489.3M in Q1 2024 alone (claim #6207). From late 2024 the mix shifted to convertibles in quick succession: a $1B offering (claim #5398), an $850M offering two weeks later (claim #5307), and a $950M offering in July 2025 (claim #4660). The conversion price fell from $26 (claim #5402) to $20 (claim #4667), so each new tranche carries more potential share issuance per dollar raised. The $37M capped call (claim #4664) partly offsets that.

The 2024 offerings named bitcoin purchases as the main use of proceeds (claims #5401, #5311), and that use appears to have been carried out at the time. Holdings of 44,893 BTC in April 2025 (claim #5023) had fallen to 35,577 by Q2 2026 (claim #612), after a 15,133 BTC sale in March 2026 (claim #9092) and a newly stated opportunistic sales strategy (claim #7971). The concurrent repurchases of the 2026 notes were small relative to the capacity raised, and one was revised from $50M (claim #4677) to $19M (claim #4672). In 2026 the direction reversed: a $1.0B convertible repurchase (claim #9265), with the 2030 and 2031 components marked delayed (claims #2154, #2155).

The recent pattern looks more reactive than planned. It includes the $785M Long Ridge debt assumption backed by a Barclays bridge (claims #928, #1871), $600M of bitcoin-backed debt arranged after Q2 (claim #11583), and a $48M AI restructuring (claim #2952). All of this sits against a $610M Q2 net loss (claim #613), negative $238M free cash flow (claim #1765), and a current ratio under 1.0 (claim #10116). Pledged collateral is reported at 9,270 BTC (claim #11575) and 18,750 BTC (claim #11143), and that gap needs explanation. A separate 50,000 BTC holdings figure (claim #2931) does not reconcile with the 35,577 reported elsewhere.

The record contains no insider transactions, so that dimension cannot be assessed, and the dataset ends mid-entry. Readers should watch for four things: how the Barclays bridge is refinanced, whether the delayed 2030 and 2031 repurchases are completed, how much of the remaining bitcoin becomes encumbered, and whether the ATM is used again.

Marathon's record shows a funding model that moved from equity to convertibles and then to bitcoin-backed debt and asset sales. The supplied record does not include share counts, so dilution cannot be measured directly. Its direction can still be traced. The Q1 2024 at-the-market program raised $489.3M (claim #6207). Three large convertible deals followed: the $1B November 2024 notes with a $26 conversion price (claims #5398, #5402), the $850M December 2024 notes (claim #5307), and the $950M July 2025 notes (claim #4660). The last of these converts at $20 (claim #4667), which means more potential shares per dollar raised. The $36.9M capped call (claim #4673) partly offsets that. Pricing weakened across successive raises.

The instrument mix looked coherent through 2025. Each convertible was paired with partial repurchases of the 2026 notes (claims #5400, #5310, #4677). That is a steady pattern of extending maturities while raising new capital. By 2026 the pattern is harder to read. Marathon repurchased $1.0B of convertibles (claim #9265), but the 2030 and 2031 repurchase figures are recorded as delayed (claims #2154, #2155). It also added bitcoin-collateralized facilities of $600M and $350M (claims #11583, #2498). Separately, it assumed a $785M Barclays bridge loan for the Long Ridge acquisition (claim #1871). This reads less like a single plan than a sequence of refinancings.

Stated use of proceeds and actual deployment have drifted apart. The 2024 convertibles were raised explicitly to buy bitcoin (claims #5401, #5311). The 2026 record instead shows bitcoin sales (claims #9092, #6101), a stated strategy of opportunistic selling (claim #7971), and a $48M restructuring toward AI initiatives (claim #2952). Recorded holdings fell from 44,893 (claim #5023) to 35,577 (claim #612). A later figure of 50,000 (claim #2931) requires explanation. Pledged collateral is also inconsistent, reported as 18,750 (claim #11143) and 9,270 (claim #11575). Earlier records have unreconciled figures of their own. The Abu Dhabi contribution appears as both $81.2M and $406M (claims #12974, #13085). The December 2024 raise appears as both $700M and $850M (claims #5348, #5307).

Several items deserve close attention. The first is how the $785M bridge is taken out, and whether equity is part of it. The second is whether the delayed 2030 and 2031 repurchases are completed. Collateral coverage matters too as holdings shrink against growing bitcoin-backed borrowing. Finally, liquidity is thin: the record shows a Q2 2026 net loss of $610M (claim #613), negative free cash flow of $238M (claim #1765), and a current ratio below 1.0 (claim #10116). Insider transaction data does not appear in the supplied record, so no read on insider behavior is possible here.

Marathon's financing record describes a company that shifted from equity to convertible debt, and then to secured and asset-backed borrowing. After a small preferred placement in 2023 (the $14.25M Series A, per claim #12532), the main equity event on record is the $489.3M at-the-market program in Q1 2024 (claim #6207). Funding then moved to convertibles in quick succession: the $1B offering completed in November 2024 (claim #5398), the $850M issue weeks later (claim #5307), and the $950M issue in July 2025 (claim #4660). The disclosed conversion price fell from $26 (claim #5402) to $20 (claim #4667) across these deals. That leaves the later notes with more potential dilution per dollar raised, even with the $37M capped call (claim #4664). The two increases to the 2018 plan reserve (claims #6539 and #5026) add a smaller, steady source of share-count growth. No insider transactions appear in the record provided, so that dimension cannot be assessed.

The 2024–2025 sequence reads as a coherent strategy. Each convertible stated two uses: buying bitcoin (claims #5401 and #5311) and retiring the 2026 notes (claims #5400, #5310 and #4672). Holdings of 44,893 by April 2025 (claim #5023) are consistent with that stated deployment. The 2026 record looks different. The 2030 and 2031 notes were repurchased (claims #2154 and #2155), with a $1.0B repurchase figure attached (claim #9265). Bitcoin sales are recorded (claims #9092 and #6101), and holdings stand at 35,577 (claim #612). The company also took on $785M of bridge-financed debt for Long Ridge (claims #1871 and #9408) and arranged $600M of bitcoin-backed credit (claim #11583). The accumulation thesis is no longer discussed in its earlier terms. The treasury now serves as a funding source and as collateral, and an 'opportunistic' sales posture is now stated (claim #7971). The $48M restructuring toward AI initiatives (claim #2952) points to a strategic change of direction rather than a continuation.

Several items require explanation. Pledged bitcoin appears as 18,750 (claim #11143) and 9,270 (claim #11575). A 50,000 holdings figure (claim #2931) conflicts with the 35,577 reported the same month. Debt-to-equity is shown as both 140 and 1.49 (claims #591 and #10115). Several offering sizes and the special interest payment start date are marked delayed (claims #4671 and #4668). Against a Q2 2026 net loss of $610M (claim #613), free cash flow of -$238M (claim #1765), and cash of $421M (claim #592), the next things to watch are how the Barclays bridge is taken out, whether the ATM is reactivated, and how the pledged collateral is reconciled.

Marathon Digital's capital strategy evolved through three distinct phases, each revealing shifting priorities and market access. The initial phase—a $489.3 million at-the-market offering in Q1 2024, per claim 6207—was modest and measured. Beginning November 2024, the company shifted sharply to convertible issuances: $1 billion priced in November, $850 million in December with a $150 million greenshoe option (claims 5398, 5307, 5423), and a further $950 million offering in July 2025 with an additional $200 million option (claim 4660, 4662). This crescendo of convertible issuance, totaling roughly $3.1 billion gross, suggests the board viewed convertible leverage as preferable to straight equity dilution—a coherent if aggressive capital structure choice.

The stated use of proceeds was unambiguous: acquire additional bitcoin, per claims 5401 and 5311. Yet actual deployment tells a different story. Bitcoin holdings peaked at $44.9 billion in April 2025 (claim 5023) but had contracted to $35.6 billion by Q2 2026 (claim 612). Critically, in March 2026 the company repurchased $1 billion of its own convertible notes (claim 9265) rather than deploying proceeds toward bitcoin accumulation. Simultaneously, Marathon assumed $785 million in debt to acquire Long Ridge's energy operations (claim 928), signaling a strategic recalibration from pure mining toward integrated energy-mining infrastructure. By Q2 2026, the company had pivoted to pledging $18.75 billion in bitcoin as collateral for new lending facilities ($600 million in bitcoin-backed debt, per claim 11583) and executing an "opportunistic" sales strategy (claim 7971). The operational picture worsened materially: Q2 2026 generated a $610 million net loss with negative $238 million free cash flow (claim 1765), while EBITDA turned deeply negative at -$255 million (claim 11137). Gross margin remained strong at 83 percent (claim 11125), suggesting the deterioration stems not from core operations but financing costs and asset valuation changes.

The coherence here is conditional: convertible issuance followed a sensible playbook, and Long Ridge's acquisition was strategically defensible. But the shift from equity issuance to bitcoin-collateral-dependent lending in a rising-rate environment introduces structural risk. A sophisticated reader should monitor whether bitcoin pledging remains available at acceptable terms if BTC prices compress, whether the $421 million cash balance (claim 11138) is adequate given negative operating cash flow, and whether management's transition from accumulation to opportunistic sales reflects conviction about bitcoin price trajectory or distress liquidity management.

Marathon Digital's capital structure shifted from equity-dilutive opportunism to convertible-dominant stacking in 2024–2025, then broke course entirely with the Long Ridge acquisition. Through mid-2025, the company raised $2.8 billion in convertible senior notes across three offerings (November and December 2024, July 2025), per claims 5398, 9275, and 9273, while simultaneously repurchasing existing 2026 convertibles—a refinancing tactic that defers dilution but locks in higher debt costs. The preceding $489 million ATM offering in Q1 2024, paired with a $200 million credit facility, had established bitcoin accumulation as the stated use of proceeds. This thesis held through 2025: convertibles funded bitcoin purchases while the maturity ladder pushed refinancing risk forward.

The April 2026 Long Ridge acquisition introduced a structural fracture. The $785 million debt assumption, per claim 928, collided with deteriorating operations. Q2 2026 disclosed a $610 million net loss on $175 million revenue and negative $224 million operating cash flow, per claims 1764 and 11141. The digital asset fair value loss alone was $343 million, per claim 10125. This undermined the bitcoin-accumulation thesis: the treasury's mark-to-market deteriorated, reducing collateral adequacy and forcing repositioning.

By August 2026, Marathon pivoted to bitcoin pledging—$18.75 billion in collateral arranged—and arranged $600 million in bitcoin-backed credit facilities with a $150 million refinance, per claims 11143 and 11583. The shift from buyer to pledger signals collateral-based survival rather than strategic optionality. The announcement of an "opportunistic bitcoin sales strategy for 2026," per claim 7971, codifies this urgency: sales now require explanation rather than patience. A current ratio under 1.0 and $421 million cash against $2.59 billion liabilities, per claims 10116 and 11139, leave minimal room for deterioration.

The coherent strategy—leveraged bitcoin accumulation via convertibles—failed when Long Ridge proved value-destructive rather than accretive. Readers should monitor: bitcoin collateral adequacy as prices fluctuate, the maturity schedule on $1.93 billion long-term debt per claim 11140, whether sales become mandatory rather than opportunistic, and whether Q3 2026 shows operational inflection or continued burning. The company is refinancing into tighter constraints.

Marathon Digital's capital strategy shifted from coherent debt-management through refinancing to operationally misaligned during 2024-2026. The company executed multiple convertible offerings across late 2024 and 2025—a $1 billion offering completed November 2024 (claim 5398), additional offerings totaling approximately $850-950 million through July 2025 (claims 5307, 4675, 4660)—alongside a $489 million ATM equity offering in Q1 2024 (claim 6207). Capital flowed systematically into repurchases of maturing 2026 convertible notes ($461 million cumulative through late 2024 per claims 5425, 5400, 5349) and bitcoin accumulation, reaching 35,577 BTC by Q2 2026 (claim 612). This dual use—extend debt maturity, accumulate reserves—was internally consistent.

Strategy fractured in spring 2026. Marathon repurchased $1 billion of convertible notes in March (claim 9265) immediately before Q2 results disclosed $610 million net loss and negative $255 million EBITDA (claims 11136, 11137). The April Long Ridge acquisition assumed $785 million in debt (claim 928) yet contributed only claimed $144 million in annualized EBITDA (claim 10650), insufficient to offset consolidated deterioration. The March capital deployment preceded visible operational stress by weeks, indicating either forecasting failure or delayed disclosure of headwinds.

Current financial stress is acute. Cash of $421 million (claim 11138) against $2.59 billion in liabilities (claim 11139), current ratio below 1.0 (claim 10116), and debt-to-equity ratio of 1.49 (claim 10115) create refinancing urgency. Recent shift to bitcoin-collateralized facilities ($600 million arranged per claim 11583, with $9,270 BTC pledged per claim 11575) and August 2026 announcement of opportunistic Bitcoin sales strategy (claim 7971) indicate forced monetization replacing disciplined deployment.

The conversion price declined from $26 per share in November 2024 (claim 5402) to $20 in July 2025 (claim 4667), documenting equity value deterioration and embedded dilution from future conversion. Readers should watch: maturity wall timing from the 2024-2025 convertible cohort; whether declining bitcoin prices force accelerated collateral liquidation; and Long Ridge integration performance, where claimed $144 million contribution against negative overall EBITDA suggests integration complexity or strategic misalignment.

Marathon Digital pursued a coherent if increasingly aggressive capital strategy from 2024 through mid-2025, then shifted materially toward collateralized borrowing. The $489.3M at-the-market offering in May 2024 (per claim 6207) was followed by a $200M credit line in October (claim 5574), but the primary vehicle became convertible debt: four large issuances between November 2024 and July 2025—$1 billion (claim 5398, net $980M per claim 5399), $850M (claim 5307, net $835M per claim 5308), $850M (claim 4675), and $950M (claim 4660, net $940M per claim 4661)—totaling approximately $3.65 billion in less than nine months. The announced use of proceeds was bitcoin acquisition (claims 5401, 5311), and treasury holdings did grow during this period. Parallel to issuance, the company repurchased maturing 2026 convertible notes, culminating in a $1 billion repurchase in March 2026 (claim 9265) partly funded by a $15.1M bitcoin sale (claim 9092). The April 2026 Long Ridge acquisition for $785M in debt assumption (claims 928, 1871) marked a strategic pivot. Rather than raise fresh equity, Marathon assumed liabilities and by August-September 2026 had established $600M in bitcoin-backed credit (claim 8507), refinanced $150M through Coinbase (claim 8508), and arranged an additional $350M bitcoin-collateralized facility (claim 2498), with approximately $18.75B in bitcoin pledged as collateral (claim 11196). However, the underlying business deteriorated markedly. Q2 2026 showed a $610M net loss on $175M revenue (claims 584, 585), with negative operating cash flow of $224M (claim 11128) and negative free cash flow of $238M (claim 11129). Trailing EBITDA was negative $255.1M (claim 11137). Despite announcing $144M annualized adjusted EBITDA from Long Ridge (claim 10650), total company performance suggests the acquisition added liabilities without offsetting operational headwinds. A $48M restructuring for AI initiatives in June (claim 2952) signals operational reconsidering mid-crisis. Insider selling persisted through deterioration at $4.9M in Q1 2026 and $5.4M in Q2 (Form 4 data). The capital structure now depends on bitcoin collateral securing over $600M in borrowings, with a debt-to-equity ratio of 1.49 (claim 10115) and current ratio below 1.0 (claim 10116). Sophisticated readers should monitor covenant compliance if bitcoin prices decline, sustainability of negative free cash flow, and whether additional deleveraging or restructuring becomes necessary given the compressed liquidity position.

Marathon has pursued a capital-raising blitz since late 2024 that appears increasingly reactive rather than strategic. The company raised approximately $2.8 billion in convertible securities between November 2024 and July 2025—a $1 billion offering in November at a $26 conversion price, an $850 million offering in December, and a $950 million offering in July at a $20 conversion price (claims 5398, 5307, 4660). Alongside this, the company issued $1.3 billion in Series X Preferred Stock (claim 5212) and deployed a $489 million ATM offering in Q1 2024 (claim 6207). All stated a unified thesis: accumulate bitcoin as a treasury asset.

The thesis appears to have fractured. Bitcoin holdings did grow from approximately 44.9 thousand coins in April 2025 to 50 thousand by late August 2026 (claim 5023 and claim 2931), but the company simultaneously incurred a $343 million unrealized loss in Q2 2026 alone (claim 10125) and initiated an "opportunistic" sales strategy (claim 7971), generating $413 million in proceeds during that quarter (claim 6101). Rather than a patient accumulation plan, the pattern reads as tactical opportunism constrained by deteriorating fundamentals.

The Long Ridge acquisition in April 2026 layered $785 million in debt onto a company already burning capital (claim 928). Q2 2026 generated a $610 million net loss with negative operating cash flow of $224 million and negative free cash flow of $238 million (claims 613, 11128, 11129). TTM diluted earnings per share stand at negative $11 (claim 588). The cash balance has contracted to $421 million against $1.93 billion in long-term debt (claims 11138, 11140), pushing the current ratio below 1.0 (claim 10116) and debt-to-equity above 1.4 (claim 10115).

Insider sales have continued modestly but consistently—15 transactions totaling $4.9 million in Q2 2026 and 17 transactions totaling $5.4 million in Q3 2026. The company has pledged 9,270 bitcoin as collateral against credit facilities (claim 11142), effectively mortgaging the treasury asset.

A sophisticated reader should monitor whether the Long Ridge acquisition begins generating sufficient cash flow to stabilize the burn, and whether lender covenants on bitcoin-backed facilities force additional sales if collateral values decline. Near-term solvency depends on operational improvement rather than capital markets access at current multiples.

Marathon's capital strategy across 2024-2026 pursued leveraged bitcoin accumulation through convertible debt, then pivoted opportunistically to operational assets when the underlying business deteriorated. From November 2024 through July 2025, the company issued approximately $2.8 billion in convertible notes—$1 billion in November, per claim 5398; $850 million in December, per claim 5307; and $950 million in July, per claim 4660. Stated proceeds funded bitcoin treasury growth to 35,577 units by Q2 2026, per claim 612. Conversion prices of $26 and $20, per claims 5402 and 4667, reflected management's implicit view of stock upside. The prior year's $489 million ATM offering in Q1 2024, per claim 6207, provided direct equity dilution alongside this converted-debt loading.

Operational results invalidated the accumulation thesis. Q2 2026 reported a $610 million loss on $175 million revenue, per claims 613 and 585. Trailing twelve-month net income turned negative, and free cash flow declined $(238) million, per claims 587 and 10113. In March 2026, faced with maturity walls, the company repurchased $1 billion of due-2026 convertibles, per claim 9265—apparently financed by accelerated bitcoin sales ($413 million proceeds, per claim 6101) executed under a stated opportunistic strategy, per claim 7971. The company simultaneously layered new secured borrowing ($600 million bitcoin-backed facility, per claim 8507; $350 million additional bitcoin collateral line and $150 million revolving by September, per claims 2498 and 2479).

The Long Ridge acquisition in April 2026 ($785 million debt assumption, per claim 928) marked a strategic reset: operational assets and energy cash flows replaced treasury leverage as the financing rationale. Whether this trade stabilizes the business depends on whether Long Ridge operations generate the cash flow to service both acquisition debt and maturing convertibles. A debt-to-equity ratio of 1.49 with a current ratio below 1.0, per claims 10115 and 10116, leaves limited room for error. Insider selling began mid-2026 ($5.4 million in Q3, per Form 4 data) and remains selective—a potential sign of opportunistic rebalancing rather than systemic doubt, but worth monitoring alongside quarterly cash flow trends.

Marathon Digital's capital strategy from late 2024 through mid-2026 separates into two phases marked by divergent execution. The first phase—November 2024 through July 2025—consisted of systematic convertible issuance totaling approximately $2.65 billion: $850M in November, $850M in December (claims 5398, 5307), and $950M in July (claim 4660). To manage share dilution, the company deployed $37M in capped call hedges (claim 4673) and repurchased $69M of existing convertible notes (claims 4677, 4672) before completing a $1B repurchase of convertible notes in March 2026 (claim 9265). The announced use of proceeds was bitcoin treasury accumulation. Within those parameters, the strategy appeared coherent and disciplined.

Execution, however, diverged materially from stated intent. Bitcoin holdings peaked at 44,893 units in April 2025 (claim 5023) but fell to 35,577 units by August 2026 (claim 1219), concurrent with opportunistic bitcoin sales (claim 7971) totaling $15.1M in March 2026 alone (claim 9092). Instead of treasury accumulation, capital was redirected toward the Long Ridge acquisition in April 2026, which carried $785M in debt assumption (claim 928), followed by a $48M AI restructuring charge in June 2026 (claim 2952). By September 2026, the balance sheet had reorganized around new secured credit facilities: a $350M bitcoin-collateralized line (claim 2498) and a $150M credit facility (claim 2479).

Financial deterioration accelerated in Q2 2026: a $610M net loss (claim 613) driven largely by $343M in fair value losses on digital assets (claim 614). Debt-to-equity reached 1.49 (claim 10115) and free cash flow fell to -$238.5M (claim 10113). Insider selling accelerated to $4.9M in April and $5.4M in July.

The operative question for investors is whether Long Ridge represents calculated portfolio rebalancing or capitulation to operational constraints. The gap between stated bitcoin-accumulation strategy and actual liquidation, paired with substantial losses and depleting cash, suggests either a fundamental shift in capital allocation intent or financial pressure driving tactical necessity.

Marathon's capitalization trajectory reveals a deliberate but fragile bet: raise low-cost convertibles to fund bitcoin accumulation, then opportunistically de-lever when the stock appreciates. Through 2024 and into 2025, the company raised approximately $2.65 billion across three convertible offerings (per claims 5398, 5307, 9275 for the $850M November 2024 close; claim 4660 for the $950M July 2025 close). A May 2024 ATM offering of $489 million (claim 6207) provided additional equity capacity. The stated use consistently emphasized bitcoin acquisition (claims 5311, 5401), and the treasury reached $44.893 billion by April 2025 (claim 5023).

But 2026 marked a pivot. In March, the company repurchased $1 billion in convertible notes at appreciated valuations (claim 9265), suggesting management confidence in the stock price or pressure to simplify leverage. The same month brought a $15.133 million bitcoin sale (claim 9092)—modest from a $44.9B position, yet it preceded a harder shift. The April Long Ridge acquisition required assumption of $785 million in debt (claim 928), introducing operational leverage precisely when Q2 reported a $610 million net loss and $343 million digital asset fair value decline (claims 613, 614). Cash generation has deteriorated: trailing twelve months' levered free cash flow stands at negative $1 million (claim 594).

By August through September, Marathon reoriented. Rather than liquidating bitcoin, the company established $600 million in bitcoin-backed credit facilities and a $350 million bitcoin-collateralized line of credit (claims 8507, 2498), pledging the treasury as collateral for liquidity. A $413 million bitcoin sale materialized in early September (claim 6101). A June restructuring plan for AI initiatives ($48 million, claim 2952) hints at diversification beyond mining. Insiders sold $4.9 million in April and $5.4 million in July 2026 per Form 4 volume.

What requires explanation: the mismatch between a massive treasury and negative free cash flow suggests collateral, not cash, is the binding constraint. Sophisticated readers should track whether asset-backed lending facilities meet convertible maturity schedules, whether Long Ridge's cash flows justify its debt load, and whether the announced "opportunistic bitcoin sales strategy" (claim 7971) reflects tactical rebalancing or emerging capital need. The company's ability to refinance without further equity dilution now depends on bitcoin volatility and lender appetite for asset-backed facilities.

Marathon Digital's capital strategy shifted markedly across three phases, each prompted by market conditions rather than a coherent long-term thesis. From early 2024 through mid-2025, the company pursued consistent dilution through two channels: a $489.3 million at-the-market offering in Q1 2024 (claim 6207) and a cascade of convertible issuances—$1 billion in November 2024 (claim 5398), $850 million in December 2024 (claim 5307), and $950 million in July 2025 (claim 4660). Each convertible cycle explicitly funded Bitcoin acquisition (claims 5401, 5311), creating a coherent if leveraged strategy. However, March 2026 marked an inflection: the company repurchased $1 billion of existing convertibles (claim 9265) while simultaneously assuming $785 million in debt through the Long Ridge acquisition (claim 928) and securing a $785 million Barclays bridge facility (claim 1871). This simultaneous liability management and debt accumulation suggests capital structure arbitrage rather than operational confidence. The stated use of proceeds—accumulating digital assets—held through mid-2026, but recent disclosures reveal strain. Q2 2026 saw a $343 million fair value loss on the $35.6 billion Bitcoin treasury (claims 614, 612), a net loss of $610 million on $175 million revenue (claims 613, 585), and negative trailing twelve-month free cash flow (claim 594). The company has since announced opportunistic Bitcoin sales (claim 7971) and layered new credit facilities: a $600 million bitcoin-backed credit line (claim 8507) and a $350 million bitcoin-collateralized facility (claim 2498). These moves signal liquidity pressure rather than optionality. Insider selling has accelerated—17 transactions totaling $5.4 million in Q2 2026 (quarterly Form 4 data). The debt-to-equity ratio of 140 (claim 591) leaves little room for error. Watch whether Long Ridge's energy hedges ($76 million per claim 2517) generate sufficient recurring cash to stabilize the leverage pyramid, or whether the company must continue discretionary Bitcoin sales to service obligations—a dynamic that becomes self-defeating should asset prices weaken further.

Marathon Digital's capital architecture over eighteen months pivots sharply from disciplined equity-raising to balance-sheet liabilities hedged by operational and treasury assets. The company executed a $489.3M at-the-market offering in Q1 2024, per claim 6207, then reversed course into convertibles: $1 billion in November 2024 (claim 5398), $850 million in December 2024 (claim 5342), and $950 million in July 2025 (claim 4671), for a total convertible issuance of roughly $2.8 billion over nine months. This stack was partially un-done—Marathon repurchased approximately $120 million of its 1.00% convertible notes due 2026 in mid-2024 and early 2025 (claims 5349, 5310, 4677, 4672)—suggesting either opportunistic buying back or refinancing pressure. The stated use of proceeds was consistent: acquire additional bitcoin. Yet the actual trajectory of holdings contradicts this thesis. Bitcoin treasury stood at approximately $44.9 billion in April 2025 (claim 5023) and declined to $35.6 billion by August 2026 (claim 612), a contraction of roughly $9 billion despite cumulative raises of $2.8 billion in convertibles plus $489 million in ATM proceeds. A $343 million digital asset fair value loss in Q2 2026 (claim 614) explains part of the gap, but Marathon also executed strategic bitcoin sales: $15.1 million in March 2026 (claim 9092) and $413 million in September 2026 (claim 6101), per the company's stated "opportunistic" sales strategy (claim 7971). The disconnect warrants explanation: did bitcoin prices decline sharply, were sales forced by operational needs, or was the acquisition thesis itself revised? The late-stage capital mix tells a different story. In April 2026, Marathon acquired Long Ridge Energy—$785 million in debt assumption (claim 928)—financed by a Barclays bridge facility (claim 1871), not equity. By August 2026, the company had layered in $600 million in bitcoin-backed credit facilities (claim 8507) and a $150 million line of credit (claim 2479). Negative free cash flow of $238 million in Q2 2026 (claim 1765) and negative trailing twelve-month EPS of $11 (claim 588) coincided with this shift. Sophisticated readers should monitor whether the Long Ridge mining operations stabilize the company's cash generation and whether the bitcoin-backed lending facilities expand—both would indicate the treasury is becoming a financing liability rather than a strategic asset.

Marathon Digital's capital structure from late 2024 through mid-2026 tells a story of announced strategy meeting operational friction. The company raised roughly $2.65 billion in gross convertible proceeds across three offerings—$1 billion in November 2024 (claim 5398), $700 million in December 2024 (claim 5348), and $950 million in July 2025 (claim 4660)—with stated use of proceeds allocated toward acquiring additional bitcoin. Yet bitcoin holdings declined from approximately 44,893 BTC in April 2025 (claim 5023) to 35,577 BTC by August 2026 (claim 612), suggesting either deployment into operations or tactical liquidation rather than accumulation. The conversion prices on successive offerings compressed from $26 per share (claim 5402) to $20 per share (claim 4667), reflecting stock weakness despite nearly $3 billion raised.

This dilution trajectory shifted abruptly in April 2026. Rather than continuing the convertible-equity mix, Marathon pivoted to debt-funded inorganic growth, assuming $785 million of debt in the Long Ridge acquisition (claim 928) after securing a Barclays bridge facility (claim 1871). Two months later came a $48 million restructuring plan earmarked for AI initiatives (claim 2952), signaling recognition that bitcoin mining alone no longer drove value creation. The company's financial performance corroborated this urgency: Q2 2026 posted a $610 million net loss (claim 613), with $343 million attributed to digital asset fair value declines (claim 614). The trailing twelve-month net loss deteriorated further (claim 587).

By September 2026, Marathon had assembled a multi-tranche credit stack—$350 million in bitcoin-collateralized facilities (claim 2498), a $150 million line of credit (claim 2479), and $600 million in aggregate bitcoin-backed facilities (claim 8507)—a structure that inverts typical credit relationships. Insiders sold $4.9 million worth of stock in Q2 2026 and $5.4 million in Q3 2026 (Form 4 volumes), without corresponding accumulation, suggesting diminished confidence despite stated AI pivot. A sophisticated reader should monitor whether the company can stabilize operations before the collateral-heavy credit facilities tighten, and whether announced AI restructuring translates to positive EBITDA. The delayed convertible repurchases announced in March 2026 (claims 2154, 2155) remain unfunded.

Marathon Digital's capital raising over eighteen months reflects a company managing a refinancing cycle through convertible debt, with material divergence between announced intent and actual deployment. The company raised approximately $2.6B through three consecutive convertible offerings (November 2024: $980M net per claim 5399; December 2024: $835M net per claim 5344; July 2025: $940M net per claim 4661), plus a $489M at-the-market offering in May 2024 (claim 6207). Stated use of proceeds focused on bitcoin acquisition (claims 5401, 5311), yet reported holdings declined from $44.9B in April 2025 (claim 5023) to $35.6B by mid-August 2026 (claims 612, 1762) before recovering to $50B by late August (claim 2931), with the company announcing opportunistic bitcoin sales by August (claim 7971).

The capital structure reveals defensive refinancing through aggressive repurchases of 2026-maturity convertibles totaling approximately $170M (claims 5349, 5310, 4677, 4672), clearing the near-term maturity wall. However, delayed repurchases of 2030 and 2031 convertibles totaling $913M (claims 2154, 2155) indicate the debt stack was extended rather than eliminated, presenting a coherent near-term strategy masking longer-term refinancing risk.

The strategic frame shifted in 2026 with the April Long Ridge acquisition assuming $785M in debt (claim 928) and June AI restructuring ($48M, claim 2952), pivoting from treasury accumulation toward operational leverage in energy and computing. This repositioning occurs against deteriorating fundamentals: Q2 2026 showed a $610M net loss (claim 613), negative trailing free cash flow (claim 594), and debt-to-equity of 1.40 (claim 591). The company accessed $150M conventional credit plus $350M in bitcoin-collateralized facilities by early September (claims 2479, 2498), indicating active liquidity management.

Insider selling remained steady at $4.9M and $5.4M in Q1 and Q2 2026 respectively. For sophisticated readers, the key question is whether convertible tranches (priced at $26 per claim 5402 and $20 per claim 4667) convert at planned levels or face refinancing pressure as maturities compound. The critical test: whether Long Ridge's energy hedges ($76M and $39M per claims 2517, 2518) generate sufficient cash flow to service debt and stabilize leverage against current negative free cash flow.

Marathon raised roughly $3 billion through convertible offerings ($1 billion November 2024, $850 million December 2024, $950 million July 2025, per claims 5398, 5307, 4660) and an ATM program ($489.3 million Q1 2024, claim 6207). The stated deployment thesis was bitcoin accumulation, which the company pursued aggressively, building holdings to $44.9 billion by April 2025 (claim 5023). That thesis is now reversing: bitcoin holdings declined to $35.6 billion (claim 1219), and $413 million was liquidated in September 2026 (claim 6101). Operating performance has deteriorated sharply, with a $610 million net loss in Q2 2026 (claim 613) and negative trailing free cash flow (claim 594). Separately, Marathon assumed $785 million in debt related to the Long Ridge acquisition (claim 928), introducing a second major liability alongside the convertible stack. Insider selling has been modest and consistent ($4.9 million in Q2 2026, $5.4 million in Q3 2026), reflecting neither panic nor strong conviction. The coherence of capital strategy is difficult to establish. Proceeds raised for bitcoin accumulation are being liquidated; convertible funding is flowing into operating losses rather than the declared asset class; and a separate acquisition debt burden has materialized. The debt-to-equity ratio stands at 140 (claim 591). Sophisticated readers should focus on two near-term stress points. First, the recent convertibles carry specific conversion prices (claims 4667, 5402) and maturity windows; if the stock trades below conversion levels or approaches maturity, holders may demand redemption rather than conversion, forcing refinancing at potentially adverse terms. Second, Long Ridge's cash generation becomes critical—the combined debt load (convertible stack plus $785 million acquisition debt) requires material EBITDA to service; shortfalls in integration would complicate near-term debt servicing given the negative free cash flow position. The bitcoin liquidations suggest those holdings are pledged for credit facilities or earmarked for debt service rather than held strategically.

Marathon Digital pursued convertible debt issuance across 2024 and 2025 according to an opportunistic rather than methodical cadence. The company issued approximately $1.85 billion in convertibles: $1 billion in November 2024 (claim 5398), $850 million in December 2024 (claim 5348), and $950 million in July 2025 (claim 4660), alongside a $489.3 million ATM offering in Q1 2024 (claim 6207). Layered within these issuances were repurchases of 2026-due convertibles totaling roughly $310 million across multiple tranches (claims 5349, 5310, 4672), suggesting the company managed near-term debt maturities while extending duration. The pattern of rapid-succession convertible offerings followed by refinancing of earlier tranches is consistent with tactical opportunism rather than strategic capital planning.

Stated use of proceeds centered on bitcoin acquisition (claims 5401, 5311). Bitcoin holdings reporting across disclosures, however, is materially inconsistent: $44.9 billion as of April 2025 (claim 5023), $35.6 billion in August 2026 (claim 612), and $50 billion by late August 2026 (claim 2931). These contradictions require explanation from management. What is evident is that the company recognized $343 million in digital asset fair value losses in Q2 2026 (claim 614) and reported $413 million in bitcoin sales proceeds (claim 6101), indicating active asset liquidation rather than passive accumulation.

Capital deployment subsequently pivoted. The Long Ridge acquisition brought $785 million in assumed debt (claim 928), and a $48 million restructuring charge for AI initiatives appeared in June 2026 (claim 2952). This M&A and restructuring activity occurred as core financials deteriorated sharply: Q2 2026 net loss totaled $610 million (claim 613), TTM levered free cash flow was negative $1 million (claim 594), and debt-to-equity reached 140 percent (claim 591). Insider sales accelerated in April and July 2026, aggregating $10.3 million, a potential signal of executive confidence withdrawal.

Most recently, the company drew upon a bitcoin-collateralized credit facility for $350 million (claim 2498) and a $150 million unsecured line (claim 2479), both announced in September 2026. The capital structure trajectory shows opportunistic issuance and tactical refinancing rather than coherent strategy, with proceeds deployed into bitcoin, subsequently liquidated amid losses, and capital redeployed into M&A against a background of negative free cash flow. Watch for Long Ridge cash generation realization, convertible refinancing execution if equity prices decline, and bitcoin collateral haircuts should digital asset prices compress materially.

Marathon's record contains almost no customer agreements in the commercial-diligence sense, and the claim count overstates how many distinct relationships exist. Roughly forty claims collapse to about fifteen counterparties once repeated filings are de-duplicated. The Compute North letter of intent alone appears in claims 14590, 14304, 14311 and 14331, and the Beowulf power agreement repeats across 14791, 14647 and 14639. The two earliest items, the UN Women (20896) and EY (20895) partnerships, refer to the Mara Foundation, a different organization, and should be excluded from the MARA record entirely.

Every counterparty is named. Nowhere does the record describe unidentified 'strategic partners' or 'anchor customers.' The issue is direction, not anonymity: nearly every named party is a supplier, lender, or service provider that Marathon pays. In the patent era, Zhone (21022) and Siemens Technology (21018) were sellers of portfolios acquired alongside licensing agreements, and IP Nav (21024) was a deal-flow source. In the mining era, the counterparties are operating vendors. These include Block Maintain, Securitas and Chord Advisors (15198–15200), Beowulf and 2P1 for power and operations (14791, 14649), Compute North for hosting (14861, 14333), and DMG as a licensor that Marathon paid $500,000 (14305). ViaBTC (14888) and Cumberland DRW (14889) are a mining pool and a liquidity desk, not buyers of a product. The NYDIG securities loan (14306, 14290) is the closest thing to a revenue counterparty, and it is a treasury yield arrangement.

On signed versus provisional language, the Beowulf power purchase agreement (14791, 14801) and the services agreement (14738) are executed contracts. The Compute North relationship began as a binding letter of intent (14590). It was later formalized through a master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347), so its status did firm up over time.

Concentration has shifted from patent sourcing to dependence on two infrastructure suppliers, Beowulf at Hardin and Compute North for 73,000 hosted miners. That is supplier concentration, not customer concentration.

No revenue arithmetic can be reconciled from these claims, because the dollar figures are all cost-side. They include the $67 million hosting commitment (14331), the $14.6 million deposit (14333), the $11.22 million equity payment to Beowulf and 2P1 (14649), and the DMG license fee. The '$67' and '$500' labels on claims 14311 and 14315 appear to be truncated extractions of those same figures, not separate deals. Marathon's actual revenue comes from block rewards and fees through the pool, so its customer disclosures are best read as a map of operating dependencies.

Marathon's record contains almost no signed customer agreements in the conventional sense. Nearly every named counterparty is a supplier, host, lender or financing agent, so the file documents what Marathon buys rather than who pays it. The two 2014 items, UN Women (claim 20896) and EY (claim 20895), refer to the Mara Foundation, an African entrepreneurship body. They appear to be an entity-matching error and should be excluded. The patent-era record names Zhone (claim 21022) and Siemens Technology (claim 21018) only as sources of acquired portfolios, and IP Navigation Group (claim 21024) as a deal-flow supplier. The 30% special-purpose-entity waterfall (claim 15223) leaves both the licensees and the entity's partners unnamed. That is the only material unidentified counterparty in the file.

After the move to mining, every counterparty is named, but none is a customer. Block Maintain, Securitas Canada and Chord Advisors (claims 15198-15200, 15068-15070, 14859-14860) are operating vendors. H.C. Wainwright (claim 15202) and NYDIG (claims 14306, 14290, 14316) are financing relationships. ViaBTC (claim 14888) is the pool that pays block rewards, and Cumberland DRW (claim 14889) is the liquidity desk. The Beowulf power purchase agreement and lease (claims 14791, 14801, 14802, 14647, 14639) and the Data Facility Services Agreement with 2P1 and Liefern (claims 14756, 14738, 14734, 14649, 14640) are executed contracts. The DMG Blockseer licence, at $500,000 (claims 14305, 14288, 14330), is also executed. The "$500" figure in claim 14315 is an extraction artifact.

The Compute North relationship needs the most careful reading. It began as a "binding letter of intent" for 73,000 miners worth $67 million (claims 14590, 14304, 14331). Claim 14311's "$67" is truncated. The relationship then became a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). The commitment was therefore formalised in stages, and only the September 2021 agreement and the joint venture clearly amount to definitive contracts.

Concentration moved from a single Granby operator to two dominant counterparties by 2021: Beowulf in Hardin and Compute North for the bulk of the hashrate. Compute North's later bankruptcy falls outside this record, but it shows what that host concentration risked. On revenue, the record contains no disclosed revenue figures to reconcile. All of the dollar amounts it does contain ($67 million, $14.6 million, $11.22 million in equity, $500,000) are cost or capital commitments. Revenue therefore cannot be tested against these announcements, and it depends on hashrate and the bitcoin price rather than on any contracted customer.

None of the claims on record for Marathon Digital is a signed customer agreement in the conventional sense, so the record should not be read as evidence of a customer franchise. The two 2014 items (claims 20896, 20895) concern the "Mara Foundation," an African entrepreneurship body that appears to be a different entity from the Nasdaq issuer, and they should be excluded. The 2013-2014 patent-era items are inbound transactions. The Siemens and Zhone "licensing agreements" (claims 21018, 21022) were the means by which Marathon acquired patent portfolios, and IP Nav (claim 21024) is described explicitly as a supplier of deal flow, not a licensee. The 2019 special-purpose-entity arrangement (claim 15223) is a contingent 30% interest in litigation proceeds, not contracted revenue. From 2019 onward, every named counterparty is a vendor, landlord, lender or financing agent. These are Block Maintain, Securitas, Chord Advisors, H.C. Wainwright, Beowulf, 2P1, Liefern, Compute North and DMG (claims 15198-15200, 15202, 14791-14802, 14756, 14333, 14305). The only parties that pay Marathon are the ViaBTC mining pool (claim 14888), which is protocol rewards rather than a negotiated customer contract, and NYDIG, which pays interest on loaned bitcoin (claims 14306, 14316). To its credit, the record names its counterparties throughout. There is no "anchor customer" or unnamed "strategic partner" language. On contract form, the Beowulf power purchase agreement and the data facility services agreement read as executed, with defined terms and share consideration (claims 14639, 14640, 14649). By contrast, the Compute North hosting relationship began as a "binding letter of intent" (claims 14590, 14331), and its later references are restatements of that letter rather than independent confirmation. It was subsequently formalized through a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). Concentration on the supply side has moved from a single patent-sourcing relationship (IP Nav) to Block Maintain at Granby, and then to dependence on Beowulf for power at Hardin and Compute North for hosting 73,000 machines. Many entries are duplicates of the same filings, and claims 14311 and 14315 carry truncated dollar figures ($67, $500) that are extraction artifacts of $67 million and $500,000. No revenue figures appear in these claims, so the deal list cannot be reconciled against disclosed revenue. Any such test would have to run against hashrate and pool payouts rather than contract values.

Nothing in this record is a signed customer agreement in the sense that someone buys a product from Marathon and books revenue. Every named counterparty is a vendor, licensor, lender, landlord or service provider. The closest thing to a revenue source is ViaBTC, the mining pool paying Marathon's daily block rewards (claim 14888), and Cumberland DRW, its liquidation counterparty (claim 14889). Both are operational channels rather than contracted buyers. The two earliest items, UN Women (claim 20896) and EY (claim 20895), describe the Mara Foundation, an African entrepreneurship body with no apparent link to Marathon Patent Group. They look misattributed and should be removed rather than read as partnerships. To Marathon's credit, the file contains no unnamed 'strategic partners' or 'anchor customers'. Every counterparty is identified: Zhone and Siemens Technology (claims 21022, 21018), IP Nav (claim 21024), Securitas Canada, Block Maintain and Chord Advisors (claims 15198-15200), Beowulf, 2P1 and Liefern (claims 14791, 14756), Compute North, NYDIG and DMG. The language-quality issue is with Compute North. The 73,000-miner hosting commitment began as a 'binding letter of intent' (claim 14590), and the March 2022 filing still described it that way (claims 14304, 14331). A September 2021 master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347) then formalized parts of it. The record does not show whether the LOI itself was ever replaced by a definitive hosting contract. Counterparty concentration has moved through three phases. The patent era depended on IP Nav for deal flow. In 2019 the business was a single Quebec site run entirely by outsourced vendors. From 2020 to 2022, operating capacity concentrated in two counterparties: Beowulf for power, land and operations at Hardin (claims 14791, 14802, 14649), and Compute North for hosting. The many restatements of the same Beowulf, Securitas and Compute North terms (for example claims 14639, 14647 and 14791) are repeat disclosures, not new deals. No revenue arithmetic can be reconciled from these claims, because none states revenue. The dollar figures are all costs or consideration: $11.22 million in equity for operations services (claim 14640), the Compute North deposit, the $500,000 DMG license (claim 14305), and NYDIG's 3% BTC loan (claim 14316). The labels '$67' and '$500' on claims 14311 and 14315 appear to drop units and should not be cited as figures.

Marathon's record contains almost no customer agreements in the conventional sense. Nearly every named counterparty is a supplier, licensor, financier or service vendor. As a bitcoin miner, the company's revenue comes from pool rewards, described as a daily reward from ViaBTC (14888), with liquidation through Cumberland DRW (14889). Neither is a contracted customer. The 2014 UN Women and EY items (20896, 20895) concern the Mara Foundation, which appears to be an unrelated entity, and should be treated as likely misattributed.

The signed instruments are fairly clear. They include the Siemens and Zhone patent licensing deals (21018, 21022), the H.C. Wainwright at-the-market agreement (15202), the Beowulf exclusive power purchase agreement and land lease (14791, 14801, 14802), the DMG Blockseer license (14305), and the NYDIG Master Securities Loan Agreement (14306). Vendor relationships with Securitas, Chord Advisors and Block Maintain (15198–15200) are described as operating arrangements rather than contracts with disclosed terms. The softer language clusters around Compute North. That relationship began as a 'hosting partner' reference (14861), became a binding letter of intent for 73,000 miners (14590), and was still described as a letter of intent in the March 2022 filing (14304, 14311, 14331). This persisted even though a master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347) had been disclosed by then. The 30% special-purpose-entity interest (15223) is a contingent waterfall, not a contract with a counterparty.

Every counterparty in the record is named, and none is described only as a 'strategic partner' or 'anchor customer.' Identification is inconsistent, however. The Hardin services agreement is attributed to 2P1 and Liefern (14756, 14738) and elsewhere to Beowulf and 2P1 (14792, 14649). The past-tense phrasing 'were t[o] provide' (14734) suggests the arrangement may have changed.

Counterparty concentration has shifted from patent sources, chiefly IP Nav (21024), to a single power and site provider in Beowulf. By 2021–22 it had moved to a far larger dependence on Compute North for hosting capacity.

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed. The $11.22 million equity consideration for 3,000,000 shares implies about $3.74 per share (14649). The $67 million tag on 14331 is not supported by its quoted text. The hosting terms in 14289 pair a $0.50 per machine monthly fee with $0.044 per kWh, and they cannot be tied to the $14.6 million deposit without the truncated schedule.

Marathon's record contains almost no customer agreements in the conventional sense, and the labelling overstates commercial traction. As a bitcoin miner, its revenue comes from block rewards paid through a pool rather than from contracted buyers, so most claims tagged as partnerships are actually vendor, financing or infrastructure arrangements. The two earliest entries, UN Women (20896) and EY (20895), concern the Mara Foundation, an unrelated African entity, and should be excluded entirely. The genuine signed instruments are mostly cost-side: the Zhone and Siemens licensing agreements tied to patent acquisitions (21022, 21018), the Beowulf power purchase agreement with a five-year term (14791, 14801), the land lease (14802), the Data Facility Services Agreement with Beowulf, 2P1 and Liefern (14756, 14738, 14649), the DMG Blockseer licence (14305), the NYDIG securities loan (14306), the Compute North master agreement (14333) and the H.C. Wainwright ATM facility (15202). The weakest language is the Compute North hosting arrangement. It was announced as a binding letter of intent (14590), and filings nearly a year later still describe it that way (14304, 14331), even though a joint venture (14347) and a master agreement (14333) have since been signed. Readers should treat the 73,000-miner figure as LOI-anchored capacity, not a closed definitive contract. To Marathon's credit, every counterparty is named, and no unidentified strategic partners or anchor customers appear. The claim count is inflated by repetition, however. Block Maintain, Securitas and Chord Advisors each appear three times (15198, 15068, 14859; 15199, 15069, 14860; 15200, 15070), and the Beowulf and DMG items recur across several 2021 and 2022 filings. Counterparty concentration has moved from patent sourcing dependent on IP Nav (21024) to a Granby vendor stack, and then to heavy reliance on two infrastructure providers. Beowulf supplies power, land and operations at Hardin, while Compute North supplies hosting and is a JV partner. On the revenue side, a single pool (ViaBTC, 14888) and one liquidity venue (Cumberland, 14889) give way to Marathon's own OFAC pool built on DMG technology. Revenue reconciliation is not possible from this record because every disclosed dollar figure is an outflow or a consideration paid. These include the $14.6 million deposit (14333), the $500,000 licence (14305), roughly $11.2 million in equity for services (14649) and the hosting rates (14289). None of the announced deals carries revenue to which reported results could be tied.

The first point about Marathon's record is that almost none of it describes customers. A bitcoin miner earns its revenue from block rewards routed through a pool, and the only revenue-side counterparty on file is ViaBTC (14888), with Cumberland DRW as the liquidation venue (14889). Everything else is supply-side, financing, or inbound licensing. The 2013 Zhone and Siemens 'licensing agreements' (21022, 21018) were patent acquisitions, IP Nav was a deal-flow supplier (21024), and the H.C. Wainwright agreement (15202) is an equity facility. The 2014 UN Women and EY items (20896, 20895) refer to the 'Mara Foundation,' which appears to be a different entity, so they should be treated as a probable entity-matching error rather than Marathon partnerships. On a positive note, the record contains no anonymous 'strategic partners' or 'anchor customers.' Every counterparty is named. The weaker language sits instead in instrument type. Compute North first appears in a binding letter of intent to host 73,000 miners (14590), and later filings still describe that LOI as the operative document (14304, 14331, 14311). It was supplemented by a master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347), so that relationship did progress past LOI. The Hardin facility rests on signed instruments: an exclusive power purchase agreement with Beowulf (14791, 14801), a lease (14802), and a services agreement. Filings describe the services counterparties inconsistently, as Liefern and 2P1 (14756, 14738) or as Beowulf and 2P1 (14649), and one filing's past-tense 'were to provide' (14734) suggests the arrangement may have been restructured. Counterparty concentration moved from a Granby vendor trio of Block Maintain, Chord and Securitas (15198–15200) to Beowulf at Hardin in 2020, and then to heavy dependence on Compute North for hosting in 2021–22. The dollar figures in this set cannot be reconciled with revenue, because they are costs. They include the $67 million hosting commitment, the $11.22 million equity payment for services (14640), and the $500,000 DMG license (14305). Several are duplicated across claims (14288, 14315, 14330 and 14290, 14316), and some carry truncated labels ('$67', '$500'). The only income-generating contracts are the NYDIG bitcoin loan, whose interest rate is truncated in the quote (14306), and the contingent 30% SPE patent interest (15223). Neither is quantified enough to test against reported revenue.

The record on Marathon Digital contains almost no customer agreements, signed or otherwise, and most of what is filed as partnership is vendor and financing plumbing. The two earliest items, the UN Women and EY tie-ups (claims 20896, 20895), belong to the Mara Foundation, an unrelated African entrepreneurship group, and should be disregarded as an entity-match error. The 2013 Siemens Technology and Zhone licensing agreements (claims 21018, 21022) are the closest thing to signed counterparty contracts from the patent-monetization era, and both were bundled with patent acquisitions rather than standalone revenue deals. IP Nav (claim 21024) was a deal-flow supplier, and the 30% waterfall interest in a patent special purpose entity (claim 15223) is a contingent residual, not a contract with a named payer.

After the pivot to bitcoin mining, every named counterparty is a supplier, host, lender or service provider. These are Block Maintain, Chord Advisors and Securitas at Granby (claims 15198, 15200, 15199, repeated in 15068-15070 and 14859-14860), Cumberland DRW as liquidity venue and ViaBTC as mining pool (claims 14889, 14888), and Beowulf, 2P1 and Liefern at Hardin (claims 14791, 14792, 14756, 14649). None of them is described anonymously as a strategic partner or anchor customer, which is to Marathon's credit. Language strength varies, however. The Beowulf power purchase agreement is executed for 30 MW, but the 100 MW Phase II is optional (claim 14791), and extensions are only 'as mutually agreed' (claim 14801). Compute North starts as a binding letter of intent for 73,000 miners and $67 million (claims 14590, 14331). It becomes a master agreement with a $14.6 million deposit in September 2021 (claim 14333) and then an 82/18 joint venture (claim 14347). That is a genuine progression from LOI to contract and equity, but it also deepens Marathon's dependence on a single host.

Concentration risk sits on the supply side. Hosting capacity moved from a single Quebec site to Hardin plus Compute North, and the claims disclose no other comparably sized host. The DMG Blockseer licence and NYDIG securities loan (claims 14288, 14306) add one technology licensor and one lender. Some line items are mislabelled in the extraction ('$500' in claim 14315, '$67' in claim 14311).

Revenue arithmetic cannot be reconciled from this record because it discloses no revenue figures, and mining rewards come from the protocol through ViaBTC rather than from contracted customers. The figures that can be checked are cost-side: the $11.22 million share issuance to Beowulf/2P1 (claim 14649), the $0.006/kWh operating cost (claim 14650) and Compute North's $0.044/kWh rate (claim 14289).

Marathon's record contains almost no customer agreements in the conventional sense. A bitcoin miner earns revenue from block rewards rather than contracted buyers, and nearly every named counterparty here is a supplier, host, lender or service vendor. The closest thing to customer-facing revenue is the mining-pool relationship with ViaBTC (claim 14888) and the liquidation channel through Cumberland DRW (claim 14889). Both are disclosed as operating arrangements, not contracts with stated value. In the earlier patent-licensing era, the Siemens Technology and Zhone licensing agreements (claims 21018, 21022) were signed transactions, but each was paired with acquiring the counterparty's portfolio. That makes them two-way deals rather than clean licensing revenue. The UN Women and EY items (claims 20896, 20895) concern the Mara Foundation, an apparently unrelated African entity, and should be treated as misattributed to MARA.

On the signed-versus-soft distinction, the most material instrument is Compute North. It began as a binding letter of intent to host 73,000 miners (claims 14590, 14304, 14331). It was followed by a master agreement with a $14.6 million deposit (claim 14333), stated hosting terms (claim 14289), and an 82/18 joint venture (claim 14347). Filings dated March 2022 still describe the arrangement through the LOI framing, so investors should not assume the full $67 million scope was ever executed as definitive contracts. The Beowulf power purchase agreement (claims 14791, 14801, 14639) is signed. However, its 100 MW Phase II is contingent, and the five-year term's extensions are only "as mutually agreed."

Counterparties are consistently named; there are no anonymous "strategic partners." The one opaque item is the unnamed special purpose entity holding a 30% patent waterfall interest (claim 15223). Concentration has migrated across eras. It moved from IP Nav as the dominant patent sourcer (claim 21024), to Block Maintain at Granby (claims 15198, 14859), to the Beowulf/2P1/Liefern cluster at Hardin (claims 14756, 14649), and finally to heavy dependence on Compute North.

Revenue arithmetic cannot be reconciled because no revenue figures are attached to any claim. The dollar amounts all represent outflows or financing: the $67 million hosting commitment, the $14.6 million deposit, the $11.22 million equity issuance to Beowulf and 2P1 (claim 14640), the $500,000 DMG license (claims 14288, 14330), and the NYDIG bitcoin loan (claim 14306). The "$67" and "$500" figures in claims 14311 and 14315 are unit-truncation artifacts and should be read as millions and thousands respectively.

Marathon's record contains almost no signed customer agreements in the conventional sense. Nearly every named counterparty is a supplier, vendor, financier or host that Marathon pays, not one that pays Marathon. The two 2014 items (claims 20896, 20895) describe UN Women and EY partnerships with the Mara Foundation, an African entrepreneurship body that appears unrelated to the Nasdaq issuer, and should be treated as an entity-matching error rather than commercial evidence. The patent-era licensing agreements with Zhone and Siemens Technology (claims 21022, 21018) are the closest to genuine counterparty contracts, but each was disclosed as the vehicle through which Marathon acquired a patent portfolio rather than as a recurring royalty stream. IP Navigation Group (claim 21024) is explicitly a supplier of deal flow. The 30% waterfall interest in a patent special purpose entity (claim 15223) is contingent and leaves the licensees unnamed. To its credit, the record contains no anonymous 'anchor customer' or 'strategic partner' language. Every operating counterparty is identified, including Block Maintain, Chord Advisors, Securitas Canada, Beowulf, 2P1, Liefern, Compute North, DMG and NYDIG. The 2019–2020 vendor claims (15198/15068/14859, 15199/15069/14860, 15200/15070) are the same three relationships repeated across filings and should not be counted as separate deals. Execution status varies. The Beowulf power purchase agreement (claims 14791, 14801), the adjacent land lease (claim 14802) and the data facility services agreement compensated with 3,000,000 shares valued at $11.22 million (claims 14649, 14640) are signed. The Compute North relationship began as a binding letter of intent to host 73,000 miners (claims 14590, 14331). It hardened into a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). Counterparty concentration therefore moved from a scattered patent-licensing base to heavy operational dependence on two infrastructure providers, first Beowulf at Hardin and then Compute North, with the latter carrying the larger fleet commitment. The revenue arithmetic cannot be reconciled from these announcements because none of the dollar figures is inbound revenue. The roughly $67 million attached to the Compute North letter of intent (claims 14311, 14331) is a hosting cost. The DMG Blockseer figure is a $500,000 license fee paid by Marathon (claims 14288, 14305), and claim 14315 mis-scales it as '$500'. Actual revenue derives from pool rewards via ViaBTC (claim 14888) and liquidation through Cumberland DRW (claim 14889), plus 3% bitcoin lending interest from NYDIG (claim 14316). None of these is quantified in the record.

Marathon's record contains almost no customer agreements in the conventional sense, and that is the first thing a reader should understand. As a bitcoin miner, the company's revenue comes from block rewards paid through a mining pool, ViaBTC (claim 14888), and from liquidating coins through Cumberland DRW (claim 14889). Nearly every other named counterparty is a supplier, host, landlord, licensor or lender. The two earliest items, the UN Women and EY partnerships (claims 20896, 20895), refer to the Mara Foundation, which appears to be an unrelated African entity, and should be treated as an entity-matching error rather than Marathon disclosures.

On the signed-versus-soft distinction, the record leans toward executed instruments. The Beowulf power purchase agreement, with a five-year term and 30 MW expandable to 100 MW (claims 14791, 14801), the land lease (claim 14802), the Data Facility Services Agreement with 2P1 and Liefern (claims 14756, 14738), the DMG Blockseer license (claims 14288, 14305), the NYDIG securities loan (claims 14306, 14290) and the Compute North master agreement (claim 14333) are all described as entered into. The notable soft item is the Compute North 'binding letter of intent' to host 73,000 miners (claims 14590, 14331). It is binding in name but still an LOI, and was later supplemented by an 82/18 joint venture (claim 14347). Every counterparty in the record is named. No anonymous 'strategic partner' or 'anchor customer' language appears.

Concentration has migrated with the business model. The 2013-14 patent-licensing era depended on Siemens, Zhone and IP Nav as a deal source (claims 21018, 21022, 21024). In 2019-20 operations centered on a single Granby site run by Block Maintain (claims 15198, 14859). In 2020-21 the company was concentrated in Beowulf and 2P1 at Hardin, and by 2021-22 the bulk of planned hash capacity sat with one host, Compute North (claims 14304, 14333). Many of these entries are refilings of the same agreements (for example, claims 14791, 14647 and 14639), which inflates the apparent number of relationships.

The revenue arithmetic cannot be reconciled from this record because no revenue figures are disclosed, and the dollar amounts attached to deals are cost-side. The $11.22 million tied to Beowulf and 2P1 is equity compensation for 3,000,000 shares (claims 14649, 14640). The $14.6 million is a hosting deposit (claim 14333), and the $500,000 is a license fee paid to DMG (claim 14330; the '$500' in claim 14315 is a truncation). The $67 million attached to the Compute North LOI (claims 14331, 14311) does not appear in any quoted text and should not be relied upon. Only the NYDIG loan, at three percent interest (claim 14316), implies income, and its principal is not given.

Marathon's record contains almost no customer agreements in the conventional sense. Nearly every named counterparty is a supplier, lender, licensor, or service provider. This matters because the claims can read like commercial traction when they describe cost and financing relationships. The two 2014 Mara Foundation items (claims 20896, 20895) appear to concern an unrelated African entity that shares the name, and should be excluded from MARA's record. In the patent-monetization era, the Siemens Technology and Zhone 'licensing agreements' (21018, 21022) were tied to patent portfolio acquisitions by Marathon, and IP Nav is explicitly described as a supplier of deal flow (21024). None of these establishes a paying licensee. The 2019 waterfall interest (15223) is contingent: a 30% stake in a special purpose entity that could share in net proceeds.

After the pivot to mining, the counterparties are all named, and none is hidden behind 'strategic partner' or 'anchor customer' language. The arrangements still sit on the cost side. Block Maintain, Securitas Canada and Chord Advisors are operating vendors, repeated almost verbatim across 15198–15200, 15068–15070 and 14859–14860. ViaBTC (14888) is the mining pool that pays block rewards, and Cumberland DRW (14889) is a liquidation venue. Neither is a customer under contract. The Beowulf power purchase agreement (14791, 14801, 14639, 14647) and the Hardin services agreement with 2P1 and Liefern (14756, 14738) are signed. However, the April 2021 phrasing that 2P1 and Liefern 'were to provide' services (14734) suggests that arrangement changed. Compute North moved from a binding LOI (14590) to a master agreement with a $14.6 million deposit (14333) and then an 82/18 joint venture (14347). That is a genuine progression from LOI to definitive terms.

Concentration moved from a single deal-flow source, IP Nav, to heavy reliance on two infrastructure counterparties. Beowulf supplies power and land at Hardin, and Compute North hosts 73,000 miners. The vendor concentration is high, but no customer concentration can be measured.

The revenue arithmetic cannot be reconciled from these claims, because none states revenue. The dollar figures are outflows or financing: the $67 million Compute North commitment (14331), the $500,000 DMG license fee paid by Marathon (14288), and $11.22 million of equity issued for services (14649). The NYDIG loan (14306) and the H.C. Wainwright ATM (15202) are also financing. Claims 14311 and 14315 record these as '$67' and '$500', which are unit-extraction errors and should be corrected before publication.

Marathon's record contains no signed customer agreements in the conventional sense. Every named counterparty is a supplier, service provider, lender, or financing agent, and readers should not count any of them as revenue relationships. The two 2014 items, UN Women (claim 20896) and EY (claim 20895), refer to the Mara Foundation, an apparently unrelated African entity, and look like entity-resolution errors rather than Marathon disclosures. In the patent-monetization era, the Zhone (21022) and Siemens Technology (21018) licensing agreements were vehicles for acquiring patent portfolios, and IP Nav (21024) is explicitly described as a supplier of acquisition opportunities. The 2019 Wainwright ATM agreement (15202) is a capital-raising arrangement. Block Maintain, Chord Advisors, and Securitas (15198-15200, repeated in 15068-15070 and 14859-14860) are outsourced operators. To its credit, the company names every counterparty and does not lean on unidentified strategic partners or anchor customers. The softer language sits elsewhere. The Compute North hosting arrangement began as a binding letter of intent for 73,000 miners (14590) and only later firmed into a master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347). Beowulf, by contrast, was a signed exclusive power purchase agreement from the outset (14791, 14801). Concentration has shifted from operator dependence on a single Granby site toward heavy reliance on two infrastructure counterparties, Beowulf/2P1 at Hardin (14649, 14756) and Compute North for the bulk of fleet hosting. On the revenue side, the only counterparty that actually pays Marathon in the mining model is the pool, ViaBTC (14888), with Cumberland DRW as the liquidation venue (14889). That makes payment flow concentrated in a single pool relationship. The revenue arithmetic cannot be reconciled from this record, because none of the dollar figures attached to claims represent inbound revenue. The $67 million Compute North figure (14311, 14331), the $11.22 million equity issued to Beowulf and 2P1 (14640), and the $500,000 DMG Blockseer license (14288) are all outflows or commitments. The NYDIG loan (14290) yields interest income, but it is not quantified here. Heavy duplication across the March 2022 claims also inflates the apparent volume of deal activity.

None of the forty-odd claims on record for Marathon Digital describes a customer agreement in the ordinary sense. Every counterparty is a supplier, host, service provider, financier or licensor whose payments flow away from Marathon. Some items should not be attributed to the company at all. The 2014 UN Women and EY announcements (claims 20896, 20895) concern the Mara Foundation, an unrelated African entity, and look like a name-collision error in the record. The patent-era agreements with Siemens Technology and Zhone (21018, 21022) were licensing deals tied to Marathon buying patent portfolios, and IP Nav (21024) is explicitly described as a supplier of patent opportunities rather than a buyer.

In the mining era, the record separates cleanly into signed contracts and softer instruments. The Beowulf power purchase agreement (14791, 14801) is executed, with a five-year term. So are the Data Facility Services Agreement with Beowulf, 2P1 and Liefern (14756, 14738, 14649), the DMG Blockseer licence (14288), the NYDIG securities loan (14306) and the September 2021 Compute North master agreement carrying a $14.6 million deposit (14333). Compute North's 73,000-miner arrangement, however, first appears as a binding letter of intent (14590). It is restated in the March 2022 filings as an LOI rather than a definitive hosting contract (14304, 14331), even as the related joint venture was formed (14347). Every counterparty is named. The record contains no unidentified 'strategic partners' or 'anchor customers.'

Concentration has narrowed over time. The company moved from relying on IP Nav for deal flow, to a single-vendor stack at Granby (Block Maintain, Chord Advisors and Securitas; 15198, 15200, 15199), and then to heavy dependence on two parties, Beowulf for power and Compute North for hosting. Its revenue has effectively come from a single mining pool, ViaBTC (14888), with liquidation through Cumberland DRW (14889).

The revenue arithmetic cannot be reconciled because the record contains no revenue figures at all. Each disclosed dollar amount is a cost: the $67 million hosting commitment, the $14.6 million deposit, the $500,000 licence and the $11.22 million share-based service fee. Claims 14311 ('$67') and 14315 ('$500') appear to be unit-truncation errors. Repeated restatements, such as 14639, 14647 and 14791, also inflate the apparent deal count and should be deduplicated before any tally.

Marathon's record contains almost no customer agreements in the conventional sense, and the counterparties it does name are overwhelmingly suppliers, service providers and financing parties. The two earliest items (claims 20896, 20895) describe UN Women and EY partnerships with the 'Mara Foundation', an African entrepreneurship vehicle with no evident connection to Marathon Patent Group, and should be treated as misattributed rather than as evidence of commercial traction. The patent-era record is more substantive. Licensing agreements with Siemens Technology (claim 21018) and Zhone (claim 21022) were closed and executed, though each was paired with a portfolio acquisition, which makes them closer to structured asset purchases than arm's-length licensing revenue. IP Navigation Group is described as a key supplier of deal flow (claim 21024), not a customer. The 30% waterfall interest in a patent special purpose entity (claim 15223) is contingent, not contracted income.

After the pivot to bitcoin mining, every named counterparty is identified by name, and the filings do not rely on unidentified 'strategic partners' or 'anchor customers'. The language does need to be read carefully, however. Compute North first appears as a hosting partner (claim 14861) and then under a 'binding letter of intent' for 73,000 miners (claims 14590, 14304, 14331). That LOI was later formalized through a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). The Beowulf power purchase agreement (claims 14791, 14801) and the data facility services agreement with 2P1 and Liefern (claims 14756, 14738, 14649) are signed contracts, but they represent costs paid by Marathon, partly through 3,000,000 shares valued at $11.22 million.

Counterparty concentration has moved from a single facility in Granby, operated by Block Maintain with Securitas providing security (claims 15198, 15199), to a heavy dependence on one hosting provider, Compute North, alongside a single mining pool, ViaBTC (claim 14888), and a single liquidity venue, Cumberland DRW (claim 14889). Revenue reconciliation is not possible from this record because no revenue figures are disclosed. The largest dollar amounts, roughly $67 million for Compute North and $500,000 for the DMG Blockseer license, are outflows. Readers should also note extraction errors, since claims 14311 and 14315 render these amounts as '$67' and '$500'. Mining revenue comes from block rewards, not contracted customers, so none of these deals should be read as booked demand.

Marathon's record contains almost no customer agreements in the conventional sense, and that is the main finding. In the patent-licensing era, the only counterparties that resemble revenue sources are Siemens Technology (claim 21018) and Zhone (claim 21022). Both are described as licensing agreements tied to patent-portfolio acquisitions, so they are closer to settlement-and-acquire structures than to recurring customer contracts. IP Navigation Group (claim 21024) was a supplier of deal flow. The 2014 UN Women and EY items (claims 20896, 20895) refer to the 'Mara Foundation,' which appears to be a different, Africa-focused entity. They should be treated as likely misattributions rather than Marathon partnerships. After the pivot to bitcoin mining, the economic counterparty for revenue is a mining pool, ViaBTC (claim 14888), plus a liquidity venue, Cumberland DRW (claim 14889). Every other named relationship is a vendor, host, power supplier or financing counterparty. These include Block Maintain, Securitas and Chord Advisors (claims 15198-15200, repeated in 15068-15070 and 14859-14860), H.C. Wainwright (claim 15202), and NYDIG, to which Marathon lends its bitcoin (claims 14290, 14306, 14316). On firmness, the Beowulf power purchase agreement is executed, with a five-year term and extensions only 'as mutually agreed' (claims 14791, 14801). The DMG Blockseer license is signed at $500,000; the '$500' in claim 14315 is a truncation (claims 14288, 14305, 14330). The weakest instrument is the Compute North arrangement. It began as a 'binding letter of intent' for 73,000 miners (claims 14590, 14331, 14304, 14311), and only later became a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). Notably, no counterparty is ever left unnamed as a 'strategic partner' or 'anchor customer.' Disclosure is specific, but it is also heavily duplicated across filings. Concentration has moved from patent licensees and one deal-sourcing intermediary toward heavy operational dependence on a few infrastructure parties: Beowulf, 2P1 and Liefern at Hardin (claims 14756, 14738, 14649), and Compute North for hosting. The arithmetic cannot be reconciled from this record because no revenue figures are disclosed alongside the deals. The dollar amounts that do appear are costs rather than revenue: the $67 million hosting commitment, the $14.6 million deposit, the $11.22 million equity payment to Beowulf and 2P1 (claim 14640), and the hosting and operating rates (claims 14289, 14650). Revenue depends on hash rate and bitcoin price through the pool, not on any announced contract.

Marathon's record contains almost no customer agreements in the conventional sense, and readers should not treat its named counterparties as revenue relationships. The two earliest items, the UN Women and EY partnerships (claims 20896, 20895), refer to the Mara Foundation, an African entrepreneurship group. They appear to be a name collision rather than anything Marathon signed, and should be excluded. The patent-era entries describe Marathon as a buyer, not a seller. It acquired portfolios in connection with licensing agreements with Siemens Technology and Zhone (claims 21018, 21022) and sourced deal flow from supplier IP Nav (claim 21024). The 30% special-purpose-entity interest (claim 15223) is a contingent waterfall, not a contracted customer.

After the pivot to mining, every counterparty is identified by name, and none is hidden behind an anonymous 'strategic partner' or 'anchor customer' label. Nearly all of them, however, are vendors, financiers or liquidity venues. Block Maintain, Chord Advisors and Securitas are service providers, and they recur across filings as restatements rather than new deals (claims 15198/15068/14859, 15200/15070, 15199/15069/14860). The H.C. Wainwright ATM (claim 15202) and the NYDIG securities loan (claims 14306, 14316, 14290) are financing arrangements. Cumberland DRW is a trading venue (claim 14889). The closest thing to a revenue source is the ViaBTC mining pool (claim 14888), which pays protocol rewards rather than contracted fees.

The firmest commitments are on the cost side. The Beowulf power purchase agreement is described as exclusive, with a five-year term and defined capacity of 30 MW up to 100 MW (claims 14791, 14801, 14647). The related services agreement with 2P1, Liefern and Beowulf was paid in 3,000,000 shares valued at $11.22 million (claims 14649, 14640, 14738). The Compute North relationship shows the clearest progression from soft to hard language. It began as a 'binding letter of intent' for 73,000 miners (claims 14590, 14331) and became a master agreement with a $14.6 million deposit (claim 14333), then an 82/18 joint venture (claim 14347). The result is heavy operational concentration in one hosting supplier and one power supplier.

The revenue arithmetic cannot be reconciled from this record. No disclosed revenue figure ties to any named contract. The $67 million tag attached to the Compute North claims (claims 14331, 14311) does not appear in the quoted text. The DMG license is recorded as both $500 and $500,000 (claims 14315, 14305), and its March 2021 date surfaces only in a 2022 filing (claim 14288). Both inconsistencies warrant source verification before the figures are relied on.

Marathon's record contains almost no customer agreements in the conventional sense, and reading it as a customer pipeline would mislead. Nearly every claim names a supplier, financier, or service vendor rather than a buyer of Marathon's output. The two earliest entries are a likely entity mismatch: the UN Women and EY partnerships (claims 20896, 20895) are with the Mara Foundation, an African entrepreneurship body, not Marathon Patent Group, and should be excluded. The patent-era licenses with Siemens Technology and Zhone (21018, 21022) were signed, but they were tied to Marathon acquiring those companies' patent portfolios. IP Nav (21024) is described as a deal-flow supplier, and the 30% special-purpose-entity interest (15223) is a contingent waterfall, not contracted revenue.

The mining era is similarly vendor-heavy, though the counterparties are consistently named rather than hidden behind 'strategic partner' language. Block Maintain, Securitas Canada, and Chord Advisors (15198-15200, repeated in 15068-15070 and 14859-14860) provide operations, security, and finance. Beowulf's exclusive 30 MW to 100 MW power purchase agreement (14791, 14801) and land lease (14802) are executed supply contracts. The 2P1/Liefern services agreement (14756, 14738) was paid in 3,000,000 shares valued at $11.22 million (14649, 14640). The April 2021 wording that those parties 'were to provide' services (14734) suggests the arrangement was later altered. The H.C. Wainwright ATM (15202) is financing, and the NYDIG securities loan (14290, 14316) is a lending arrangement earning interest.

The one instrument whose label needs scrutiny is Compute North. It began as a 'binding letter of intent' to host 73,000 miners, valued at $67 million (14590, 14331, 14311). It was then documented as a master agreement with a $14.6 million deposit (14333) and a 82/18 joint venture (14347). That chain is a hosting commitment Marathon pays for, not revenue it receives. The DMG Blockseer license (14330, 14305, 14288) is likewise an outflow of $500,000. Claim 14315's '$500' figure appears to be an extraction error.

Concentration is therefore a supplier and treasury risk rather than a customer risk. Block reward income flows through a single pool, ViaBTC (14888), and liquidation runs through a single desk, Cumberland DRW (14889). Hosting capacity sits heavily with Compute North. No revenue figures appear in this record, so the deal arithmetic cannot be reconciled against reported revenue. Readers should also note that heavy duplication across the March 2022 filings inflates the apparent number of distinct agreements.

Marathon's record contains no customer agreements in the conventional sense. Nearly every named counterparty is a supplier, lender, licensor or service provider, and none is a buyer of Marathon's output. Two 2014 items should be excluded outright. The UN Women and EY partnerships (claims 20896, 20895) name the "Mara Foundation," an unrelated African entity, and appear to be an entity-matching error rather than Marathon disclosures. The patent-era entries are cost-side. The Zhone and Siemens licensing agreements (21022, 21018) were vehicles through which Marathon acquired portfolios, and IP Nav is described explicitly as a supplier of patent opportunities (21024). The company leaned heavily on that single source, which was the first concentration point in the record. The 2019 SPE item (15223) is a 30% residual interest, not a customer. In the mining era, Marathon's revenue comes from a pool payout. ViaBTC pays the daily reward and Cumberland DRW is the liquidation venue (14888, 14889), so there is no customer book to name. Every counterparty is identified; nowhere does the record lean on unnamed 'strategic partners' or 'anchor customers.' The language-quality issue sits elsewhere. The Compute North relationship began as a 'binding letter of intent' for 73,000 miners (14590, 14304, 14331). It became a master agreement with a $14.6 million deposit only in September 2021 (14333) and then an 82/18 joint venture (14347). Readers should date the firm commitment from 14333, not from the LOI. Counterparty concentration moved in sequence. It ran from IP Nav, to Block Maintain as sole operator at Granby (15198, 15068, 14859), to Beowulf and 2P1 at Hardin, and finally to Compute North. Beowulf combined power supply, land and O&M in one party (14791, 14802, 14649), and Compute North carried the bulk of planned hashrate. Repeated filings of identical language (for example, the Beowulf PPA in 14791, 14647 and 14639) inflate the apparent deal count. The record supports no revenue reconciliation, because none of these claims discloses revenue. The figures that can be checked are cost or financing items. The Beowulf equity consideration of $11.22 million for 3,000,000 shares implies roughly $3.74 per share (14649, 14640). The NYDIG bitcoin loan (14306, 14316) produces interest income, not operating revenue. Several amounts are recorded inconsistently. DMG appears as both $500 and $500,000 (14315 versus 14288), and Compute North as $67 versus $67,000,000 (14311 versus 14331). These units should be normalised before any figures are published.

Marathon's record contains almost no customer agreements in the conventional sense, and diligence should read it that way. As a bitcoin miner, its revenue comes from block rewards paid through a pool, ViaBTC (claim 14888), and liquidation through a trading counterparty, Cumberland DRW (claim 14889). Nearly everything else on file is a supplier, financing or infrastructure contract. The two 2014 items, UN Women (claim 20896) and EY (claim 20895), refer to the Mara Foundation, an apparently unrelated African entity, and should be treated as entity-resolution errors rather than Marathon partnerships. The patent-era licensing deals with Siemens Technology (claim 21018) and Zhone (claim 21022) were signed, but they functioned as portfolio acquisitions rather than revenue contracts. IP Nav (claim 21024) is a deal-flow supplier.

On signed versus provisional language, the Beowulf power purchase agreement is executed. It is exclusive, 30 MW expanding to 100 MW, with a five-year term (claims 14791, 14801, 14639, 14647). The Data Facility Services Agreement is likewise dated October 6, 2020 (claims 14738, 14734). The filings disagree on who the service providers are: claims 14756 and 14738 name 2P1 and Liefern LLC, while claims 14792 and 14649 name Beowulf and 2P1.

The Compute North relationship escalated in three steps. It began as a binding letter of intent for 73,000 miners (claims 14590, 14304, 14331). It then became a master agreement with a $14.6 million deposit (claim 14333), and finally an 82/18 joint venture (claim 14347). Only the later instruments are definitive.

The one fully unnamed counterparty is the special purpose entity holding the '798 patent, in which Marathon owns 30% (claim 15223). The H.C. Wainwright ATM (claim 15202) is a capital-raising arrangement, not a commercial relationship.

Counterparty concentration has moved from Granby-era vendors (Block Maintain, Securitas, Chord; claims 15198 to 15200) to heavy dependence on a single host, Compute North, by 2021 to 2022. That is an operational exposure rather than revenue concentration.

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed and the dollar amounts are costs rather than income. The NYDIG bitcoin loan at 3% (claims 14316, 14306) is the only income-bearing contract. Several extracted values are corrupted: '$500' in claim 14315 against $500,000 in claim 14305, and '$67' in claim 14311 against $67 million in claim 14331. The $11.22 million equity payment for 3,000,000 shares implies about $3.74 per share (claim 14649). The $0.50 per machine monthly hosting fee (claim 14289) also merits checking against the source filing.

Marathon's record contains almost no customer agreements in the conventional sense. Nearly every named counterparty is a supplier, licensor, financier or service provider, which is consistent with a business that moved from patent monetization to bitcoin mining, where the effective revenue source is the mining pool (ViaBTC, claim 14888) and the liquidity venue (Cumberland DRW, claim 14889) rather than identifiable buyers. The two earliest entries, the UN Women and EY partnerships (claims 20896, 20895), refer to the Mara Foundation, an unrelated African entrepreneurship group. They appear to be entity-matching errors and should not be attributed to MARA. Among genuine counterparties, the firmly executed instruments are the Zhone and Siemens patent licensing agreements (claims 21022, 21018), the H.C. Wainwright ATM (claim 15202), the Beowulf power purchase agreement with a defined five-year term (claims 14791, 14801), the DMG Blockseer license (claim 14305), the NYDIG master securities loan agreement (claim 14345), and the Compute North master agreement with its $14.6 million deposit (claim 14333). The Compute North relationship began as a binding letter of intent for 73,000 miners (claim 14590). That LOI language persisted in filings as late as March 2022 (claims 14304, 14331) even after a master agreement and an 82/18 joint venture were signed (claim 14347), so readers should not treat the LOI and the definitive contracts as separate deals. The Hardin services agreement with 2P1 and Liefern is described in past-conditional terms by April 2021, as services that the parties "were to" provide (claim 14734), which suggests it had been restructured or terminated. The only unidentified counterparty is the patent special purpose entity in which Marathon held 30% (claim 15223). Counterparty concentration has shifted sharply over time. The company first depended on IP Nav for deal flow (claim 24) [sic: claim 21024], then on a Granby-era vendor stack of Block Maintain, Securitas and Chord Advisors (claims 15198-15200), then on Beowulf at Hardin, and finally on Compute North as the hosting provider for essentially the entire planned fleet. The record contains no disclosed revenue figures, so deal values cannot be reconciled against reported revenue. The dollar figures that do appear are costs or consideration rather than revenue: $11.22 million in stock for 3,000,000 shares (claim 14649, implying about $3.74 per share), a hosting rate of $0.044 per kWh (claim 14289), and a $67 million label on claim 14331 that the truncated quote does not support. Heavy duplication across filings inflates the apparent claim count, and the input list was cut off mid-record, so later claims were not assessed.

Marathon's record contains almost no customer agreements in the conventional sense, because for most of its history the company has not sold to customers. The patent-era filings name two real revenue counterparties, Siemens Technology (claim 21018) and Zhone (claim 21022). Both were signed licensing agreements, and each was paired with a portfolio acquisition. IP Navigation Group (claim 21024) is described as a supplier of deal flow, not a buyer. The two 2014 items crediting UN Women and EY (claims 20896, 20895) refer to the 'Mara Foundation.' They look like an entity-resolution error and should not count toward Marathon's commercial history. After the 2019 pivot to mining, the named counterparties are vendors and service providers rather than customers. These include Block Maintain, Chord Advisors and Securitas (claims 15198-15200, 14859-14860), Cumberland DRW as a liquidity venue (claim 14889), and NYDIG as a borrower of Marathon's bitcoin (claims 14345, 14290). The only revenue-adjacent counterparty is the mining pool ViaBTC (claim 14888). To Marathon's credit, every counterparty in the record is identified by name, and no unnamed 'strategic partners' or 'anchor customers' appear. The label discipline is weaker, however. The Compute North hosting arrangement for 73,000 miners began as a 'binding letter of intent' (claim 14590). The March 2022 filing still describes it that way (claims 14331, 14304) even though claim 14311 is tagged as a 'hosting agreement.' The relationship firmed up through a September 2021 master agreement with a $14.6 million deposit (claim 14333), stated hosting rates (claim 14289) and an 82/18 joint venture (claim 14347). Beowulf is the cleanest signed relationship. It covers an exclusive power purchase agreement of 30-100 MW with a five-year term (claims 14791, 14801), a land lease (claim 14802), and an operations agreement with 2P1 and Liefern paid in 3,000,000 shares valued at $11.22 million (claims 14649, 14738). Concentration therefore moved from a few patent licensees to dependence on two suppliers: Beowulf for power at Hardin and Compute North for most of the planned fleet. Revenue arithmetic cannot be reconciled from this record because it discloses no revenue figures. The DMG Blockseer license is also inconsistent, recorded at $500,000 in claims 14305 and 14330 but $500 in claim 14315. The claim list provided is also truncated after claim 14289.

Marathon's record contains almost no customer agreements in the conventional sense, and that is the most important finding. As a bitcoin miner, its revenue comes from block rewards paid through a mining pool, so the named counterparties are mostly suppliers, hosts, financiers and service vendors rather than buyers. The one revenue-side relationship disclosed is ViaBTC, described only as the pool "to which we point our miners" (claim 14888), alongside Cumberland DRW as a liquidation venue (claim 14889). Neither is a signed offtake contract.

The 2014 entries should be set aside. The UN Women and EY partnerships (claims 20896, 20895) belong to the Mara Foundation, an unrelated African entrepreneurship organisation, and appear to be name-match misattributions. The patent-era items are real but are acquisition-side relationships: the Siemens and Zhone licensing agreements were the vehicles through which Marathon bought portfolios (claims 21018, 21022), and IP Nav was a supplier of deal flow (claim 21024). Licensees who paid Marathon are never named.

From 2020 the signed instruments are identifiable and specific. These are the exclusive Beowulf power purchase agreement for 30 MW, expandable to 100 MW, over a five-year term (claims 14791, 14801), the Data Facility Services Agreement with 2P1 and Liefern paid in 3,000,000 shares valued at $11.22 million (claims 14756, 14649), the DMG Blockseer license at $500,000 (claim 14305; claim 14315's "$500" is a parsing error), and the NYDIG securities loan (claim 14290). The Compute North arrangement needs care. Its May 2021 announcement was a "binding letter of intent" for 73,000 miners at roughly $67 million (claims 14590, 14331), not a definitive contract, though it was later supplemented by an 82/18 joint venture (claim 14347).

Concentration has therefore evolved on the supply side, not the customer side. Exposure has moved from a single Granby operator, Block Maintain (claim 15198), to heavy dependence on two counterparties, Beowulf for Hardin power and Compute North for most planned hashrate hosting. Notably, the record contains no unnamed "anchor customer" or "strategic partner" language; every counterparty is identified.

The revenue arithmetic cannot be reconciled from this record. Every dollar figure disclosed is a cost or outflow: equity compensation, license fees, hosting prepayments and the $0.044/kWh hosting rate (claim 14289). No revenue figure ties to any named deal. Many entries are also duplicates of the same 2022 filing (for example claims 14288, 14305 and 14330), and the record is truncated at claim 14333.

Marathon's record contains no signed customer agreements, because Marathon has no customers in the conventional sense. Mining revenue arrives as a daily pool reward from ViaBTC (claim 14888), and bitcoin is liquidated through Cumberland DRW (claim 14889). Nearly every other named counterparty is a supplier, financier or vendor, so this file is a map of supply-chain dependency rather than demand. Two early entries, the UN Women and EY tie-ups with the "Mara Foundation" (claims 20896, 20895), appear to concern an unrelated African organisation and should be treated as misattributed. The 2013 patent-era deals with Siemens and Zhone (claims 21018, 21022) were licensing arrangements that coincided with patent acquisitions, and IP Nav (claim 21024) was a deal-flow supplier. None of these was a revenue customer.

To its credit, Marathon names its counterparties. There is no anonymous "strategic partner" or "anchor customer" language. The real distinction is between binding and preliminary instruments. The Beowulf power purchase agreement (claims 14791, 14801) and the Data Facility Services Agreement with 2P1 and Liefern (claims 14756, 14738, 14734) are executed contracts. The Compute North relationship began as a binding letter of intent to host 73,000 miners (claim 14590). It firmed into a master agreement carrying a $14.6 million deposit (claim 14333) and later an 82/18 joint venture (claim 14347). The NYDIG loan (claim 14306) and the DMG Blockseer license (claim 14305) are signed agreements.

Counterparty concentration has tightened over time. In 2019 the Granby operation depended on one operator, Block Maintain (claims 15198, 15068). From 2020 to 2021 the company shifted to a single power-and-site sponsor at Hardin, Beowulf. By 2022 the bulk of the planned fleet was committed to one hosting provider, Compute North, at $0.044/kWh under a three-year term (claim 14289). Any operational failure at that host would therefore strand most of the installed capacity.

The revenue arithmetic cannot be reconciled from this record. No revenue, hashrate or pool-output figures appear against which the hosting commitments could be tested. Some claim titles also overstate the underlying quotes. Claims 14311 and 14315 carry "$67" and "$500" where sibling claims show $67,000,000 and $500,000, and the $67 million figure itself does not appear in any quoted excerpt. The $11.22 million services consideration (claims 14649, 14640) implies about $3.74 per share for the 3,000,000 shares issued, which is equity dilution rather than cash cost. Heavy duplication across the 2022 filings also inflates the apparent deal count, and the source list ends mid-record.

Nothing in Marathon's record is a customer agreement in the conventional sense. Every named counterparty is a supplier, host, lender, licensor or service vendor, so the question is less which deals are signed than which way the money flows. Two early entries should be set aside entirely. The UN Women and EY partnerships (claims 20896, 20895) concern the Mara Foundation, an unrelated African entrepreneurship body, and look like an entity-matching error rather than Marathon disclosures.

In the patent-licensing era, the Siemens and Zhone licensing agreements (21018, 21022) were the vehicles through which Marathon acquired portfolios. They were not revenue licensees, and IP Nav (21024) is described as a deal-sourcing supplier. In the mining era, the documents are largely executed contracts: the Beowulf power purchase agreement with defined MW tranches and term (14791, 14801), the land lease (14802), the Data Facility Services Agreement with 2P1 and Liefern (14756, 14738), the NYDIG loan (14345), and the DMG Blockseer licence (14305). The softest instrument is the Compute North arrangement. It began as a 'binding letter of intent' to host 73,000 miners (14590) and was later papered as a master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347). The $67 million attached to it in claim titles (14331, 14311) does not appear in any quoted text and should not be treated as a disclosed contract value.

All counterparties are identified by name. The record contains no unnamed 'strategic partner' or 'anchor customer' language, which is to the company's credit. Concentration has moved from a Quebec vendor cluster at Granby (Block Maintain, Securitas, Chord; 15198-15200) to a Hardin cluster built around Beowulf and 2P1. It then shifted to dependence on Compute North for hosting at scale. The apparent breadth of the record is inflated by repetition. The Beowulf PPA alone appears in 14791, 14647 and 14639, and the 3,000,000-share, $11.22 million services consideration appears in both 14649 and 14640.

Revenue cannot be reconciled against these deals because none of them generate revenue. Mining income arrives as pool rewards from ViaBTC (14888) and is realised through Cumberland DRW (14889). Neither relationship carries disclosed contract value. The dollar figures on record are costs or financing: hosting rates (14289), facility opex of $0.006/kWh (14650), equity paid to vendors, and the DMG fee. The claim list supplied here is truncated after the March 2022 filings, so later counterparties are not assessed.

Marathon's record contains no signed customer agreements in the conventional sense, because a bitcoin miner sells into a protocol rather than to buyers. Nearly every named counterparty is a supplier, financier or service provider. The closest thing to a revenue counterparty is ViaBTC, the pool that pays Marathon's daily block reward (claim 14888). Cumberland DRW, the liquidity desk for bitcoin sales, is the only other (claim 14889). The 2014 UN Women and EY items (claims 20896, 20895) concern the Mara Foundation, an unrelated African entrepreneurship body. They look like an entity-resolution error and should not be attributed to MARA. The patent-era licences with Siemens Technology and Zhone (claims 21018, 21022) were really patent acquisitions, and IP Nav was a deal-flow source (claim 21024), not a licensee.

On naming, the record is unusually clean. Every counterparty is identified by name, and none are described as unnamed 'strategic partners' or 'anchor customers.' The softer language sits elsewhere. The Compute North hosting deal for 73,000 miners began as a 'binding letter of intent' dated May 21, 2021 (claims 14590, 14304, 14331). It hardened into a master agreement with a $14.6 million deposit on September 3, 2021 (claim 14333) and then a joint venture split 82/18 (claim 14347). By contrast, the Beowulf power purchase agreement (claims 14791, 14639) and the 2P1/Liefern services agreement dated October 6, 2020 (claims 14738, 14734) are executed contracts.

Concentration has moved through three stages. It started with patent sellers, then a small Quebec vendor stack of Block Maintain, Securitas and Chord Advisors (claims 15198-15200, repeated in 15068-15070 and 14859-14860). By 2021-22 it had settled on two dependencies: Beowulf for power and Compute North for hosting. Many claims are restatements of the same deal across filings, such as NYDIG in claims 14316, 14345, 14306 and 14290. They should not be counted as separate relationships.

The revenue arithmetic cannot be reconciled from this record. No claim discloses revenue, and the economics captured are almost entirely on the cost side: $0.044/kWh hosting (claim 14289), $0.006/kWh operations (claim 14650), and 3,000,000 shares valued at $11.22 million (claim 14649). Claim 14331 is labelled $67 million, but its quote does not contain that figure. Claim 14311 shows '$67', and claim 14315 shows '$500' against $500,000 in claims 14305 and 14288. These are extraction unit errors that need fixing before publication. The supplied record also ends mid-entry after claim 14315, so later contracts are not assessed here.

Marathon's record contains few customer agreements in the ordinary sense, and the claim count overstates how many distinct relationships exist. Most entries are suppliers, hosts, lenders or financing counterparties, and many are the same disclosure repeated across filings. The Granby vendor trio of Block Maintain, Chord Advisors and Securitas Canada appears three times (15198/15068/14859, 15200/15070, 15199/15069/14860). The Beowulf power purchase agreement recurs as 14791, 14647 and 14639, and the NYDIG loan as 14345, 14306, 14316 and 14290. The two earliest items, UN Women (20896) and EY (20895), concern the Mara Foundation, an African entrepreneurship body, and look like entity-resolution errors rather than Marathon relationships. They should be excluded.

What is signed is clear. The patent-era licensing agreements with Siemens (21018) and Zhone (21022) were executed and came with portfolio acquisitions. The Beowulf PPA is described as an exclusive, executed agreement for 30 MW, expandable to 100 MW (14791), with a five-year initial term (14801). The DMG Blockseer licence (14305, 14330) and the NYDIG Master Securities Loan Agreement (14345) are also signed. By contrast, the largest operational commitment, hosting 73,000 miners with Compute North, began as a "binding letter of intent" (14590, 14304, 14331). It was only later papered as a master agreement with a $14.6 million deposit (14333) and a joint venture split 82/18 (14347). The counterparty set on the services contract is inconsistent. Some filings name Beowulf and 2P1 (14792, 14649), while others name 2P1 and Liefern (14738, 14756). The past-tense "were t[o]" in 14734 hints the arrangement may have changed. Every counterparty in the record is named; none is described only as an unidentified "strategic partner."

Concentration moved from a patent pipeline leaning heavily on IP Nav (21024), to a single Quebec site run by one operator (15198), to dependence on two counterparties for power and hosting: Beowulf at Hardin and, above all, Compute North. On the revenue side, the only revenue counterparty disclosed is the mining pool ViaBTC (14888), with Cumberland DRW handling liquidations (14889). That is effectively single-source concentration.

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed alongside the deals. The dollar amounts that do appear are costs, not revenue: the $67 million Compute North commitment (14331), the $14.6 million deposit (14333), the $11.22 million share issuance (14640), and the $500,000 DMG licence. Two extracted figures, "$67" (14311) and "$500" (14315), are unit errors. The input also ends mid-record at claim 142xx, so later developments are not assessed here.

Marathon's record contains almost no customer agreements in the conventional sense, and readers should weigh it that way. As a bitcoin miner, its revenue comes from block rewards paid through a mining pool, ViaBTC (claim 14888), and from bitcoin sales to Cumberland DRW (claim 14889). Nearly every other entry is a supplier, host, financing, or licensing counterparty. These include Block Maintain, Chord Advisors and Securitas at Granby (claims 15198, 15200, 15199), H.C. Wainwright's at-the-market offering (claim 15202), and DMG's $500,000 Blockseer license (claim 14305). These are costs or capital sources, not demand. The one revenue-bearing contract with a named counterparty is the NYDIG Master Securities Loan Agreement, which lends Marathon's bitcoin at interest (claim 14316).

On naming, the record is unusually clean. Every counterparty is identified, and none is described only as a 'strategic partner' or 'anchor customer.' The signed-versus-provisional distinction still matters. The Compute North arrangement for 73,000 miners began as a 'binding letter of intent' (claims 14590, 14304). It was later carried at $67 million (claim 14331) before a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347) formalized part of it. The Beowulf power purchase agreement is signed (claim 14791), but its extensions beyond five years are only 'as mutually agreed' (claim 14801). The Hardin services arrangement also shifts across filings. It first named Beowulf and 2P1 (claim 14792), then 2P1 and Liefern (claim 14756). A later filing says services 'were to' be provided (claim 14734), which implies the arrangement changed or ended.

Counterparty concentration has migrated rather than diversified. Operations depended on three Granby vendors in 2019, then on Beowulf for power, land and operations at Hardin in 2020–21 (claims 14791, 14802, 14649). By 2022 the dependence had moved to Compute North for hosting (claim 14289). In 2023 an Abu Dhabi joint venture with FS Innovation appears (claims 13084, 13074). The count of distinct claims overstates breadth, because the same agreements are re-disclosed repeatedly: the Beowulf power agreement alone appears in claims 14791, 14647 and 14639.

The revenue arithmetic cannot be reconciled from this record. No claim ties an announced agreement to a revenue figure. The largest dollar amounts run against Marathon's revenue: $11.22 million in equity paid to Beowulf and 2P1 (claim 14640) and the Compute North commitments. The supplied record also ends mid-entry at claim 13074, so later disclosures are not assessed here.

Marathon's record contains almost no customer agreements in the conventional sense, and the counterparty list should be read as a map of suppliers, hosts, lenders and joint-venture partners. The only revenue-side counterparty named is ViaBTC, the mining pool that pays Marathon's daily block rewards (claim 14888). Cumberland DRW is a liquidation venue rather than a buyer under contract (claim 14889). Every other named party either receives Marathon's money or equity or provides it capital: Block Maintain, Securitas Canada and Chord Advisors as service vendors (claims 15198, 15199, 15200), Beowulf under an exclusive power purchase agreement (claims 14791, 14801), DMG under a Blockseer license (claim 14305), NYDIG under a bitcoin securities loan (claim 14345), and H.C. Wainwright under an at-the-market offering (claim 15202). To Marathon's credit, the record contains no unnamed 'strategic partner' or 'anchor customer' language. Every counterparty is identified. The apparent breadth is inflated by restatement, however. Block Maintain appears three times (claims 15198, 15068, 14859), and the NYDIG loan (claims 14345, 14306, 14316, 14290) and DMG license (claims 14305, 14330, 14315, 14288) appear four times each from a single March 2022 filing. The firmness of the documentation varies. The Beowulf power agreement and the DMG, NYDIG and FS Innovation shareholders' agreement (claim 12951) are executed contracts. The 73,000-miner Compute North arrangement began as a 'binding letter of intent' (claims 14590, 14331), not a definitive hosting contract, and was only later supplemented by a master agreement carrying a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). The 2P1 and Liefern services agreement is described in April 2021 as services those parties 'were to provide' (claim 14734), which implies the arrangement had been restructured within six months of signing. Operational concentration has migrated in sequence: first Block Maintain at Granby (2019), then Beowulf and 2P1 at Hardin (2020 to 2021), then Compute North for most of the planned fleet (2021 to 2022), then FS Innovation in Abu Dhabi (2023, claims 13056, 12951, 12972). Each shift left Marathon dependent on a single host or power counterparty. The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed, and all tagged dollar amounts are costs or financing: the $11.22 million equity payment to Beowulf and 2P1 (claim 14649), the Compute North deposit, and the $500,000 DMG license. The '$67' and '$500' tags on claims 14311 and 14315 appear truncated relative to their companion claims.

Marathon's record contains almost no customer agreements in the conventional sense. As a bitcoin miner, its revenue comes from block rewards paid through a mining pool, and the only disclosed revenue channel is ViaBTC, the pool to which it points its miners (claim 14888). Nearly every other entry is a supplier, landlord, financing or technology relationship in which Marathon is the paying party. These include Block Maintain, Chord Advisors and Securitas Canada at Granby (15198, 15200, 15199), the Wainwright ATM facility (15202), and the DMG Blockseer license (14305). The one arrangement where a counterparty pays Marathon is the NYDIG securities loan, which bears three percent interest on lent bitcoin (14316). Legacy patent economics survive only as a contingent 30% interest in a special purpose entity (15223).

Every counterparty is named. The file contains no unidentified 'strategic partners' or 'anchor customers'. The more important distinction is between executed and preliminary language. The Beowulf power purchase agreement (14791) and the Data Facility Services Agreement with 2P1 and Liefern (14756, 14738) are described as entered into. So are the NYDIG loan and the DMG license. The Compute North relationship began as a 'binding letter of intent' for 73,000 miners (14590, 14304). It became a master agreement with a $14.6 million deposit (14333) and then an 82/18 joint venture (14347), so its $67 million headline (14331) should be read as LOI-stage until then. The FS Innovation Abu Dhabi venture is a signed shareholders' agreement, but its capital contributions remain subject to conditions (12977, 13056). The Applied Digital hosting item is phrased only as 'entering into' relationships (13055).

Concentration moved through three phases. In 2019 Marathon depended on a single outsourced Granby operator. In 2020–21 it depended on Beowulf and 2P1 at Hardin, where the $11.22 million services fee was paid in stock (14649). In 2021–22 it depended on Compute North, which alone accounted for the largest disclosed dollar commitments. By 2023 hosting and development were spread across Applied Digital and the Abu Dhabi venture. Throughout this period, pool revenue concentration in ViaBTC was never disclosed as having diversified.

The revenue arithmetic cannot be reconciled from this record, because no claim states revenue attributable to any counterparty. The disclosed dollar figures ($67 million, $14.6 million, $11.22 million, $500,000) are all costs or commitments, not income. Extraction units also look inconsistent: claims 14311 and 14315 render $67 million and $500,000 as '$67' and '$500'. Many entries are duplicate restatements of the same agreements across filings (14647/14639, 14288/14305/14330), which inflates the apparent deal count.

Marathon's record is unusual for a customer-diligence read because almost none of the named counterparties are customers. They are suppliers, landlords, lenders, licensors and joint-venture partners. The only revenue-side relationships disclosed are ViaBTC, the mining pool that pays the company's daily block rewards (claim 14888), and Cumberland DRW, the desk used when bitcoin is liquidated (claim 14889). Both are described operationally rather than as contracted offtake, so the true 'customer' is the Bitcoin network itself, intermediated by a single named pool.

On the cost and infrastructure side, the language is mostly signed. Beowulf's exclusive Power Purchase Agreement for 30 MW, expandable to 100 MW, has a stated five-year term (claims 14791, 14801). The adjacent land lease (claim 14802) is also signed, as is the Data Facility Services Agreement with 2P1 and Liefern. That agreement is paid partly in equity valued at $11.22 million (claims 14649, 14756). The April 2021 restatement shifts to 'were to provide' (claim 14734), which hints at a changed arrangement worth confirming. The NYDIG securities loan (claims 14316, 14306) and the $500,000 DMG Blockseer licence (claims 14305, 14330) are executed agreements. The FS Innovation Abu Dhabi shareholders' agreement is a signed joint-venture formation (claims 13074, 12972).

The softer instruments sit with Compute North and Applied Digital. The 73,000-miner hosting arrangement began as a 'binding letter of intent' (claims 14590, 14331). It was later papered as a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). The Applied Digital 'hosting relationships' (claim 13055) carry no disclosed term, price or capacity. Every counterparty on record is identified by name, and there are no anonymous 'strategic partners'.

Concentration has migrated rather than diversified. In 2019 it sat with the Granby operator Block Maintain, a vendor repeated verbatim across three filings (claims 15198, 15068, 14859). By late 2020 it had moved to Beowulf and Hardin, and through 2021–22 to Compute North, which hosted the bulk of the planned fleet. Only in 2023 do Applied Digital and the Abu Dhabi venture begin spreading the risk.

Revenue arithmetic cannot be reconciled from this record. None of the claims discloses revenue, and the $67 million associated with Compute North (claims 14331, 14311) is a hosting outlay, not income. Several extracted amounts also appear unit-truncated ('$67', '$500', '$81,200' in claims 14311, 14315, 13074), so they should be checked against the source filings before use.

Marathon's record contains almost no customer agreements in the conventional sense. As a bitcoin miner, its only revenue counterparty on file is the ViaBTC mining pool that pays its daily block reward (claim 14888), with Cumberland DRW named as the desk it trades with when it sells bitcoin (14889). Every other entry is a supplier, host, lessor, financier or joint-venture partner. The useful question is therefore which cost and capacity commitments were actually signed and which were preliminary. The signed instruments are clear enough. They include the H.C. Wainwright at-the-market agreement (15202), the exclusive Beowulf power purchase agreement for 30 MW scaling to 100 MW, with a five-year term (14791, 14801), the October 6, 2020 Data Facility Services Agreement paid in 3,000,000 shares valued at $11.22 million (14738, 14649), the NYDIG master securities loan (14316), the DMG Blockseer license at $500,000 (14330), and the FS Innovation shareholders' agreement for the Abu Dhabi entity (13074). The Compute North relationship is the one that moved through stages. It began as a hosting partner described in 2020 (14861), became a 'binding letter of intent' for 73,000 miners in May 2021 (14590), was formalized in a September 2021 master agreement with a $14.6 million deposit (14333), and became an 82/18 joint venture that December (14347). The Applied Digital arrangement is described only as 'hosting relationships', with no date, capacity or price (13055). To Marathon's credit, every counterparty is named, and there is no reliance on unidentified 'strategic partners'. The disclosure does have two inconsistencies. The services agreement is attributed to Beowulf and 2P1 in some filings (14792, 14640) and to 2P1 and Liefern in others (14756, 14738). The April 2021 filing also says those parties 'were to' provide services (14734), which hints at a change the record does not explain. Concentration shifted from Granby vendors (Block Maintain, Securitas, Chord) in 2019, to Beowulf at Hardin in 2020-21, to heavy dependence on Compute North through 2022, and then toward Applied Digital and the Abu Dhabi venture in 2023. The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed, and the dollar amounts attached are outflows, not inflows. The units are also inconsistent: the Compute North figure appears as both $67,000,000 (14331) and '$67' (14311), and the Abu Dhabi figure appears as '$81,200' (13074). The record also repeats claims heavily, which overstates how many distinct deals there are.

Marathon's record contains almost no customer agreements in the conventional sense. The counterparties are overwhelmingly suppliers, service providers, financiers, and joint-venture partners of a bitcoin miner, and the nearest thing to a revenue counterparty is the mining pool ViaBTC, which pays block rewards (claim 14888), plus Cumberland DRW as the liquidation venue (claim 14889). Read as a diligence file, the useful question is not who is buying from Marathon but how firmly its capacity-side relationships were papered. Every counterparty in the file is named. No 'strategic partner' or 'anchor customer' appears without identification, which is a point in the disclosures' favor. Firmness varies considerably, however. The Beowulf arrangement at Hardin is a signed, exclusive Power Purchase Agreement with a stated five-year term and extension options (claims 14791, 14801), paired with a land lease (claim 14802) and a Data Facility Services Agreement compensated in equity valued at $11.22 million (claims 14649, 14640). The Compute North relationship began as a 'binding letter of intent' to host 73,000 miners (claims 14590, 14331, 14304). It was later described as a master agreement with a $14.6 million deposit (claim 14333) and a joint venture owned 82/18 (claim 14347), a progression from LOI toward definitive terms that the record documents only in part. The Applied Digital entry is framed loosely as 'third-party hosting relationships' with no disclosed contract value or term (claim 13055). The FS Innovation Abu Dhabi entity rests on a signed Shareholders' Agreement (claims 13074, 12972). Concentration moved in stages. Early operations relied on Block Maintain, Chord Advisors, and Securitas at a single Quebec site (claims 15198–15200). Hosting and power then consolidated around Beowulf and, far more heavily, Compute North, which carried the largest disclosed dollar commitment ($67 million for the 73,000-miner program). By 2023 the file shows diversification toward Applied Digital and the Abu Dhabi venture, but it is silent on what happened to the Compute North commitments, which is a material gap. The revenue arithmetic cannot be reconciled from this record because no revenue, hashrate-to-revenue, or per-counterparty revenue figures are disclosed; the dollar amounts are all cost, deposit, or equity-consideration items. Headline counts are also inflated by repetition, since roughly half the claims restate the same Compute North, NYDIG, and DMG agreements from a single March 2022 filing. Some amounts appear truncated: '$67' in claim 14311 and '$500' in claim 14315 correspond to $67 million and $500,000 elsewhere, and '$81,200' in claim 13074 is unexplained. These amounts should not be summed as independent commitments.

The first thing to establish about this record is that almost none of it describes customers. Marathon's named counterparties are overwhelmingly vendors, hosts, lenders and landlords: financial back-office from Chord Advisors, site operations from Block Maintain, guarding from Securitas Canada (claims 15198-15200, repeated verbatim at 15068-15070 and 14859-14861), power from Beowulf (14791, 14639, 14647), operations and maintenance from Beowulf/2P1/Liefern (14792, 14756, 14738, 14734, 14649, 14640), hosting from Compute North (14861, 14289) and later Applied Digital (13055). The only counterparties on the inbound side of the ledger are ViaBTC, which pays the daily pool reward (14888), Cumberland DRW, which buys the coin when Marathon liquidates (14889), and NYDIG, which pays three percent on borrowed bitcoin (14290/14306/14316/14345). A reader scanning headlines for demand-side validation will find two names and a lending desk.

On identification, the record is unusually clean: there is no "anchor customer" or unnamed "strategic partner" language anywhere. Every operating counterparty is named. The two unnamed entities are structural rather than commercial, being the patent special purpose entity in which Marathon holds a 30% waterfall interest (15223) and the ADGM vehicle formed with FS Innovation (13084, 13074).

Contract quality varies more than the headlines suggest. The Compute North relationship was announced as a "binding letter of intent" (14590, 14304, 14331) and only hardened into a master agreement with a $14.6 million staged deposit (14333) and then an 82/18 joint venture (14347). The $67 million figure at 14311 is an obligation, not a booking. Note also that the March 9, 2022 cluster restates the same three agreements across roughly eleven claim ids, which inflates apparent deal count without adding counterparties.

Concentration has migrated rather than diversified: from small Quebec service vendors in 2019, to a single power landlord at Hardin in 2020-21, to a single host in 2021-22, to Applied Digital and an Abu Dhabi JV by 2023. Each era carries essentially one load-bearing name.

The arithmetic does not reconcile because it cannot. The disclosed dollar terms - $11.22 million in shares, $500,000 for the Blockseer license (14288/14305), $0.006/kWh O&M (14650), $0.044/kWh plus $0.50 per machine hosting (14289) - are all cost inputs. Revenue remains a function of hashrate, difficulty and bitcoin price, arriving through ViaBTC, and no sum of these announcements produces it.

The first thing to establish is that almost nothing in this record is a customer agreement. Across roughly forty claims spanning 2019 to 2023, Marathon appears overwhelmingly as the buyer, lessee, licensee or lender of assets rather than as a vendor. Securitas Canada, Chord Advisors and Block Maintain (claims 15198, 15199, 15200, repeated verbatim in 15068-15070 and again in 14859-14860) are service suppliers. Beowulf and Two Point One supply power and O&M (14791, 14792, 14801, 14802). Compute North supplies hosting (14590, 14304). DMG Blockchain licenses technology to Marathon, not the reverse (14305). The company's only genuinely revenue-adjacent counterparties are ViaBTC, the mining pool that pays the daily block reward (14888), Cumberland DRW, the liquidation venue (14889), and NYDIG, which borrows bitcoin at three percent (14290, 14306, 14316, 14345). Anyone reading this as a customer roster is misreading a procurement list.

On execution status, the distinction is sharper than the language suggests. The Beowulf power purchase agreement, the Data Facility Services Agreement, the DMG license, the NYDIG master securities loan agreement and the FS Innovation shareholders' agreement (12972, 13074, 13084) are all executed instruments with stated terms. Compute North is the exception worth flagging: the 73,000-miner arrangement was announced in May 2021 as a "binding letter of intent" (14590) and was still being described in those terms in the March 2022 filings (14304, 14311, 14331), even after a separate master agreement and a $14.6 million deposit (14333) and an 82/18 joint venture (14347). Binding LOI is not the same as a definitive hosting contract, and the persistence of that phrasing across three reporting periods is the single most useful tell in the file.

One counterparty is never identified: the special purpose entity holding the '798 patent, in which Marathon retains a 30% waterfall interest (15223). That is the only inbound economic claim in the 2019 record and it is attached to an unnamed vehicle with no quantified recovery.

The arithmetic therefore cannot reconcile to revenue. The disclosed figures — $67 million, $14.6 million, $500,000, $11.22 million in stock — are outflows. Concentration migrated from several small vendors toward near-total dependence on Compute North and Beowulf by 2021, which is a supply-side single-point risk, not customer diversification.

The first thing worth stating plainly is that almost nothing in this record is a customer agreement. Marathon's named counterparties are overwhelmingly vendors and suppliers to which the company pays money: security from Securitas Canada (15199, 15069, 14860), bookkeeping from Chord Advisors (15200, 15070), site operations from Block Maintain at Granby (15198, 15068, 14859), power and O&M from Beowulf and Two Point One at Hardin (14791, 14792, 14801, 14802, 14756, 14738, 14734, 14647, 14649, 14640), and hosting from Compute North (14861, 14590, 14304, 14289). A miner does not sell to customers in the ordinary sense; its revenue counterparty is the protocol. The only genuine inbound-revenue relationships disclosed are ViaBTC, the pool that pays the daily block reward (14888), and NYDIG, which pays three percent on borrowed bitcoin under the Master Securities Loan Agreement (14306, 14290, 14345, 14316). Cumberland DRW is a liquidation venue, not a customer (14889).

On identification, Marathon is unusually specific by name and unusually vague where value sits. The patent special purpose entity holding the '798 Patent is never identified, and neither are any licensees or defendants behind the 30% waterfall interest (15223). The Abu Dhabi vehicle formed with FS Innovation is described only as 'the ADGM Entity' (12972, 13084, 13074), and the $81,200 figure attached to it is too small to be a capitalization of consequence without further context. 'Hosting partner' language (14861) preceded any binding instrument by nine months.

Concentration moved decisively in one direction. The 2019 vendor set was diffuse and small-dollar; by 2021 the entire asset base ran through two dependencies, Beowulf/2P1 for Hardin energy and Compute North for hosted fleet, with the latter carrying 73,000 machines under a binding letter of intent (14590), a $14.6 million staged deposit (14333), an $18-month, $67 million commitment (14311, 14331), and an 82/18 joint venture (14347). That is a single point of failure dressed as diversification.

The revenue arithmetic does not reconcile because it was never meant to. The headline dollar figures — $67 million, $14.6 million, $11.22 million in stock to Beowulf/2P1 (14649, 14640), $500,000 to DMG for Blockseer (14305, 14288, 14330) — are outflows, not bookings. No disclosed revenue line attaches to any named counterparty other than the pool.

The first thing a reader should understand about Marathon's counterparty record is that it contains almost no customers. Nearly every named relationship runs the other direction: Block Maintain for site operations and Securitas Canada for security at Granby (claims 15198, 15199, 14859, 14860), Chord Advisors for finance (15200, 15070), Beowulf and Two Point One plus Liefern for power and O&M at Hardin (14791, 14792, 14756, 14738), Compute North and later Applied Digital for hosting (14861, 13055), and DMG Blockchain for a pool license (14288, 14305, 14315). These are vendors Marathon pays. The only revenue-facing names in the whole set are ViaBTC, described as the pool from which Marathon receives its daily reward (14888), Cumberland DRW as the liquidation venue (14889), and NYDIG, to which Marathon lends bitcoin at three percent (14290, 14306, 14316) — the single genuinely income-producing named agreement on record, and one with no disclosed notional.

On signed versus soft, the record is cleaner than most. The Compute North relationship begins as a binding letter of intent dated May 21, 2021 for 73,000 miners (14590, 14304, 14311, 14331), then hardens into a master agreement with a $14.6 million staged deposit (14333) and an 82/18 joint venture entity (14347) — an LOI that actually converted, with priced terms of $0.50 per machine per month and $0.044/kWh (14289). The Abu Dhabi arrangement with FS Innovation is a shareholders' agreement rather than an MOU (13074, 13063), but Marathon holds only 20 percent and the disclosed figure is $81,200, a formation-level number that does not support the geographic headline it carries.

Concentration moved sharply in one direction and then partially back. The 2019 posture was a scatter of small Granby vendors; by 2021-2022 essentially all operating capacity routed through Compute North across four instruments, a single point of failure that the March 2023 turn to Applied Digital (13055) implicitly acknowledges.

The arithmetic cannot be reconciled because the announced deals are costs, not revenue: $67 million of hosting, $0.006/kWh at Hardin (14650), and $11,220,000 in stock to Beowulf and 2P1 (14649, 14640) — non-cash consideration funded, ultimately, by the H.C. Wainwright ATM (15202). The supplied record stops at claim 13056 in March 2023; nothing later is assessable here.

The first thing to say about Marathon's disclosure record is that almost none of it describes customers. The counterparties named across 2019-2023 are vendors, hosts, power suppliers, lenders and a placement agent: Block Maintain for Granby mining operations (15198, 15068, 14859), Chord Advisors for finance (15200, 15070), Securitas Canada for site security (15199, 15069, 14860), ViaBTC as the mining pool (14888), Cumberland DRW as the liquidation venue (14889), and H.C. Wainwright under an At The Market equity agreement (15202) that is capital markets plumbing rather than a commercial relationship at all. Marathon is the payer in essentially every one of these. The only revenue-side item in the entire set is the 30% waterfall interest in the patent special purpose entity (15223), and there the counterparty is the least identified thing on record — the SPE is unnamed, the licensees behind any net proceeds are unnamed, and the entitlement is contingent rather than contracted.

On the signed-versus-soft distinction, the record is cleaner than most: these are executed agreements with dates and terms, not MOUs. The one genuine hedge is the Compute North arrangement, which entered as a binding letter of intent on May 21, 2021 for 73,000 miners (14590, 14331, 14304, 14311) and only became a master agreement on September 3, 2021 with a $14.6 million staged deposit (14333). Diligence should also discount the apparent deal count: the March 9, 2022 filing alone generates roughly ten claim ids (14288-14345) covering four underlying agreements — NYDIG, DMG, Compute North, and hosting economics — so repetition is filing artifact, not deal flow.

Concentration moved sharply and then broke. 2019 is diffuse and small-dollar; 2020-21 concentrates into Beowulf and 2P1 at Hardin, with an exclusive PPA at 30MW scaling to 100MW (14791, 14647, 14639), $0.006/kWh operating cost (14650), and $11.22 million paid in stock rather than cash (14649, 14640); 2021-22 concentrates further into Compute North at $0.044/kWh plus an 82/18 joint venture (14347, 14289); 2023 shows the pivot to Applied Digital (13055) and the Abu Dhabi vehicle with FS Innovation, where Marathon holds only 20% (13063, 13074 — the $81,200 tag there is a label artifact, not deal value).

Nothing here reconciles to revenue because none of it produces revenue. The dollar figures are deposits, licenses and share issuances on the cost side; disclosed revenue arrives as block rewards through ViaBTC (14888), a counterparty that appears exactly once.

Marathon's record contains essentially no signed customer agreements in the conventional sense, because a bitcoin miner's revenue comes from block rewards rather than from contracted buyers. The nearest thing to a revenue counterparty is ViaBTC, described as the pool that pays a daily reward (claim 14888), with liquidation routed through Cumberland DRW (claim 14889). Neither is framed as a contract with committed volume. Almost everything else on file is supply-side: facility operators, power, hosting, financing and licensing. These include Block Maintain, Securitas Canada and Chord Advisors at Granby (claims 15198-15200, repeated in 15068-15070 and 14859-14860), the H.C. Wainwright ATM (claim 15202), the NYDIG securities loan (claims 14290, 14306) and the DMG Blockseer license (claims 14288, 14305).

On identification, the record is clean. Every counterparty is named, and there are no unidentified 'strategic partners' or 'anchor customers'. The residual opacity is the 2019 patent special purpose entity, in which Marathon holds a 30% waterfall interest but whose other holders are not disclosed (claim 15223).

The signed-versus-soft distinction matters most for Compute North. The 73,000-miner hosting arrangement began as a 'binding letter of intent' (claims 14590, 14304, 14331). It was later supported by a September 2021 master agreement with a $14.6 million deposit (claim 14333), stated hosting rates (claim 14289) and an 82/18 joint venture (claim 14347). Readers should note that claims 14304, 14311 and 14331 are the same LOI restated, not three separate deals. Beowulf's Power Purchase Agreement is plainly executed, with a five-year term and 30 MW firm (claims 14791, 14801). However, Phase II up to 100 MW is expansion capacity, not committed load. The Data Facility Services Agreement is described in inconsistent terms across filings. Beowulf and 2P1 appear in claim 14792, and Liefern and 2P1 appear in claim 14756. By 2021 the services are something 2P1 and Liefern 'were t[o]' provide (claim 14734), a tense shift worth diligencing.

Concentration moved from a single Quebec site dependent on Block Maintain (2019), to Beowulf at Hardin (2020), to heavy reliance on Compute North for fleet deployment (2021-22), and then to a minority 20% position in an FS Innovation-controlled Abu Dhabi vehicle (claims 13063, 13084).

The revenue arithmetic cannot be reconciled from this record. No claim discloses mining revenue, hashrate or pool payouts that could be tied to the announced capacity. Several dollar tags are also ambiguous about units: '$67' (claim 14311), '$500' (claim 14315) and '$81,200' (claim 13074). They should be checked against source filings before anyone relies on them.

Marathon's record contains no revenue customers in the conventional sense. Every named counterparty is a supplier, host, service provider, financier or co-investor, and the only revenue-side relationship disclosed is the ViaBTC mining pool that pays the company's daily block rewards (claim 14888), with Cumberland DRW named as the liquidity venue for selling bitcoin (claim 14889). Read accordingly, the question is whether these cost-side agreements are signed or aspirational. Most are signed. The Beowulf power purchase agreement is described as entered into, with a five-year initial term and stated extension mechanics (claims 14791, 14801, 14639). The Hardin data facility services agreement with 2P1 and Liefern is dated October 6, 2020 (claims 14738, 14756). The DMG Blockseer licence (claim 14305), the NYDIG securities loan (claim 14290) and the FS Innovation shareholders' agreement (claim 13084) are all executed documents. The main softer instrument is the May 2021 Compute North arrangement. It was a binding letter of intent to host 73,000 miners (claim 14590), later firmed up by a September 2021 master agreement carrying a $14.6 million deposit (claim 14333) and by an 82/18 joint venture (claim 14347). The April 2021 filing refers to services 2P1 and Liefern 'were to' provide (claim 14734). That phrasing suggests the Hardin operating arrangement changed, and the record does not explain how. Every counterparty is named, and the file contains no unnamed 'strategic partner' or 'anchor customer' language. The one exception is the Applied Digital hosting relationship (claim 13055), which is named but has no disclosed terms. Concentration has shifted three times. The first phase was a single Quebec site run by Block Maintain (claim 15198). The second was a Montana dependence on Beowulf for power, land and operations (claims 14791, 14802, 14649). The third was a heavy Compute North concentration covering 73,000 machines. Applied Digital and the Abu Dhabi venture (claims 13055, 13074) show only partial diversification by 2023. The record gives no revenue figures, so the deal arithmetic cannot be reconciled against revenue. The dollar amounts disclosed are outflows or consideration: $67 million of hosting prepayment (claim 14331), $11.22 million in equity paid to Beowulf and 2P1 (claim 14649), and a $500,000 licence (claim 14305). The Abu Dhabi figure is captured as '$81,200' (claim 13074), and its unit is ambiguous. Readers should also note that many claims restate the same four or five agreements across multiple filings, and that the record appears truncated after early 2023.

Marathon's record contains almost no customer agreements in the conventional sense, and it should not be read as a customer book. As a bitcoin miner, its operating revenue comes from protocol block rewards routed through a mining pool, disclosed as a daily reward from ViaBTC (claim 14888), with liquidation through Cumberland DRW (claim 14889). Nearly every other entry is a supplier, financing, or joint-venture arrangement where Marathon is the paying party: security from Securitas Canada (15199), Granby operations from Block Maintain (15198), financial operations from Chord Advisors (15200), an ATM facility with H.C. Wainwright (15202), and a DMG Blockseer license under which Marathon pays $500,000 (14305). The count is also inflated by repetition across filings. The Beowulf power purchase agreement alone appears as 14791, 14647 and 14639, and the Compute North letter of intent recurs in 14590, 14304, 14311 and 14331.

On signed versus preliminary language, the record is mostly executed documents. The Beowulf PPA is described as entered into, with a five-year term (14801), and the NYDIG securities loan (14290) and the FS Innovation shareholders' agreement (13084) are likewise executed. The main exception is Compute North. It began as a 'binding letter of intent' to host 73,000 miners (14590), and was only later papered as a master agreement with a $14.6 million deposit (14333) and a 82/18 joint venture (14347). The 2P1/Liefern services agreement warrants attention: a later filing refers to services the parties 'were to provide' (14734), which implies the arrangement did not run as announced. The patent SPE interest (15223) is a contingent 30% waterfall, not a contracted stream.

Every counterparty on record is identified by name, and there are no anonymous 'strategic partners' or 'anchor customers'. Concentration has migrated rather than diversified. Dependence ran from Block Maintain at Granby in 2019, to Beowulf and 2P1 at Hardin in 2020, and then to Compute North as the dominant host in 2021–22. By 2023 it had begun spreading across Applied Digital (13055) and the Abu Dhabi venture with FSI (13074).

The revenue arithmetic cannot be reconciled from this record. No revenue figures are disclosed, and the dollar amounts that do appear are costs or equity consideration. These include the $11.22 million share issuance to Beowulf and 2P1 (14649) and the Compute North figures, which are inconsistently scaled between $67 (14311) and $67,000,000 (14331). The FSI amount of $81,200 (13074) has an unstated unit. Mined-bitcoin revenue should be tested against hashrate and pool payouts, not against these agreements.

Marathon's record contains almost no customer agreements in the conventional sense. Mining revenue comes from pool rewards, and the only disclosed revenue-side counterparty is ViaBTC, described as the pool that pays a daily reward (claim 14888). Cumberland DRW is a liquidation venue, not a buyer of services (claim 14889). The rest of the record describes vendors, hosts, power suppliers, financiers and joint-venture partners, so it should be read as a map of supply-side dependencies rather than a demand book.

The record does not lean on anonymous 'strategic partners' or unnamed 'anchor customers'. Every counterparty is identified by legal name, and several items reappear across filings rather than representing new relationships. Block Maintain, Securitas Canada and Chord Advisors appear three times each (claims 15198, 15068, 14859), and the NYDIG loan and DMG licence each recur several times in the March 2022 filing (claims 14290, 14306, 14316, 14345; 14288, 14305, 14315, 14330).

The distinction between executed and preliminary language matters most for Compute North. The 73,000-miner hosting arrangement was first disclosed as a 'binding letter of intent' (claims 14590, 14304, 14331). It was subsequently backed by a master agreement with a $14.6 million deposit (claim 14333) and an 82/18 joint venture (claim 14347). That sequence is firmer than a bare LOI, but the $67 million headline figure traces to LOI language. Beowulf's power purchase agreement is described as entered into, with a five-year term (claims 14791, 14801). The Hardin operations and maintenance counterparties are named inconsistently: 'Beowulf and 2P1' in some filings (claims 14792, 14649) and '2P1 and Liefern LLC' in others (claims 14756, 14738). The April 2021 wording that services 'were to' be provided (claim 14734) warrants checking for termination. The FS Innovation Abu Dhabi entity rests on a signed shareholders' agreement (claims 12972, 13084).

Concentration has moved through three phases. In 2019 the business depended on one site's vendors at Granby. In 2020 it depended on a single power host at Hardin. From 2021 Compute North became the dominant hosting relationship, before the 2023 FSI venture added a foreign counterparty.

The revenue arithmetic cannot be reconciled from this record, because no revenue figures are disclosed alongside the deals. Several amounts also lack consistent units. Claims 14311 and 14315 show $67 and $500 against $67,000,000 and $500,000 elsewhere, and the $81,200 attached to claim 13074 has no stated unit and a truncated quote. The claim set also ends mid-2023, so later developments are not covered.

Marathon's record contains no customer agreements in the conventional sense, which is the first thing a diligence reader should register. A bitcoin miner earns block rewards, and the only revenue-side counterparties named are the ViaBTC mining pool (14888) and Cumberland DRW as the liquidation desk (14889). Every other claim describes a supplier, host, landlord, lender, licensor or joint-venture partner. None of these is a buyer of Marathon's output. Every counterparty is identified by name, and the record contains no unnamed strategic partner or anchor customer. The record is also heavily duplicated: roughly forty claims reduce to about a dozen distinct arrangements.

The signed-versus-soft distinction matters most for Compute North. The 73,000-miner hosting arrangement was first disclosed as a binding letter of intent (14590, 14331, 14311, 14304). A master agreement with a $14.6 million deposit followed (14333), and later an 82/18 joint venture (14347). The record contains nothing on Compute North after March 2022. The Beowulf power purchase agreement is described as entered into, with defined 30 MW and 100 MW phases and a five-year term (14791, 14801, 14639). The Hardin services agreement is signed, but its named parties shift across filings. It lists Beowulf and 2P1 in 14792 and 14649, and Liefern and 2P1 in 14756 and 14738. By April 2021 it is phrased in the past tense, as services that were to be provided (14734), which implies a change the record never explains. The FS Innovation Abu Dhabi venture rests on an executed shareholders' agreement (13056, 12972). The Applied Digital relationship is described only as entering into hosting relationships (13055), with no terms disclosed.

Counterparty concentration has migrated rather than diminished. In 2019 a single Quebec site depended on Block Maintain (15198, 15068). The 2020 build-out depended on Beowulf's Hardin station. In 2021 and 2022 essentially all incremental capacity was routed through Compute North. Only in 2023 do Applied Digital and FSI appear as a second and third leg.

The revenue arithmetic cannot be reconciled from this record because no revenue figure is disclosed alongside any named deal. The only dollar amounts on record are costs. These are the $67 million Compute North commitment (14331), set against a $14.6 million deposit (14333) that covers only part of it; $11.22 million of equity paid to Beowulf and 2P1 (14649, 14640); and a $500,000 DMG pool licence (14305). Two extraction labels, $500 (14315) and $67 (14311), are unit-dropping errors and should not be relied on.

Marathon's record contains no customer agreements in the conventional sense. Every named counterparty is a supplier, lender, host, licensor or joint-venture partner, and the business earns its revenue from block rewards paid through a mining pool, identified as ViaBTC (14888), with liquidations run through Cumberland DRW (14889). On this record the diligence question is how firm the supply-side relationships are, not how firm demand is.

Most of these agreements read as executed. They include the Beowulf exclusive power purchase agreement, with 30 MW rising to 100 MW over a five-year initial term (14791, 14801), the Hardin land lease (14802), the October 2020 Data Facility Services Agreement (14756, 14738) and the NYDIG securities loan (14306). They also include the $500,000 DMG Blockseer licence (14305) and the H.C. Wainwright ATM (15202). The softer items need care. The Compute North relationship began as a "binding letter of intent" for 73,000 miners (14590), became a master agreement with a $14.6 million deposit (14333), and then became an 82/18 joint venture (14347). The Abu Dhabi vehicle with FS Innovation is a shareholders' agreement whose capital contributions remain "subject to the satisfaction or waiver of certain conditions" (12972, 12977). The Applied Digital hosting relationship is described only as "entering into" arrangements, with no disclosed terms (13055). The April 2021 restatement of the services agreement shifts to the past tense, saying 2P1 and Liefern "were to" provide services (14734), which suggests those terms were revised.

All counterparties are named, and none are described only as unidentified strategic partners. The March 2022 filings, however, repeat 2021 deals several times over (14304, 14311, 14331; 14288, 14305, 14330; 14290, 14316, 14345). Those repeats should not be read as new relationships.

Operational concentration has moved in steps. In 2019 the company relied on a single Quebec site run by Block Maintain (15198). From late 2020 it depended on Beowulf for power and land at Hardin (14791). During 2021–22 its expansion capacity rested heavily on Compute North (14590, 14347). By 2023 that exposure had spread across Applied Digital and an overseas joint venture (13055, 12972).

The revenue arithmetic cannot be reconciled, because the record discloses no revenue figures. The dollar amounts are costs: the $67 million hosting prepayment, the $14.6 million deposit, $11.22 million in share-based compensation to Beowulf and 2P1 (14649), and power and operating rates of $0.044/kWh and $0.006/kWh (14289, 14650). The extraction also shows unit errors, with "$67" (14311) and "$500" (14315) appearing where the underlying figures are $67 million and $500,000.

Marathon's record is not a customer book in the conventional sense. A self-mining bitcoin producer earns block rewards through a pool, which here is ViaBTC (14888), so every counterparty on file is a supplier, host, lender, licensor or joint-venture partner rather than a buyer. That framing matters for everything below. The good news on identification is that no counterparty is hidden behind 'strategic partner' or 'anchor customer' language. Every relationship names its entity.

The instruments behind those names vary a great deal. The executed contracts are the Beowulf exclusive Power Purchase Agreement, with a five-year term and extensions (14791, 14801), the adjoining land lease (14802) and the Data Facility Services Agreement dated October 6, 2020 (14738). Also executed are the DMG Blockseer license (14330), the NYDIG Master Securities Loan Agreement (14290), the September 2021 Compute North master agreement carrying a $14.6 million deposit (14333) and the Marathon Compute North 1 LLC joint venture (14347). The first Compute North commitment was only a 'binding letter of intent' for 73,000 miners (14590, 14331). The FS Innovation Abu Dhabi Shareholders' Agreement is signed, but its capital contributions remain 'subject to the satisfaction or waiver of certain conditions' (12977). The Securitas, Chord Advisors, Block Maintain, Cumberland DRW and Applied Digital references (14860, 14070, 14068, 14889, 13055) describe working relationships and cite no underlying agreement.

The cast around the Hardin services agreement drifts between filings. The parties are 2P1 and Liefern in 14756 and 14738, but Beowulf and 2P1 in 14649. The April 2021 filing also says services 'were to' be provided (14734), which hints that the agreement was restructured or ended.

Concentration moved in three stages. It began with a Quebec vendor cluster around Granby in 2019, which gave way to dependence on Beowulf's Hardin power and site in 2020. By 2021–22 exposure had shifted to Compute North, which hosted 73,000 miners and received sizable prepayments (14311, 14333). By 2023 the base had broadened to Applied Digital in Texas and North Dakota and to the Abu Dhabi JV (13055, 12951). That broadening lowered single-host risk after Compute North's prominence.

Revenue cannot be reconciled from this record. No revenue figures appear, and the dollar amounts are all cost or capital items. These include $0.006/kWh operating cost (14650), the $0.044/kWh hosting rate (14289), $11.22 million in equity for 3,000,000 shares (14649), $67 million tied to Compute North (14331) and the Abu Dhabi commitment (13074). Amount units are also inconsistent across duplicate extractions: DMG appears as $500,000 in 14330 but '$500' in 14315, Compute North as '$67' in 14311, and Abu Dhabi as '$81,200' in 13074. These figures should be verified against source filings before anyone relies on them. The record also truncates at claim 13056.

Marathon's record contains almost no customer agreements in the conventional sense. As a self-mining bitcoin producer, its revenue comes from block rewards paid through a pool, and nearly every named counterparty here is a supplier, host, lender or co-investor. The closest thing to a revenue counterparty is ViaBTC, which pays the daily mining reward (claim 14888). Cumberland DRW is the liquidation desk (14889). The record should be read as supply-side and financing diligence.

On the signed-versus-soft distinction, the Beowulf power purchase agreement is executed. It commits 30 MW with expansion to 100 MW over a five-year initial term (14791, 14801). It is backed by a land lease (14802) and a Data Facility Services Agreement with 2P1 and Liefern, paid with 3,000,000 shares valued at $11.22 million (14649, 14640, 14738). The Compute North relationship began softer. The 73,000-miner arrangement was a binding letter of intent (14590), and the March 2022 filing still used that label (14304, 14331). It was later reinforced by a September 2021 master agreement with a $14.6 million deposit (14333) and an 82/18 joint venture (14347). The FS Innovation Abu Dhabi vehicle is a signed shareholders' agreement. Its capital contributions, however, are subject to conditions (12977), and Marathon holds only 20% (13063). The Applied Digital arrangement is described only as hosting relationships, with no stated term or value (13055).

Every counterparty in the record is identified by name. No unnamed strategic partners or anchor customers appear. The count overstates breadth, though. The 2022-03-09 cluster restates 2021 agreements with NYDIG, DMG and Compute North several times each (14345, 14306, 14290, 14316; 14305, 14288, 14330).

Concentration has shifted over three phases. From 2020 to 2021 it ran through a small group of Canadian service vendors (14859, 14860) and then the Beowulf/2P1 Hardin complex. From 2021 to 2022, a single host, Compute North, became dominant. In 2023 exposure spread to Applied Digital and the Abu Dhabi joint venture. The record does not show how the Compute North exposure was resolved, and that gap is material.

The revenue arithmetic cannot be reconciled from this record. No disclosed revenue figures are attached to any agreement, and the stated dollar amounts are costs rather than inflows. Several extracted values also look like unit errors: DMG at $500 versus $500,000 (14315 vs 14305), Compute North at $67 versus $67,000,000 (14311 vs 14331), and FSI at $81,200, which is likely in thousands (13074). The input is also truncated at claim 12687, so later agreements are not assessed here.

The first thing to understand about MARA's record is that almost none of it describes customers. The company earns revenue from bitcoin mining, and the named counterparties are mostly suppliers, hosts, lenders, licensors or joint-venture partners. The only company-to-customer language on record is prospective AI/HPC tenancy, and no tenant has been named. That said, the record is unusually free of anonymous "anchor customer" framing. Every counterparty is identified: Compute North, DMG Blockchain, NYDIG, FS Innovation, Applied Digital, Auradine, TAE Power Solutions, EDF Pulse Ventures, Starwood and MPLX. What goes unidentified is the demand side of the Starwood platform.

The binding status varies. Compute North began as a binding letter of intent to host 73,000 miners (claims 14590, 14331, 14304). It became a signed master agreement with a $14.6 million deposit and stated per-kWh terms (14333, 14289), then an 82/18 joint venture (14347). The DMG Blockseer license (14330, 14288, 14305) and the NYDIG securities loan, in which MARA is the lender (14345, 14290, 14316, 14306), are executed contracts. The FS Innovation Abu Dhabi shareholders' agreement is signed, but its capital contributions are conditional (12951, 12977, 13063). Applied Digital appears only as a hosting relationship being entered into (13055). Auradine is a related-party governance tie rather than a disclosed commercial contract (12692, 12687). TAE is a headline partnership with no terms on record (9272). The EDF Pulse agreement is an equity investment in Exaion (9271). Starwood is an executed Strategic Agreement (3660, 11780). Its economics are elective: MARA takes a 10% to 50% interest in each joint venture, and Starwood serves as managing member (3662, 3663). The 2.5 GW figure is a target cited in secondary commentary (926, 7504), and the platform is described as "expected" (11784). MPLX is explicitly at the discussions stage and is framed as a longer-term initiative (11789).

Counterparty concentration has swung from one partner to several and back again. In 2021 and 2022, MARA's infrastructure depended heavily on Compute North. In 2023 it spread across Applied Digital and the Abu Dhabi joint venture. The 2026 Starwood agreement recentralizes the business, because it covers all U.S. properties outside existing joint ventures (3661) and gives Starwood control of permitting and power procurement (3665).

The revenue arithmetic cannot be reconciled from this record. No disclosed revenue is attributable to any announced deal other than NYDIG's 3% interest. The "$67,000,000" label on 14331 and 14311 is not visible in the quoted text. Claim 14315 records the DMG fee as $500 while sibling claims record $500,000. The date of 11780 (February 25) also conflicts with the filing date of February 26 (3660).

Marathon's record contains almost no customer agreements in the revenue sense. As a bitcoin miner, its named counterparties are hosting providers, lenders, licensors, and joint-venture partners, so the question is less whether customers are signed and more whether partnership language implies commitments that have not been made. The claim count also overstates breadth: the NYDIG bitcoin loan appears four times (14345, 14290, 14316, 14306), the DMG Blockseer license four times (14305, 14330, 14315, 14288), and the FS Innovation Abu Dhabi venture seven times (13074, 13084, 12951, 12972, 13063, 12977, 13056).

The firmest instruments are the NYDIG Master Securities Loan Agreement, the $500,000 DMG license (14315's "$500" appears to be an extraction artifact), and the 82/18 Marathon Compute North 1 LLC (14347). The Compute North hosting arrangement was a "binding letter of intent" for 73,000 miners (14304, 14331), later backed by a $14.6 million master-agreement deposit (14333) and priced terms (14289). It is binding, but the filings still call it an LOI. The Abu Dhabi entity is a signed Shareholders' Agreement, but Marathon holds 20% (13063) and contributions are "subject to the satisfaction or waiver of certain conditions" (12977). The Applied Digital hosting "relationships" (13055) and the TAE Power Solutions partnership (9272) carry no disclosed terms. Auradine (12687, 12692) is a related-party governance tie, not a commercial contract. The MPLX West Texas campus is explicitly "discussions" and "a longer-term initiative" (11789).

Every counterparty is named. The gap sits one level down. The Starwood announcement promises "Hyperscale, Enterprise and AI Capable" infrastructure (9267) and targets up to 2.5 gigawatts (7504, 926) or more than a gigawatt near term (2946), yet no tenant or offtaker is identified. The signed Strategic Agreement (3660) lets Marathon elect only a 10% to 50% interest in each venture (3662), makes Starwood the managing member (3663), and leaves permitting and power procurement to Starwood's pre-development work (3665). The EDF Pulse Ventures agreement (9271) is capital coming into the Exaion subsidiary, not a sale of compute.

Concentration has moved in stages. Compute North dominated in 2021–22, FSI and Applied Digital followed in 2023, and Starwood has anchored the AI/HPC narrative since February 2026. None of the claims on record discloses revenue, contracted backlog, or megawatt pricing, so the announced deals cannot be reconciled with reported revenue. Any Starwood economics would flow at the joint-venture level and be diluted by Marathon's elected stake. The claim list supplied was truncated at claim 3661, so later disclosures may change this read.

Across this record, Marathon has not disclosed a single signed customer agreement for AI or HPC capacity. What it has signed are joint-venture, investment and supplier-side documents, and the remaining items are letters of intent, 'discussions' or unnamed interest. The executed paper starts with the January 27, 2023 Shareholders' Agreement with FS Innovation for the Abu Dhabi mining entity (13074, 12951). In that entity Marathon held 20% against FSI's 80% (13063), and capital contributions were conditional (12977). The Applied Digital arrangement (13055) runs the other way, with Marathon buying hosting rather than selling it. Auradine is a related-party link through Fred Thiel's board seat (12687, 12692), not a commercial counterparty. The EDF Pulse Ventures document (9271) is an investment agreement in Exaion, which brings capital in rather than revenue. The TAE Power Solutions announcement (9272) uses partnership language with no disclosed terms.

The Starwood Strategic Agreement of February 26, 2026 is signed (3660), but it is a framework. It covers essentially all U.S. properties outside existing joint ventures (3661). Marathon will elect a 10% to 50% interest in each joint venture (3662), with Starwood as managing member (3663) and responsible for permits and power (3665). Marathon keeps its mining rights under a rent-free lease (3666). MPLX has moved in the opposite direction. It was described as a letter of intent (3524, 3513, 3497, 3503), and in February 2026 it was described as ongoing 'discussions' on a longer-term initiative requiring significant capital (11789). East Ohio Valley Railway is at the 'reasonable best efforts to negotiate' stage (1873). The Hannibal site at Long Ridge cites inbound interest from 'multiple potential investment-grade AI/Critical IT tenants' but names none of them (9414, 10655). The tenant expectation in claim 935 is truncated in the record and cannot be assessed.

Concentration has narrowed sharply over time. In 2023 the record showed one named joint-venture partner and one hosting vendor. In 2026 Starwood accounts for most of the new claims, many of them duplicates (3493, 3501, 3511, 3527, 9267), and controls development across the U.S. footprint. The disclosed figures do not reconcile to revenue because no contract value, tenor, pricing or megawatt-to-dollar rate appears anywhere. The one dollar figure, $81,200 in claim 13074, carries no stated unit. Capacity targets are also inconsistent, with 'up to 2.5 gigawatts' in 926 and 7504 against 'more than a gigawatt near term' in 2946. Combined with the 10% to 50% election, Marathon's attributable share could fall anywhere from roughly 100 megawatts to 1.25 gigawatts. None of that capacity is yet tied to a named tenant.

MARA's record contains no signed customer agreement for AI or HPC compute. The one executed, filed instrument is the February 26, 2026 Strategic Agreement between MARA USA Corporation and Starwood Capital Group (claim 3660). It is a development framework, not a customer contract. MARA contributes its U.S. mining sites outside existing third-party joint ventures (3661) and elects a 10% to 50% interest in each future joint venture (3662, 3494). Starwood acts as managing member (3663) and performs pre-development work, including permits and power procurement (3665). The individual joint ventures are still to be formed, and the 2.5 gigawatt figure is a target (926, 7504, 8403), set against a nearer-term goal of more than one gigawatt (2946). Everything else on file is weaker. MPLX appears first as ongoing discussions that management itself calls longer-term (11789) and then as a letter of intent (3513, 3497). The EDF Pulse Ventures agreement is an investment in the Exaion subsidiary rather than a purchase of services (9271). The TAE Power Solutions announcement gives no economic terms (9272). The Auradine ties are related-party relationships, with a board seat and an executive's roughly 5% shareholding (12687, 12692). The East Ohio Valley Railway item commits the parties only to reasonable best efforts to negotiate (1873). Every actual tenant remains unnamed. The Hannibal site at Long Ridge has drawn inbound interest from multiple potential investment-grade tenants (9414, 10655). A later site has interest from potential HPC tenants (8906). Management also expects to have a tenant in place around closing (935). The Abu Dhabi joint venture partner was never identified (12646). The 70% of Long Ridge output contracted long-term is a power offtake with unnamed counterparties, not compute demand (8504). Counterparty concentration has narrowed from a scattered 2023 to 2025 set to a near-total reliance on Starwood, which now frames development of roughly 90% of non-hosted capacity under consideration (9201, 8905, 8501). Lease guidance has also shifted, from one or more leases by year-end (9198) and at least one (11582) to at least two (8502), with no lessee named at any point. No disclosed contracted revenue, lease rate, or megawatt commitment is tied to a named counterparty, so the revenue arithmetic cannot be reconciled with the announced deals. On the record, the AI/HPC revenue base under these announcements is zero, and every gigawatt figure is capacity ambition.

Marathon Digital's record contains no signed customer agreement in the revenue-paying sense. The one executed contract is the February 26, 2026 Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures (claim 3660), and it is a development and capital arrangement rather than a customer contract. The filing language places Starwood as managing member of each joint venture (claim 3663) and makes it responsible for pre-development, permitting and power procurement (claim 3665). MARA elects a 10% to 50% interest (claims 3662, 3494) and contributes its existing US mining sites (claim 3661). The EDF Pulse Ventures agreement (claim 9271) is also signed, but it is an equity investment in the Exaion subsidiary, not a purchase of services. The TAE Power Solutions 'partnership' (claim 9272) carries no disclosed terms. The Abu Dhabi joint venture (claim 12646) names a jurisdiction but no counterparty. MPLX sits clearly on the non-binding side. It is described as a letter of intent (claims 3503, 3513) and, in management's own words, as continuing 'discussions' on 'a longer-term initiative' (claim 11789).

The counterparties that would actually generate AI/HPC revenue remain unnamed. Management cites 'inbound interest from multiple potential investment-grade AI/Critical IT tenants' at Hannibal (claims 9414, 10655) and 'interest from potential HPC tenants' at the newer site (claim 8906). It also says it 'expects to have a tenant in place' around the Long Ridge close (claim 935). The stated contracted share of Long Ridge's power output, above 70% (claim 8504), likewise comes without buyers or tenor. The Long Ridge acquisition itself is still pending (claim 1214), and its ancillary railroad terms remain at 'reasonable best efforts to negotiate' (claim 1873).

Concentration has narrowed sharply. A scattered set of mining, grid and investment relationships in 2023–2025 has given way to a strategy routed almost entirely through Starwood. Roughly 90% of non-hosted capacity is 'being considered' for conversion (claim 9201). Much of the apparent Starwood volume is repetition rather than new deals: third-party restatements (claims 926, 7504, 8403, 7973, 8246, 8244, 1214) echo the same February agreement. Claim 9090 dates a 'launch' to February 1, weeks before the agreement was signed, which looks like a date error rather than a separate event.

The revenue arithmetic cannot be reconciled because no lease revenue, pricing or megawatts under contract has been disclosed. The capacity headline varies by source, from 'up to 2.5 gigawatts' (claims 926, 8403) to 'more than a gigawatt near term' (claim 2946). The lease-signing goal moved from 'one or more tenant leases by year-end' (claims 9198, 11582) to 'at least 2 leases before year end' (claim 8502). The target rose even though no tenant had been identified.

Marathon Digital's announced partnerships reveal substantial disclosure gaps between definitive commitments and forward-looking discussions. The Starwood Digital Ventures strategic agreement, executed February 26, 2026 (claims 3660, 3661, 3662, 3663, 3665, 3493, 3494, 3501, 3511, 3527), represents the sole major signed deal, establishing MARA's 10-50% ownership in joint ventures with Starwood as managing member developing up to 2.5 gigawatts of AI/HPC capacity. Named counterparties are limited: TAE Power Solutions (claim 9272, partnership rhetoric without revenue disclosure) and EDF Pulse Ventures (claim 9271, investment agreement in subsidiary Exaion). By contrast, Long Ridge tenant claims reference only "multiple potential investment-grade AI/Critical IT tenants" (claims 9414, 10655) and claims 7670 alludes vaguely to "partnerships with leading AI and grid management companies," neither identifying actual customers. The MPLX relationship remains unexecuted; claims 3513, 3524, 3503, and 3497 explicitly designate this a "letter of intent," while claim 11789 concedes West Texas discussions "remain a longer-term initiative" requiring "significant" capital.

Starwood dominance defines counterparty concentration. This single entity appears across roughly 30 claims spanning infrastructure, ownership, and tenant expectations, with MARA structurally subordinated as minority owner (10-50%) while Starwood controls managing member status and operational decisions. This concentration amplifies execution dependency on a single partner.

No revenue figures reconcile against announced partnerships. Claims 9198, 11582, and 8502 project future lease signings—"expectations" to sign "at least 2 leases" by year-end—without documenting executed customer contracts. Long Ridge, repeatedly cited as receiving "inbound interest" since April 2026, presents no named tenant despite sustained management signaling of imminent signings. Neither the Starwood joint venture nor Long Ridge generates disclosed revenue contribution. The persistent gap between announced strategic scope and actual customer capture—particularly given Starwood's structural dominance over MARA's AI/HPC strategy—signals material execution risk and opacity regarding whether marketed infrastructure capacity will translate to near-term revenue accretion.

Marathon Digital's disclosed partnership activity reveals a sharp distinction between formally executed agreements and preliminary discussions. The Starwood Capital Group/Starwood Digital Ventures strategic agreement, signed February 26, 2026 (claims 3660, 3493, 3501), constitutes MARA's sole named, signed customer-facing arrangement of material scale. This agreement specifies binding terms: joint venture structure with MARA retaining 10–50% ownership, Starwood as managing member, up to 2.5 gigawatts of capacity, and MARA's retention of existing mining rights via rent-free lease (claims 3662, 3663, 3665, 3666). EDF Pulse Ventures represents a signed investment agreement in MARA's Exaion subsidiary (claim 9271), though this is a capital transaction rather than a customer relationship. By contrast, TAE Power Solutions is described only as a "partner" for a grid efficiency platform (claim 9272) without indication of contract execution or binding terms. The MPLX LP arrangement is explicitly styled a "letter of intent" (claims 3513, 3524), signaling pre-contractual status and no firm commitment. Tenant-level customer acquisition remains entirely unnamed. MARA repeatedly emphasizes inbound interest from "multiple potential investment-grade AI/Critical IT tenants" (claims 9414, 10655) and "potential High-Performance Computing tenants" (claim 8906), yet no specific customers are identified. Management expressed confidence in signing "at least 2 leases before year end" 2026 (claim 8502), and separately committed to "one or more tenant leases by year-end" (claim 9198), but these are forward-looking aspirations, not executed deals. Counterparty concentration has consolidated dramatically around Starwood, which accounts for the majority of capacity and partnership announcements by mid-2026, effectively replacing diversified small-scale partnerships. Revenue reconciliation presents a material gap: MARA discloses no customer revenue attributable to any announced deal. Infrastructure announcements emphasize capacity targets (2.5 GW under Starwood) and Long Ridge's 70% contracted power (claim 8504), but fail to translate these into named customer contracts or revenue guidance. The repeated emphasis on tenant-signing targets by year-end, coupled with the complete absence of named customers or revenue confirmation, suggests that despite high-profile Starwood execution, MARA's underlying revenue-generating customer base remains undisclosed and possibly not yet committed.

Marathon Digital has disclosed a limited set of definitively signed customer agreements versus substantially more material announcements framed as partnerships or frameworks. The Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures (claims 3660, 3511) constitutes the sole signed major infrastructure partnership, with Marathon retaining 10-50% ownership in joint ventures while Starwood serves as managing member and handles day-to-day development (claims 3663, 3665). By contrast, the 2026-03-01 Letter of Intent with MPLX LP for power access (claims 3513, 3524) explicitly represents non-binding dialogue rather than committed offtake or capacity. The EDF Pulse Ventures investment in the Exaion subsidiary (claim 9271) is described as signed but narrowly scoped to capital deployment, not operational customer relationships. The TAE Power Solutions partnership (claim 9272) contains no deal-structure language and reads as a joint product announcement absent contractual specificity.

Counterparty concentration has become acute. Starwood dominates Marathon's announced partnership narrative, appearing across 20+ claims spanning February through August 2026, effectively collapsing multiple announcements into a single underlying Strategic Agreement. By contrast, Marathon names no operational AI or HPC tenants; all references to "investment-grade AI/Critical IT tenants" (claims 9414, 10655) or "potential High-Performance Computing tenants" (claim 8906) remain unnamed and described as "inbound interest" or "potential" rather than signed leases. The Long Ridge facility announcements (claims 9414, 935) project a tenant placement by year-end 2026 but cite no executed agreement as of August 2026. The August 2026 statement that Marathon "remain[s] confident in [its] ability to sign at least 2 leases before year end" (claim 8502) restates an expectation, not a closed deal.

No claims disclose revenue, contract values, or financial contribution from any announced partnership or potential tenant, rendering reconciliation impossible. The disclosed metrics focus exclusively on capacity (gigawatts) and ownership percentages within joint ventures (claims 3662), not realized or projected customer revenue. This represents a material gap between partnership announcements and financial substance—the firm is announcing infrastructure framework agreements and stating lease-signing intentions without quantifying customer commitments or revenue streams underlying its AI/HPC pivot.

Marathon Digital's disclosed partnership and customer announcements reveal a significant gap between stated capacity development and actual signed revenue-generating agreements. Of the named counterparties, only Starwood Digital Ventures and Starwood Capital Group represent a formalized contractual relationship—structured as a strategic agreement (claims 3660, 3493, 3501) establishing joint ventures where MARA retains 10–50% ownership interest while Starwood manages operations and engineering. This arrangement is fundamentally a capital and development partnership, not a customer agreement. EDF Pulse Ventures represents an investment in MARA's Exaion subsidiary (claim 9271), which is financial in nature. TAE Power Solutions is described only as a 'partner' for a grid efficiency platform (claim 9272) without contractual detail. Critically, the MPLX LP relationship remains a letter of intent as of March 2026 (claims 3513, 3503, 3524)—not signed—despite repeated claims framing it as an announced commitment.

The actual end-customer landscape is almost entirely unnamed. MARA describes receiving 'inbound interest from multiple potential investment-grade AI/Critical IT tenants' at its Hannibal facility (claims 9414, 10655) and 'interest from potential High-Performance Computing tenants' at other sites (claim 8906), but provides no customer names or signed lease commitments. Forward-looking statements reference expectations to 'sign one or more tenant leases by year-end' (claim 9198, dated May 2026) and confidence in signing 'at least 2 leases before year end' (claim 8502, dated August 2026)—these are forecasts, not closed deals.

Counterparty concentration is extreme. From February 2026 onward, the Starwood partnership dominates every major announcement, with 26 distinct claim IDs referencing the same February 26 strategic agreement and its ongoing implications. No other customer or partner achieves comparable visibility. Revenue reconciliation is impossible: no revenue figures are disclosed anywhere in the dataset, and no signed customer agreements are named. The disclosed arithmetic centers on capacity targets (2.5 gigawatts via Starwood, claims 7504, 926, 7505, 8403) and power contracted under long-term agreements at Long Ridge (over 70%, claim 8504), but these do not constitute actual data center tenant revenue. MARA's expansion narrative rests substantially on prospective customer relationships and named-partner development capacity rather than executed customer contracts.

Marathon Digital's customer and partnership disclosures over 2025-2026 reveal a sharp distinction between signed agreements and preliminary commitments. Only two material deals qualify as executed contracts: the EDF Pulse Ventures investment agreement in the Exaion subsidiary (claim 9271, signed 2025-08-11) and the Starwood Capital Group / Starwood Digital Ventures strategic agreement (claims 3660-3663, executed February 26, 2026). The Starwood deal includes specific operational terms—MARA retains 10-50% ownership in joint ventures, Starwood serves as managing member, and detailed pre-development obligations are enumerated. By contrast, TAE Power Solutions (claim 9272) is described only as a "partnership" without evidence of a signed agreement, and MPLX LP (claims 3513, 3524, 3503, 3497) is explicitly identified as a "letter of intent," not a binding contract. Across claims dated March-August 2026, MARA repeatedly references "multiple potential investment-grade AI/Critical IT tenants" (claims 9414, 10655) and expects "one or more tenant leases by year-end" (claim 9198, 8502), yet no tenant names appear in the record. Claims 9201 and 8501 shift language to describe Starwood as enabling customer "confidence in execution" rather than guaranteeing customer wins. Counterparty concentration is extreme: Starwood dominates 16 of the 41 claims, appearing in every major announcement from February onward (claims 2946, 3660-3666, 3501, 3511, 3527-3528, 8403, 8246, 1214, 603, 7557, 7505). This concentration intensifies after the February strategic agreement. Revenue reconciliation is impossible because no financial figures are disclosed. Claims reference "2.5 gigawatts of capacity," "70% of Long Ridge's power output contracted" (claim 8504), and expected lease signings, but no dollar values, AUM, or revenue numbers appear. The pattern suggests MARA has one signed infrastructure partnership (Starwood), one signed capital partner (EDF), one preliminary supply commitment (MPLX LOI), and zero named revenue-generating tenant customers as of August 2026, despite 18 months of announcements.

Marathon Digital's partnership announcements from 2025 through mid-2026 reveal a sharp distinction between signed infrastructure agreements and prospective customer commitments. The February 26, 2026 Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures (claims 3660, 3511, 3493, 3501) represents the sole large-scale, formally executed arrangement on record. That deal specifies MARA will retain 10–50% ownership in joint ventures targeting up to 2.5 gigawatts of AI and HPC capacity, with Starwood as managing member and responsible for development, permitting, and procurement (claims 3662, 3663, 3665). This is architecturally a capital partnership, not a customer contract. By contrast, the March 1, 2026 announcement involving MPLX LP is explicitly a letter of intent (claims 3513, 3524, 3497, 3503)—an early-stage framework lacking binding commitment. Earlier partnerships with TAE Power Solutions (claim 9272, June 2025) and an investment agreement with EDF Pulse Ventures in Exaion (claim 9271, August 2025) involved named counterparties but remain vaguely described as partnership and investment arrangements without customer-revenue clarity. The narrative's largest void concerns tenant relationships at the Long Ridge facility. MARA disclosed "inbound interest from multiple potential investment-grade AI/Critical IT tenants" (claims 9414, 10655, April 2026) and stated an expectation to sign "one or more tenant leases by year-end" (claim 9198, May 2026; claim 8502, August 2026). No tenant names appear in any claim; these remain prospective and unsigned as of the August 2026 cutoff. Separately, claim 8504 reports that 70% of Long Ridge's power output is "contracted under long-term agreements" but identifies no counterparties. Counterparty concentration is extreme: Starwood dominates the disclosed framework across 23+ claims spanning infrastructure development and customer confidence rhetoric. MARA has publicly announced zero named customer contracts and only one LOI with MPLX. Revenue reconciliation is impossible. The company reports no customer revenue figures against the announced capacity, partnerships, or tenant pipelines. Starwood appears to be a capital vehicle and development collaborator rather than a revenue-generating customer, and the Long Ridge tenant pipeline remains unbooked and unidentified. The entirety of MARA's commercial disclosures consists of infrastructure partnerships and prospective capacity rather than confirmed revenue agreements.

Marathon Digital's partnership announcements from June 2025 through August 2026 conflate three distinct commitment levels, creating narrative fog around actual customer wins. The EDF Pulse Ventures investment agreement of August 2025 (claim 9271) represents a true signed deal—capital deployment into the Exaion subsidiary. The Starwood partnership, announced February 26, 2026 and reiterated through August (claims 3660, 3501, 3527, 8501), is also signed but structurally distinct: MARA contributes properties and retains 10–50% ownership in joint ventures managed entirely by Starwood (claims 3662–3663), making this a capital and operational control handoff rather than a customer relationship. In contrast, MARA's March 1, 2026 announcement with MPLX LP is explicitly a letter of intent (claims 3513, 3524, 3503, 3497)—no signed agreement exists, only "reasonable best efforts" language around future negotiations for natural gas and power capacity. TAE Power Solutions, cited June 3, 2025 (claim 9272), uses vague "partner" language without clarifying whether a binding agreement exists. Long Ridge tenant discussions epitomize the vagueness: MARA reports "multiple potential investment-grade AI/Critical IT tenants" with "inbound interest" (claim 9414) and states confidence in signing "at least 2 leases before year end" (claim 8502), yet names no customers and treats lease signings as forward expectations rather than accomplished commitments (claims 9198, 935). Starwood dominance pervades the portfolio—appearing in 20+ claims from February onward—yet represents joint venture architecture, not contracted customer revenue. The entire disclosure lacks revenue reconciliation; no customer contract values, revenue guidance, or lineage between announced capacity (2.5 gigawatts from Starwood) and actual contracted customer commitments appear in any claim. TAE Power, EDF (confined to a subsidiary), and the unnamed Long Ridge tenants constitute the only distinct counterparties. MARA conflates partnership announcements with customer wins, frames aspirational tenant discussions as strategic achievements, and discloses zero customer revenue, making it impossible to reconcile the narrative of customer momentum with actual signed customer economics.

Marathon Digital's partnership announcements from mid-2025 through August 2026 reveal a sharp distinction between executed agreements and forward-looking commitments, with pronounced concentration risk emerging around a single counterparty. The company has disclosed two clearly signed agreements: an investment agreement with EDF Pulse Ventures in the Exaion subsidiary (claim 9271, August 2025) and a strategic agreement with Starwood Digital Ventures dated February 26, 2026 (claims 3660, 3511, 3493). The Starwood Strategic Agreement includes binding specifics—joint venture ownership capped at 10–50%, Starwood's role as managing member, contribution of identified US data center assets, and defined pre-development and operational obligations (claims 3661–3665)—distinguishing it from framework language. By contrast, the March 1, 2026 letter of intent with MPLX LP (claims 3513, 3524, 3503, 3497) is explicitly non-binding and targets only future negotiation of natural gas and power arrangements. TAE Power Solutions' June 2025 partnership (claim 9272) lacks contractual specificity and remains ambiguously described. The East Ohio Valley Railway arrangement (claim 1873) uses aspirational "best efforts" language characteristic of LOI rather than executed terms.

Counterparty concentration has shifted dramatically toward Starwood, which dominates announcements from February 2026 forward across at least twenty distinct claims—often restating the same February 26 announcement (claims 7504, 9267, 2946, 8403, 8501, 7505, 7557, 603, 579). This concentration extends to strategic positioning: claims 9201 and 8246 position Starwood as central to the company's pivot toward AI and HPC infrastructure. Notably, tenant commitments remain prospective: claims 9414, 8906 reference "potential" tenants with "inbound interest," while claims 935, 9198, 8502 describe "expectations" to sign leases by year-end 2026, using conditional language rather than naming executed counterparties. Generic references to "multiple potential investment-grade AI/Critical IT tenants" (claim 9414) and "leading AI and grid management companies" (claim 7670) disclose no named customers.

No revenue figures appear in the disclosed claims, making it impossible to verify whether announced megawatt deployments or joint venture structures reconcile to any quantified revenue guidance or backlog. The absence of named end-customers, combined with the reliance on Starwood as development partner rather than revenue customer, leaves the commercialization pathway opaque. Marathon's deal-making architecture appears heavily dependent on Starwood execution and tenant acquisition by year-end 2026, neither of which has been contractually consummated.

Marathon Digital's disclosed partnership announcements conflate several distinct deal structures that require careful separation. The company has signed a formal strategic agreement with Starwood Capital Group and its Starwood Digital Ventures platform, announced February 26, 2026 across claims 3660-3663, 3493-3494, and 3511. This is the only named customer agreement presented as signed. However, this is structurally a joint venture in which MARA contributes assets and retains 10-50% ownership while Starwood assumes managing member and operational control—it is not a traditional customer revenue agreement. EDF Pulse Ventures signed an investment agreement in MARA's Exaion subsidiary (claim 9271, August 2025), but this is a capital partner investment, not a customer engagement. TAE Power Solutions is described only as a partner (claim 9272, June 2025) with no substance on contractual commitment. MPLX LP is explicitly a letter of intent, not a signed agreement (claims 3513, 3524, 3497, 3503, March 2026).

The concentration risk is extreme. Starwood appears in over twenty claims spanning six months, dominating all announced AI and HPC infrastructure development. No other named counterparty approaches this significance. The company describes multiple potential investment-grade AI and Critical IT tenants (claim 9414, April 2026) and potential High-Performance Computing tenants (claim 8906, July 2026), but these remain prospects. May 2026 guidance expressed confidence in signing one or more tenant leases by year-end (claim 9198); by August this refined to at least two leases before year-end (claim 8502), suggesting none were signed at announcement. The Long Ridge facility cites more than seventy percent of power output contracted under long-term agreements (claim 8504, August 2026), but this refers to underlying power capacity, not customer data center leases.

No revenue figures are disclosed to reconcile against announced deal sizes or capacity targets (2.5 gigawatts cited repeatedly across claims 926, 7504, 7505). Generic references to leading AI and grid management companies (claim 7670, August 2026) obscure the actual customer concentration. The portfolio appears entirely dependent on Starwood's execution and tenant recruitment, with no independent customer commitment yet materialized into signed leases.

Marathon Digital's partnership landscape reveals significant gaps between contractual commitments and prospective engagement. The company has named four distinct counterparties with publicly disclosed relationships: Starwood Digital Ventures and Starwood Capital Group (strategic agreement announced February 26, 2026, claims 3660–3663), EDF Pulse Ventures in the Exaion subsidiary (investment agreement, claim 9271), TAE Power Solutions (partnership claim 9272), and MPLX LP (letter of intent, claims 3503, 3513). However, the binding force of these relationships diverges materially. The Starwood strategic agreement, detailed across claims 3493–3501, represents a signed binding arrangement specifying joint venture structures, ownership stakes between 10–50 percent, and defined roles for engineering and construction, making this an executed partnership. The EDF Pulse transaction similarly constitutes a signed investment agreement. By contrast, MPLX remains explicitly a letter of intent—a pre-contractual instrument with no binding obligation to close (claims 3524, 3497). The TAE Power Solutions relationship is described only as a partnership without documentation of execution or financial terms (claim 9272).

Counterparty concentration has undergone dramatic consolidation. Before March 2026, MARA disclosed relationships with multiple parties (TAE, EDF, MPLX) suggesting diversified partnerships. Since the February 2026 Starwood announcement, Starwood dominates MARA's public disclosures, appearing across dozens of claims through August 2026 (3493, 3501, 7504, 8501, 8403, 7465, 579). This concentration is compounded by reliance on unnamed prospective tenants. Claims 9414 and 8906 reference inbound interest from multiple potential AI and high-performance computing tenants but identify no specific counterparties. The Long Ridge facility tenant pipeline remains unnamed and conditional: claims 935 and 9198 state MARA expects and remains confident in lease signings by year-end 2026, but no tenant has been named or contracted. On revenue reconciliation, the disclosed claims contain no financial figures, contracted amounts, or pricing terms for any announced relationship. Capacity targets are stated—2.5 gigawatts for Starwood, 70 percent of Long Ridge power contracted (claim 8504)—but unmoored from revenue or profitability metrics, preventing verification that announced partnerships translate to material disclosed revenue impact.

Marathon Digital's partnership and customer announcements reveal a critical distinction between executed agreements and unsigned frameworks, alongside acute counterparty identification gaps. Of the named counterparties, only Starwood Digital Ventures and Starwood Capital Group represent a clearly signed strategic agreement, explicitly documented as entered into on February 26, 2026 (claims 3660, 3493, 3501, 3511, 3527). The EDF Pulse Ventures investment agreement (claim 9271) is signed, but it resides in the Exaion subsidiary rather than core MARA operations. TAE Power Solutions (claim 9272) uses non-binding partnership language without execution language. Most critically, MPLX LP is explicitly a letter of intent (claims 3513, 3503, 3497, 3524)—unsigned and non-binding, announced March 1, 2026 for natural gas and power access.

Counterparty concentration is dominated by Starwood across 24+ claims spanning February through August 2026. This single relationship encompasses up to 2.5 gigawatts of capacity (claims 7504, 926, 8403, 7505), joint venture structures with 10-50% MARA ownership (claim 3662), and Starwood as managing member controlling day-to-day operations (claim 3663). All other end customers remain entirely unnamed. MARA references "investment-grade AI/Critical IT tenants" (claim 9414), "potential High-Performance Computing tenants" (claim 8906), and vague "partnerships with leading AI and grid management companies" (claim 7670)—none with identified counterparty names. Expected lease signings cite targets of "one or more tenant leases by year-end" (claim 9198) and "at least 2 leases before year end" (claim 8502), yet provide no tenant identities.

Revenue reconciliation is impossible. No announced partnerships disclose revenue, deal values, or committed capacity pricing. Long Ridge shows 70% power output under "long term agreements" (claim 8504), but neither capacity volume nor pricing is disclosed. Starwood's 2.5 gigawatt target represents capability, not booked revenue. The absence of named end-customer contracts, combined with repeated aspirational language around tenant interest and lease signatures, suggests MARA's revenue-generating customer relationships remain unsigned or unannounced. The bulk of external communications describe partnership infrastructure and development frameworks rather than executed commercial agreements generating disclosed revenue.

Marathon Digital's partnership announcements reveal a stark concentration risk masked by volumetric claim repetition. Of genuinely binding customer agreements, only two qualify as signed: the February 2026 Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures (claims 3660, 3662-3666), and the August 2025 investment agreement with EDF Pulse Ventures in the Exaion subsidiary (claim 9271). The Starwood deal dominates the disclosure record—appearing across 25+ claims—yet consists of a joint venture structure where MARA retains only 10-50% ownership and cedes operational control to Starwood as managing member (claims 3662, 3663). This is not a customer agreement but a development partnership with diluted upside. The MPLX LP announcement (claims 3497, 3513, 3524) is explicitly a letter of intent, not binding, targeting natural gas and power access. TAE Power Solutions, named in June 2025 (claim 9272), receives no subsequent mention, suggesting dormancy or limited materiality.

Counterparty concentration has tightened sharply around Starwood since February 2026. Claims describing unnamed "partnerships with leading AI and grid management companies" (claim 7670) and "potential High-Performance Computing tenants" (claim 8906) remain entirely unidentified. The Long Ridge facility exemplifies this pattern: MARA promised tenant leases signed by year-end 2026 (claims 9198, 8502), yet as of August 2026, no named customers appear. Claims cite "70% of Long Ridge's power output contracted" (claim 8504) but provide no customer names or revenue figures.

The revenue arithmetic fails to reconcile with announced deal scale. MARA cites capacity targets (2.5 GW under Starwood; claims 7504, 8403) and power-contracting percentages, yet discloses no revenue recognition, customer identification for hyperscale tenants, or signed lease revenue impact. The Starwood agreement itself generates no direct customer revenue—instead defining a 10-50% minority ownership in joint ventures where Starwood controls engineering, procurement, and construction (claims 3665, 3666). This leaves MARA's actual end-customer pipeline entirely unnamed and aspirational. The distinction between a strategic partnership announcement and a revenue-generating customer agreement has been systematically blurred through repeated restating of the same underlying Starwood transaction across multiple claims without incremental customer wins or named tenant signings since the original February 2026 announcement.

Marathon Digital's disclosure record distinguishes sharply between framework commitments and customer revenue arrangements. The sole disclosed SIGNED counterparty is Starwood Capital Group (Strategic Agreement, February 26, 2026; claims 3660–3666). Critically, this is not a customer contract but a joint venture framework for infrastructure development targeting 2.5 gigawatts of AI and HPC capacity, with MARA retaining 10–50% ownership stakes and Starwood managing operations, procurement, and financing (claims 3661–3666, 7504). The partnership is real, but revenue-bearing customer leases remain undisclosed.

All other counterparty arrangements are explicitly non-binding or prospective. MPLX LP appears only as a "letter of intent" (claims 3503, 3513, 3524) aimed at natural gas and power access—not a commercial service contract. The East Ohio Valley Railway reference (claim 1873) constitutes only an agreement "to negotiate in good faith" on railroad assets. Beyond Starwood, Marathon discloses no named customers. Instead, the pipeline consists of unnamed tenant placeholders. As of August 13, 2026, management remained "confident in our ability to sign at least 2 leases before year end" (claim 8502); references to "potential High-Performance Computing tenants" (claim 8906) and unidentified "partnerships with leading AI and grid management companies" (claim 7670) provide no counterparty identification.

Counterparty concentration is extreme. Starwood Digital Ventures appears in over 30 claims spanning six months, dominating every material disclosure. No competing customers or anchor tenants are named.

Revenue reconciliation is impossible because no commercial pricing or volume data are disclosed. Claims reference "long term agreements" contracting Long Ridge power (claim 8504) and Starwood's role securing "hypfinancing" (claim 3494), but actual rates, volumes, contract terms, and end-customer identities remain absent. The capacity targets (2.5 GW) and JV ownership percentages provide structural indicators but no revenue validation. Without disclosed lease pricing, tenant commitments, or power contract economics, the arithmetic linking announced partnerships to realized revenue cannot be tested against these claims.

Marathon Digital's announced partnerships reveal a single signed customer agreement layered with letters of intent and unnamed prospective tenants, reflecting deepening reliance on Starwood while projected revenue remains opaque. The February 26, 2026 Strategic Agreement with Starwood Digital Ventures (claims 3660, 3661, 3662, 3663, 3665, 3666) stands as the sole substantive signed deal on record. It specifies Marathon's 10–50% ownership in joint ventures, designates Starwood as managing member and operational lead, and obligates Starwood to perform pre-development services, secure permitting and power arrangements, and deploy capital to develop up to 2.5 gigawatts of AI and HPC infrastructure. The agreement is referenced across multiple filings (claims 3511, 3493, 3501, 3527) and citations persist consistently through September 2026 (claims 2474, 2480), indicating no deterioration of the relationship.

In contrast, other commercial announcements fall structurally short. The March 2026 announcement with MPLX LP is expressly termed a letter of intent (claims 3503, 3497, 3513, 3524, 2476), remaining unsigned, and addresses power and natural gas sourcing rather than end-customer commitments. The Long Ridge Energy transaction remains prospective: Marathon "expects" to place a tenant by mid-2026 (claims 935, 6849), but by late August, management was only "confident" about signing "at least 2 leases before year end" (claim 8502), implying no signed lease as of September 1. The tenant is never named.

Counterparty concentration is extreme. Starwood is the only identified customer with a binding, material agreement. All remaining commercial initiatives feature unnamed prospective tenants, letters of intent, or generic partnership language. Claims 8906 reference "interest from potential HPC tenants" without naming a single tenant. Claim 7670 invokes "partnerships with leading AI and grid management companies" without identification. Exaion surfaces only as a vague reference to "sovereign enterprise AI hosting" (claim 2504) with no disclosed contract architecture.

Revenue reconciliation is notably absent. Marathon has not disclosed lease rates, utilization assumptions, or expected Marathon cash contribution linked to any partnership announcement. The Starwood agreement specifies governance and ownership but omits pricing, lease term, or cash generation expectations. Long Ridge's unsigned, unnamed tenant status precludes financial modeling entirely. This silence on monetary linkage—the absence of connection between announced infrastructure capacity and expected cash flows—leaves material uncertainty about whether commercial announcements materialize into revenue.

Marathon Digital's partnership and customer announcements collapse to a single named counterparty with binding obligation: Starwood Digital Ventures, via a Strategic Agreement executed February 26, 2026 (claim 3660). That agreement is unambiguously signed—the language specifies MARA's 10–50% ownership stake in joint ventures, Starwood's role as managing member, designated asset schedules, and MARA's retained mining rights (claims 3662, 3663, 3661, 3666). The Starwood deal targets up to 2.5 gigawatts of AI and HPC infrastructure over time (claims 7504, 926), with near-term capacity of more than 1 gigawatt (claim 2500).

Everything else is either pre-binding or entirely unnamed. The MPLX LP announcement (March 1, 2026) is explicitly a letter of intent—framework language for potential access to natural gas and power capacity, not a signed contract (claims 3513, 3503, 3497, 3524). The Long Ridge facility has 70% of power output under long-term contracts, yet zero named tenants appear in any claim (claim 8504); MARA states confidence in signing "at least 2 leases before year end" but provides no customer identities (claim 8502). A single reference to "partnerships with leading AI and grid management companies" supplies no names whatsoever (claim 7670). HPC tenant interest at the Long Ridge site is noted as "received interest from potential" counterparties—interest, not commitment (claim 8906).

Counterparty concentration has been severe and stable throughout. Starwood dominates from February onward; by August–September 2026, after six months, the company still has no second named partner with binding terms. MPLX remains an LOI; Long Ridge tenants remain unnamed prospects. The only material diversification attempt—the MPLX letter of intent—achieved no executed agreement.

The announced deals disclose capacity targets and ownership percentages, not revenue. Starwood contributions are structurally equity (MARA takes 10–50% stakes in joint ventures) rather than direct revenue. The claims reference "contracted" power at Long Ridge but supply no pricing, term lengths, or revenue attributions that would permit arithmetic reconciliation with MARA's reported financials. The partnership announcements emphasize scale and strategic positioning; they do not reconcile to identifiable, quantifiable customer revenue streams.

Marathon Digital's disclosed partnership portfolio as of September 2026 reveals a sharp divide between named signed agreements and prospective or preliminary arrangements that lack counterparty identification. The Starwood Capital Group partnership stands as the only material named counterparty with confirmed execution; entered into on February 26, 2026, it represents a signed Strategic Agreement for joint AI and HPC infrastructure development across up to 2.5 gigawatts of capacity, with near-term deployment targeting 1+ gigawatt (claims 3660, 2500, 2501). Starwood's role is clearly defined, including ownership stakes between 10–50% per joint venture, operational management responsibilities, and pre-development obligations (claims 3660, 3663, 3665, 3662). The MPLX LP relationship, by contrast, remains preliminary. Announced March 1, 2026, it is explicitly a letter of intent rather than a signed customer agreement, aimed at expanding access to natural gas and scalable power capacity (claims 3503, 3497, 3513). No MPLX-branded revenue or capacity commitments have materialized on record. More concerning is the prevalence of unnamed counterparties masquerading as strategic accomplishments. Marathon expects a tenant at the Long Ridge facility by year-end 2026 but has not disclosed the tenant's identity, despite describing this entity as a hyperscaler (claims 935, 6849). Similarly, the firm reports receiving interest from potential High-Performance Computing tenants without naming any of them (claim 8906). References to partnerships with leading AI and grid management companies remain similarly vague (claim 7670). These unnamed relationships, whether exploratory or near-closing, inflate the appearance of commercial traction while obscuring concentration risk and execution probability. Revenue arithmetic does not reconcile in any claim reviewed. Marathon discloses capacity targets and power contracted percentages—70% of Long Ridge's output under long-term agreements (claim 8504)—but provides no megawatt pricing, utilization assumptions, or revenue derivations to connect infrastructure capacity to actual customer economics. The lack of quarterly lease signings, named hyperscaler commitments, or revenue guidance tied to Starwood or any other named party prevents triangulation of claimed pipeline value. Counterparty concentration has intensified rather than diversified over the February–September period, with Starwood remaining the sole named, signed strategic engine for growth while all material customer relationships remain anonymous or uncommitted.

Marathon Digital's partnership disclosures from February through September 2026 reveal a heavily concentrated counterparty landscape dominated by a single named entity paired with preliminary or forward-looking commitments from unnamed actors. The Starwood Capital Group and Starwood Digital Ventures partnership, announced February 26, 2026, constitutes the only formally executed strategic agreement on record (claims 3660, 3493, 3501, 3527). The agreement's structure—documented through claims 3661–3666—shows MARA contributing specified U.S. bitcoin mining data centers in exchange for 10–50% ownership stakes in joint ventures, with Starwood managing day-to-day operations and performing pre-development services including permitting and power procurement. Claim 2480 acknowledges the agreement subjects MARA to "significant development, execution, financing and counterparty risks." By contrast, the MPLX LP engagement exists only as a letter of intent announced March 1, 2026 (claims 3503, 3513, 3524, 3497, 2476), explicitly framed as aimed at expanding access to natural gas and power capacity without binding commitment language.

The Long Ridge Energy asset (claims 935, 6849, 8502, 8504) illustrates the gap between announced capacity and contracted revenue. While claim 8504 states over 70% of Long Ridge's power output is "contracted under long-term agreements," no customer names or revenue amounts are disclosed. Claims 935 and 8502 reference expectations of "a tenant" and confidence in signing "at least 2 leases before year end"—forward-looking expressions of intent rather than completed deals. The mysterious "leading AI and grid management companies" (claim 7670) receive no identification whatsoever.

The concentration is striking: Starwood appears in 27+ claims spanning the full February–September period, while all other named counterparties appear in single digits. Unnamed or vaguely described parties dominate the tenant and partnership narrative. On revenue reconciliation, the disclosures offer capacity targets—2.5 gigawatts total, 1 gigawatt near-term (claims 2500, 2501)—but no revenue figures, pricing, or deal values. The absence of financial metrics alongside capacity announcements makes it impossible to assess whether announced megawatt deployments will translate to material revenue. The portfolio presents as heavily dependent on Starwood execution and unnamed tenant fulfillment.

Marathon Digital's partnership announcements from February through September 2026 reveal heavy concentration on a single named counterparty with material legal distinctions obscured by promotional language. The company signed a Strategic Agreement with Starwood Capital Group and its platform Starwood Digital Ventures on February 26, 2026 (claims 3660, 3662, 3663), creating joint ventures structured as minority positions: MARA retains 10 to 50 percent ownership while Starwood serves as managing member and controls day-to-day development (claims 3663, 3494). This is a definitive signed commitment, albeit one where MARA's upside is capped by its minority stake. The partnership targets more than one gigawatt near-term and a 2.5 gigawatt pathway over time (claims 2500, 2501).

Other claimed partnerships carry weaker legal standing. MARA announced a letter of intent with MPLX LP for natural gas and power access (claims 3513, 3524, 2476)—explicitly an intention to negotiate, not a binding agreement. The Long Ridge Energy facility is said to expect "a tenant in place around the time the deal closes, in the second half of the year" (claim 935), yet the customer remains unnamed. MARA references "advanced discussions" with unidentified "investment-grade tenants" (claim 2524). A later Exaion reference to "sovereign enterprise AI expansion" (claim 2504) adds a second named entity without clarity on agreement status. Concentration risk is acute: Starwood dominates substantive, signed commitments while other counterparties are either LOIs, prospective unnamed tenants, or vague partnership descriptions.

Revenue reconciliation cannot be verified. The disclosed claims attach no customer-specific revenue figures, lease terms, or capacity utilization assumptions. The Starwood joint venture's minority structure means MARA captures only its percentage share of any infrastructure economics, with further dilution if development stalls. The unnamed Long Ridge tenant and unidentified investment-grade tenants create a visibility gap between announced gigawatt targets and actual signed customer commitments. Without matching customer identities to capacity milestones and underlying lease pricing, the claimed 2.5 gigawatt pathway cannot be assessed as a revenue-bearing asset pathway.

Marathon Digital's partnership disclosure distinguishes sharply between one binding agreement and several non-binding frameworks. The Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures, signed February 26, 2026, constitutes the only named, signed customer arrangement on record (claims 3660–3665, 3501, 3511, 3527, 4937, 2474, 2480). This agreement grants MARA a 10–50% ownership stake in joint ventures where Starwood will serve as managing member and lead engineering, procurement, and construction—a structure that transfers operational control to the counterparty despite equity participation. By contrast, the March 1, 2026 announcement with MPLX LP is explicitly a letter of intent (claims 3503, 3497, 3513, 3524, 2476), not a signed agreement, and targets only "expanded access" to natural gas and power capacity with no binding obligation. The Long Ridge Energy facility tenant, described as a "hyperscaler" with expected placement in H2 2026 (claims 935, 6849), remains unnamed and unfunded—this is anticipation, not a signed customer. East Ohio Valley Railway's rail assets involve parties agreeing to "negotiate in good faith" (claim 1873), language that signals ongoing discussions rather than a closed transaction. Beyond Starwood's dominance across roughly 20 separate claims, MARA disclosed "partnerships with leading AI and grid management companies" (claim 7670) and "advanced discussions with investment-grade tenants" (claim 2524) without naming any counterparties, rendering these partnerships impossible to verify. The claim referencing "Starwood Property Trust" (claim 6104) introduces ambiguity about whether this entity differs from Starwood Capital Group and Starwood Digital Ventures or represents the same relationship under a different corporate identity. Counterparty concentration is severe: Starwood absorbs all named partnership activity, while prospective customers are either unnamed or conditional. No revenue figures appear in any claim, preventing arithmetic reconciliation between announced capacity targets (1+ gigawatt near-term, 2.5 gigawatt long-term) and actual customer commitments or revenue recognition. The portfolio shows a framework partnership with one named player controlling day-to-day operations, an unexecuted LOI, and several unnamed or exploratory relationships.

Marathon Digital's partnership disclosures reveal a stark distinction between binding commitments and preliminary discussions. The company has signed one major Strategic Agreement—with Starwood Capital Group and its platform Starwood Digital Ventures, executed February 26, 2026 (claims 3660, 926, 3501, 2474). This is the only named counterparty with explicit signed documentation; the agreement targets 1–2.5 gigawatts of AI and HPC infrastructure capacity, with MARA retaining a 10–50% ownership interest in joint ventures under Starwood's management (claims 3662, 3528, 2502). By contrast, the MPLX LP relationship remains at letter-of-intent stage (claims 3497, 3513, 2476), structured as exploratory access to lower-cost natural gas and power—no signed operating agreement.

Beyond Starwood and MPLX, counterparty disclosure becomes hazy. The company describes "advanced discussions with investment-grade tenants" (claim 2524) for its Long Ridge Energy facility without naming any tenant; announcements state MARA "expects to have a tenant in place" by mid-2026 (claims 935, 6849), but this is an expectation, not a binding agreement. Exaion appears in claims 2504 as part of "sovereign enterprise AI expansion via hosting" but without contract-level detail. A "separately announced AI data center partnership with Starwood Property Trust" (claims 6351, 6104) is mentioned by name yet lacks operational specifics. Vaguest of all are references to "partnerships with leading AI and grid management companies" (claim 7670)—no named counterparties whatsoever.

Counterparty concentration has tightened dramatically around Starwood, which dominates 25+ claims from February through September 2026. Early roadmap included MPLX (still LOI-stage), but Starwood anchors the disclosed pipeline. On revenue reconciliation: no customer revenues are disclosed across these claims. Capacity targets are contingent (claim 2480 acknowledges "significant development, execution, financing and counterparty risks"), and ownership economics remain unspecified—MARA's 10–50% stake in Starwood JVs and a separately mentioned 35,577-bitcoin treasury (claim 1214) do not translate to disclosed operating revenue. Without signed tenant commitments or Starwood carrying revenue, the announced capacity pathway remains aspirational rather than revenue-producing.

Marathon Digital's announced partnership portfolio demonstrates material concentration risk and linguistic slippage between signed agreements and preliminary engagement. The February 26, 2026 Strategic Agreement with Starwood Capital Group and Starwood Digital Ventures (claims 3660, 3661, 3493) is a signed agreement governing joint venture formation for AI and HPC infrastructure development, with MARA retaining 10–50% ownership (claim 3662) and Starwood as managing member overseeing day-to-day operations (claim 3663). This is categorically distinct from the March 1, 2026 letter of intent with MPLX LP (claims 3513, 3524), which targets expanded access to power and gas capacity rather than committed facility leases or consumption commitments.

Counterparty concentration presents acute dependency risk. Starwood entities dominate disclosed partnerships: the Strategic Agreement covers MARA's entire U.S. portfolio excluding third-party joint ventures (claim 3661), targets over 1 gigawatt near-term with a 2.5-gigawatt pathway (claims 2500, 2501), and subordinates operational control to Starwood. Starwood Property Trust appears separately in August 2026 disclosures (claims 6351, 6104), but its contractual relationship to the February agreement remains opaque. Beyond Starwood and MPLX's LOI, no tenant counterparties appear in named form.

Long Ridge illustrates the execution-framework gap. MARA expects to secure a tenant "around the time the deal closes, in the second half of the year" (claims 935, 6849)—expectation language, not commitment. By September 2026, this deteriorated to "advanced discussions with investment-grade tenants" (claim 2524), terminology indicating pre-LOI stage. Exaion receives single mention in connection with "sovereign enterprise AI" hosting (claim 2504), with no disclosure of executed status, capacity terms, or revenue commitment. Claims referencing "partnerships with leading AI and grid management companies" (claim 7670) withhold counterparty identity entirely, blocking concentration assessment.

No revenue figures appear in the claim set, precluding reconciliation of announced infrastructure capacity against disclosed customer economics. The disclosed gigawatt pathways and percentage ownership structures exist alongside an absence of named end-customer commitments, obscuring whether capacity targets reflect binding demand or optionality. The record suggests MARA has secured framework infrastructure arrangements (Starwood, MPLX LOI) but has not disclosed signed tenant agreements sufficient to validate revenue realization against infrastructure development spend.

The record supplied here ends abruptly in May 2021 at claim 14648, so this read covers Marathon's mining-era buildout only and cannot speak to the Texas, Ohio, North Dakota or Nebraska sites now listed as operational or announced. The 2013-2014 entries, including the country-footprint figures (20897, 20898) and the patent-portfolio acquisitions, date from the company's patent-licensing predecessor and say nothing about facility delivery. They should be treated as legacy noise, and the country-count items look misattributed. On scale, the trajectory runs from a single leased Quebec building of roughly 26,700 square feet (15230) to a 240 PH/s fleet target (14906, 14887, 14871), which was raised within weeks to 294 PH/s (14858). It then moves to a 100 MW Hardin plan (14789). Hardin's headline total held at 100 MW, but its phasing was quietly re-cut. Phase I went from 30 MW (14790) to 70 MW, with a 30 MW Phase II (14645, 14646). The project now sits in the cancelled column with no closing claim explaining the exit. Date discipline on hardware was good where it was tested. The mid-August 2020 S19 Pro and MicroBT deliveries (14873, 14854, 14872) and the 10,000-unit order due by March 2021 (14762) are all marked verified. The fourth-quarter 2020 batch of 1,000 S19 Pros (14855) was never confirmed either way. The Quebec lease shows a quieter inconsistency. It runs through November 2022 and is marked abandoned under the Quebec entry (15229), yet the same lease remains open under Granby (15219). That suggests one site booked twice, with only one copy closed out. The SelectGreen miner purchase is similar. It was recorded as withdrawn (15062) and later restated as completed (14851, 14883), with no acknowledgement of the reversal. Independent corroboration is thin. Nothing in this record ties Hardin to an interconnection filing, a Montana permit or site reporting, and every project row carries blank capacity and cost fields. The Hardin megawatt figures therefore rest entirely on company statements. Over the next 90 days, the priority is completing the claim history past mid-2021 so that the current portfolio can be scored at all. After that, the record should be checked against ERCOT queue entries for MAT 1177 and Matagorda County, Ohio siting and PUCO dockets for Long Ridge and Coshocton, and any Plano filings. Each announced site should also be checked for whether it ever acquires a stated megawatt figure and an in-service date that can be tested.

The record supplied here ends in May 2021, partway through the Hardin disclosures at claim 14648, so this read covers Marathon's early mining build-out rather than its current portfolio. Claims from 2013 and 2014 reflect the predecessor patent-licensing business. Two of them, 20898 and 20897, describe African and 26-country operations, and 20896 and 20895 describe UN Women and EY partnerships. None of these fit Marathon's history, and they look misattributed. They should be audited before any editorial use. Announced capacity grew steadily in the 2020 hardware cycle. The fleet target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858) within weeks, and the 10,500-unit S-19 Pro order (14842) and the 10,000-unit order (14762) scaled ambitions well beyond that. Delivery promises in this period mostly held. The mid-August 2020 shipments (14873, 14872, 14854) and the March 2021 S-19 Pro tranche (14762) are verified. The Q4 2020 commitment for 1,000 S-19s (14855) was never resolved either way. Hardin is the clearest case of quiet revision. In October 2020 it was framed as a 30 MW Phase I within a 100 MW target (14790, 14789). By May 2021 the same site was presented as a 70 MW Phase I plus a 30 MW Phase II (14645, 14646). The total held at 100 MW, but the phasing was more than doubled at the front end without any stated rationale. The project list now shows Hardin as cancelled, and nothing in the supplied claims explains when or why. Several data-hygiene problems also stand out. The Quebec lease (15229) is marked abandoned, while the parallel Granby lease (15219), with its November 2022 end date long past, still reads not_yet_verifiable. The SelectGreen purchase is tagged withdrawn in 15062 but restated as completed in 14883 and 14851. On independent evidence, the record contains no permits, interconnection queue entries or site reporting, so the pace of any facility is supported only by Marathon's own statements. That gap matters most for the projects that are still only announced: Long Ridge, Coshocton, Matagorda County, MAT 1177 and Plano. None of them carries a capacity or cost figure. Over the next 90 days, the priorities are ingesting post-2021 claims and attaching megawatt figures and in-service dates to the announced sites. Those figures should be checked against PJM queue positions for the Ohio sites and ERCOT large-load filings for the Texas sites, and the status labels on Hardin, Granby and SelectGreen should be reconciled.

The record as supplied ends in May 2021, mid-entry at claim 14645. Every capacity and timeline claim here therefore predates the projects that define Marathon today: Long Ridge, Coshocton, Matagorda County, MAT 1177 and Plano are all listed as announced with no capacity or cost attached. That gap is the main finding, and any read of the current pipeline should wait for the later claims. There is also a data-hygiene problem. The 2013-2014 entries cover the patent-licensing predecessor, which is fair context. But claims 20895-20898, the UN Women and EY partnerships and the operations across 26 countries with an African presence in 19, do not fit any Marathon business line and look like an entity-resolution error. They should be excluded before publication.

Within the window available, announced scale moved upward in small steps rather than shrinking. The fleet hashrate target went from 240 PH/s (claims 14906, 14887, 14871) to 294 PH/s (14858) within about six weeks of summer 2020. That increase was backed by specific purchase orders. Near-term delivery promises held: the mid-August 2020 S19 Pro and MicroBT deliveries (14872, 14873, 14854) and the 10,000-unit order due by March 2021 (14762) are all marked verified. Claim 14062 is wrong as written; the withdrawn SelectGreen purchase is 15062, and it was later restated and completed as 14851.

The site-level picture is weaker. Hardin was framed as a 100 MW target with a 30 MW first phase (14789, 14790) and a second 30 MW phase priced at 9 million USD (14646). It now carries a cancelled status. The record holds no claim acknowledging that the 100 MW figure was reduced or dropped, so the revision appears to have happened without notice in this dataset. Quebec shows the same pattern: the lease claim 15229 is tagged abandoned with no explanatory filing. Claim 15219, the Granby lease due to end in November 2022, is still not_yet_verifiable nearly four years after that date, which should be closed out. Nothing in the record offers independent corroboration, whether permits, interconnection-queue entries or site reporting. Every capacity figure here is self-reported.

For the next 90 days, the priority is to backfill the claim history after May 2021 and attach capacity figures to the announced Texas and Ohio sites. Those sites should be checked against ERCOT and PJM queue filings and against county permit records. Any later Hardin or Quebec restatements in 10-K filings would also clarify whether these exits were ever disclosed as such.

The supplied record ends mid-2021, so this read covers Marathon's transition from patent holder to hosted miner and its first attempt at owned infrastructure. It cannot speak to the Texas, Ohio, North Dakota or Abu Dhabi sites now listed. Several of those, including MAT 1177, Long Ridge, Matagorda County, Coshocton and Plano, carry no capacity or cost figures at all, which is itself the most notable gap. The 2013-2014 entries also need cleanup before any trend is computed. Claims 20898, 20897, 20896 and 20895, covering African operations and UN Women and EY partnerships, appear to belong to a different MARA entity. The patent acquisitions are duplicated, for example 21015 with 21028 and 21016 with 21029.

On capacity, the trajectory scaled up and then partly collapsed. The Granby hosting target held at 240 PH/s across three statements (14906, 14887, 14871), then rose to 294 PH/s (14858) once the larger S-19 Pro orders landed (14842, 14762). Hardin was framed as a 100 MW site (14789) with 30 MW phases (14790, 14646) on a Beowulf power contract (14647). Both Hardin and Quebec now sit as cancelled, and no claim in the record acknowledges either cancellation or retires the 100 MW figure.

On timing, the hardware deliveries held: 14872, 14873, 14854 and 14762 all verified on schedule. Facility commitments fared worse. The Quebec lease through November 2022 is tagged abandoned in 15229 but not yet verifiable in 15219, apparently because the same lease is attributed to two different project records. The SelectGreen purchase was marked withdrawn (15062) yet restated as completed in 14851 and 14883. The Q4 2020 delivery of 1,000 S-19 units (14855) was never closed out.

The record contains no independent corroboration of any kind, such as permits, interconnection queue positions or site reporting. Every facility figure here rests on company statements.

Over the next 90 days, the third-quarter 10-Q due in early November is the main test. Watch whether it attaches megawatt figures to the announced Texas and Ohio sites, and whether Long Ridge and Matagorda County appear in PJM or ERCOT interconnection queues at sizes consistent with any disclosed capacity. Also check whether Plano and Coshocton gain local permit filings or quietly drop out of disclosures, as Hardin did.

The record supplied here ends mid-2021, at claim 14650, so this read covers only Marathon's transition from patent holder to miner. It cannot speak to the Texas, Ohio, Nebraska, North Dakota or Abu Dhabi sites that now dominate the project list. Several pre-2019 entries also look misattributed, including the 26-country footprint (20897), the African presence (20898) and the UN Women and EY tie-ups (20896, 20895). Those should be checked before publication. Every project row also carries a blank capacity field, so there is no current baseline to measure against.

Within that window, the announced scale grew quickly but mostly through procurement, not facility build. The fleet hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858) within weeks. It then jumped well past that with the 10,500-unit S-19 Pro order (14842). The 10,000-unit tranche due by March 2021 is marked verified (14762), as are the August 2020 deliveries (14873, 14854, 14872). On hardware, then, stated dates largely held. The SelectGreen purchase is the exception: it was withdrawn (15062) and later restated as a completed transaction (14883, 14851) with no explanation in the record.

Facility claims are weaker. Hardin was introduced as a 100 MW site with a 30 MW first phase (14789, 14790). Seven months later Phase I was restated as 70 MW and Phase II as 30 MW (14645, 14646). Total capacity was unchanged, but the near-term commitment more than doubled with no acknowledgement of the change. Hardin is now listed as cancelled, and no dated completion promise for it appears anywhere in the excerpt. In Quebec, the lease running to November 2022 (15219, 15229) is marked abandoned on one record and not yet verifiable on the other. The Quebec project itself is cancelled. That leaves Granby's 26,700 square feet (15230) as the only physical footprint claim with a stated size.

Nothing in this record ties any of these claims to independent evidence such as permits, interconnection queue positions, utility filings or site reporting. The verified tags apply to equipment deliveries, not to energized megawatts.

For the next 90 days, the priority is to load the full claim history so that the announced-status sites can be assessed: MAT 1177, Matagorda County, Long Ridge, Coshocton and Plano. For the Texas sites, look for ERCOT large-load interconnection entries. For Long Ridge and Coshocton, look for PJM or Ohio siting filings. The test is whether each first disclosure carries a dated megawatt figure, and whether later filings keep it or quietly reshuffle phases the way Hardin did.

The supplied record ends partway through May 2021, at claim 14646. This read therefore covers the Quebec and Hardin era only. It cannot speak to the projects that now define MARA's footprint: Long Ridge, Coshocton, Matagorda County, MAT 1177 and Plano are all still listed as announced, and none carries a capacity or cost figure. The 2013-2014 claims also need cleaning before publication. The patent acquisitions (21015 through 21035) fit the company's patent-licensing past, but the 26-country and African-expansion footprint claims and the UN Women and EY partnerships (20895-20898) look misattributed. On scale, the trajectory runs upward through 2020 and then gets reframed. The fleet hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within about three weeks (14858) as S-19 Pro orders grew. Hardin was announced at 30 MW for Phase I within a 100 MW total (14790, 14789). By May 2021 Phase I alone was described as 70 MW (14645), with no filing in the record explaining whether the overall 100 MW envelope changed. Hardin is now marked cancelled, so that capacity ultimately went to zero at that site. On timing, short-horizon delivery promises held: the August 2020 miner deliveries (14872, 14873, 14854) and the 10,000-unit S-19 Pro order due by March 2021 (14762) are all verified. Longer-dated commitments fared worse. The Quebec lease through November 2022 is recorded as abandoned (15229), while a parallel claim about the same lease is still tagged not yet verifiable (15219) almost four years after its date passed. That is a tracking gap worth closing. The SelectGreen equipment purchase was marked withdrawn (15062) and then restated as completed (14883, 14851) without any recorded explanation. The Q4 2020 delivery of 1,000 S-19 Pros (14855) has no verification attached. On independent evidence, the record contains none: no permits, no interconnection queue entries, no site reporting. Every capacity figure here is company-sourced. Over the next 90 days, the priority is to extend the record past May 2021 and attach announced MW figures to the open sites. Watch for ERCOT and PJM queue positions behind the Texas and Ohio announcements, county permit filings at Coshocton and Matagorda, and whether quarterly disclosures reconcile energized megawatts against the totals those announcements imply.

This record is truncated at claim 14648, so the read below covers Marathon's history only through mid-2021. It cannot speak to the Texas, Ohio, North Dakota or Abu Dhabi sites that now dominate the project list. The earliest entries also need caution. Claims 20898, 20897, 20896 and 20895, about African presence across 19 countries and UN Women and EY partnerships, do not fit Marathon's patent-licensing past and look like entity contamination. The 2014 patent acquisitions (21015 through 21035) are context, not infrastructure.

On capacity, the clearest trajectory is at Hardin. Claims 14789 and 14790 set a 100 MW total with a 30 MW first phase. By May 2021, claims 14645 and 14646 restated Phase I as 70 MW and Phase II as 30 MW. The total held while the phasing more than doubled the near-term tranche, and nothing on record explains the change. Hardin is now marked cancelled, so the growing promise ended in withdrawal rather than delivery. Fleet hashrate targets moved more modestly, from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858) as S-19 Pro orders grew.

On dates, the narrow equipment deliveries held. Claims 14872, 14873, 14854 and 14762 are verified. Claim 14855, a Q4 2020 delivery of 1,000 S-19 Pros, carries no resolution. The Quebec lease is internally inconsistent. Claim 15229 is marked abandoned, while 15219 covers the same November 2022 term at Granby and is still flagged not yet verifiable nearly four years past its date. The SelectGreen purchase is likewise marked withdrawn in 15062, yet reappears as completed in 14851 and 14883. The record does not show whether that was a revision or a reversal.

On independent evidence, the record contains no permits, interconnection-queue entries or site reporting. Every claim is either company-stated or verified only against later company disclosures. None of the operational sites carries a capacity figure, so the announced pace cannot be checked against physical build-out.

Over the next 90 days, the priority is to backfill post-2021 claims, since the current portfolio is invisible here. Beyond that, watch Marathon's Q3 2026 results for megawatt and timeline figures on MAT 1177, Matagorda County, Coshocton and Plano. Check ERCOT queue filings for the Texas sites and Ohio Power Siting Board dockets for Long Ridge and Coshocton for outside confirmation. Granby also needs a status decision: it is still listed as announced even though its only dated claim has expired.

The record supplied stops at claim 14648 in May 2021, so this read covers Marathon's transition from patent holder to miner and the Hardin build-out. None of the currently active sites appears in it: Long Ridge, Matagorda, Coshocton, Plano and MAT 1177 have no capacity or date claims here. Several pre-2017 entries also look misattributed, including the 19- and 26-country footprint and the UN Women and EY partnerships (20896-20898), and should be checked before publication.

Within the visible window, announced scale mostly ratcheted upward. The fleet hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858), driven by successive S-19 Pro orders that culminated in the 10,500-unit purchase (14842). Hardin kept its 100 MW headline (14789), but its phasing flipped. It was announced as a 30 MW Phase I (14790), then restated seven months later as a 70 MW Phase I and a 30 MW Phase II (14645, 14646). Nothing in the record acknowledges or explains that change.

Short-dated hardware promises held. The mid-August 2020 miner deliveries (14872, 14873, 14854) and the 10,000-unit S-19 Pro tranche due by March 2021 (14762) are all verified. The Q4 2020 delivery of 1,000 S-19 Pros (14855) was never resolved either way. Site-level commitments fared worse. The Quebec lease through November 2022 is marked abandoned (15229), yet the same lease, tagged to Granby, still reads not yet verifiable (15219). Hardin is now cancelled, but no claim records its wind-down. The SelectGreen purchase also appears as withdrawn (15062) and later as completed (14851, 14883), a contradiction the company never reconciled publicly.

No independent evidence appears in this record: no permits, no interconnection queue positions, no on-the-ground reporting. Hardin's capacity rested entirely on the Beowulf power purchase agreement (14647, 14791). As a result, the announced pace can only be judged against Marathon's own later statements, not against third-party filings.

Over the next 90 days, the first priority is completing the claim history so the Texas, Ohio and Plano projects can be assessed at all. Beyond that, watch for ERCOT large-load interconnection entries for the Texas sites and PUCO or local zoning filings for Long Ridge and Coshocton. In the next quarterly report, check whether any megawatt or energization dates for those sites are restated without comment, as Hardin's phasing was.

The record supplied for Marathon Digital ends at claim 14647, in May 2021. Everything below therefore covers the Granby, Quebec and Hardin era. The later Texas, Ohio and Abu Dhabi sites cannot be assessed from this extract. Several pre-2018 entries also look misattributed: the 26-country footprint (20897), the African presence (20898) and the UN Women and EY partnerships (20896, 20895) do not fit the Marathon Patent Group predecessor. They should be checked for entity-resolution errors before publication.

Capacity has generally grown, though not in a straight line. The fleet hashrate target rose from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858), and then to a 10,000-plus S-19 order (14842, 14762). Hardin was announced as a 100 MW site with a 30 MW first phase (14789, 14790). By May 2021 it was described as a 70 MW Phase I plus a 30 MW Phase II (14645, 14646). The total stayed at 100 MW, but the phases were reshaped without explanation. Hardin is now marked cancelled, and no capacity figure is attached to it at closure.

Near-term hardware dates mostly held. The mid-August 2020 deliveries (14873, 14872, 14854) and the Q1 2021 S-19 delivery (14762) are verified. The fourth-quarter 2020 batch of 1,000 miners (14855) has no follow-up. The SelectGreen purchase was marked withdrawn (15062) but was later restated as completed (14851, 14883), and the record does not reconcile the two. The 2019 Quebec lease running to November 2022 appears twice: as abandoned under Quebec (15229) and as not yet verifiable under Granby (15219). Granby is also still listed as announced despite operating claims (15198, 14859). That is a tracking inconsistency, not a company slip, but it hides whether the lease actually ran its full term.

No independent evidence appears in this extract. There are no permits, utility filings, interconnection queue positions or site reporting, so the announced pace rests entirely on company statements and delivery confirmations.

Over the next 90 days, the priority is to load the post-2021 claims. Then look for capacity figures on the sites that have none: MAT 1177, Matagorda County, Long Ridge, Coshocton and Plano. Search for matching ERCOT and PJM interconnection queue entries, and check whether a capacity number for the Long Ridge site in Hannibal, Ohio, is tied to the adjacent power plant. Finally, correct Granby's status and confirm whether the Hardin miners were relocated or written off.

The record supplied here ends partway through May 2021 (it breaks off at claim 14648), so this read covers Marathon's early hosted-mining era. It cannot speak to the current pipeline of MAT 1177, Long Ridge, Matagorda County, Coshocton and Plano, none of which carries a capacity or cost figure on file. The 2013 and 2014 entries describe a patent-licensing business, and the 19- and 26-country footprint figures (20898, 20897) look misattributed to this issuer. They should be checked before anyone relies on them. On scale, the headline hashrate target moved quickly and upward. It was 240 PH/s across three filings (14906, 14887, 14871), then 294 PH/s within weeks (14858). The 10,000-unit S-19 Pro order (14762) followed, putting the fleet on a far larger trajectory. Hardin is the instructive case. The 100 MW total (14789) held steady, but Phase I was quietly re-cut from 30 MW (14790) to 70 MW (14645), with a 30 MW Phase II carved out (14646). The project now sits as cancelled, and nothing in the claims trail announces or explains that exit. On dates, the near-term hardware promises held: the mid-August 2020 delivery of 660 S-19 Pros (14873, 14854) and the March 2021 deployment of 10,000 units (14762) are both verified. The Q4 2020 delivery of 1,000 S-19 Pros (14855) was never closed out either way. The Quebec lease is internally contradictory. The same November 2022 term is tagged abandoned under 15229 but not yet verifiable under 15219, and the SelectGreen purchase is marked withdrawn (15062) yet restated as completed (14851, 14883). Those conflicts point to revision without clean acknowledgement and should be reconciled. Independent corroboration is absent from this record. There are no permits, no interconnection queue entries, and no site reporting for Hardin, Granby or any announced Texas or Ohio site, so the announced pace rests entirely on company statements and third-party hosts such as Compute North and Beowulf. Over the next 90 days, the priority is to pull the post-2021 claims history and test whether the current announced sites ever gained stated megawatts. Analysts should look for ERCOT large-load queue entries tied to Matagorda County and MAT 1177, PJM or local utility filings for Long Ridge and Coshocton, and whether third-quarter results attach energization dates to any of them.

The supplied record ends mid-2021, cutting off at claim 14650, so this read covers Marathon's early pivot from patent licensing to mining. It cannot assess the operational Texas, North Dakota, Nebraska and Abu Dhabi sites or the newer announced projects such as MAT 1177, Long Ridge, Matagorda County, Coshocton and Plano, none of which carries a capacity or cost figure in the project table. Claims from 2013 and 2014 (21042 through 21020, 20898, 20897) describe intellectual property and corporate presence, not facilities, and say nothing about delivery capacity.

On scale, the hashrate target moved upward quickly. A 240 PH/s goal (14906, 14887, 14871) became 294 PH/s within weeks (14858), driven by successive S-19 Pro orders that swelled to 10,000 units (14762) and then 10,500 (14842). The unit arithmetic does not reconcile cleanly. A 500-unit add-on (14870) sits beside a 1,660-unit total (14853) that only works as 660 plus 1,000 (14854, 14855), and the MicroBT model naming varies across 14868, 14872 and 14852. Hardin shows the clearest restatement. It was announced as 100 MW with a 30 MW Phase I (14789, 14790), then recast with a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The headline stayed at 100 MW, but the phasing was inverted without stated explanation, and the project now carries a cancelled status.

Dated promises are sparse, and the near-term ones held. The mid-August 2020 delivery of 660 S-19 Pros (14873, 14854) and the 700-unit MicroBT order (14872) are verified, as is the 10,000-unit purchase targeted for March 2021 (14762). The Quebec lease running to November 2022 is marked abandoned (15229), while the parallel Granby framing (15219) remains unverifiable. The SelectGreen equipment purchase was withdrawn (15062) and then reappears as stated (14851, 14883). That is a reversal the record does not reconcile.

No independent evidence appears in this extract: no permits, no interconnection queue positions, no site reporting. Hardin's power arrangement (14647, 14791) rests entirely on company disclosure.

Over the next 90 days, three things matter. First, whether the full post-2021 claim history surfaces capacity figures for the announced Texas and Ohio projects. Second, whether those figures can be matched to ERCOT and PJM queue entries. Third, whether Marathon explains Hardin's cancellation alongside the earlier phase restatement.

The record supplied for Marathon ends mid-entry at claim 14649 in May 2021. This read therefore covers the 2019 to 2021 build-out in depth. For the sites that now define the portfolio it can mostly only flag gaps. The earliest entries (20896, 20897, 20898, and the 2014 patent cluster around 21015 to 21035) come from the patent-monetization predecessor. Some, such as the UN Women and 19-country African presence claims, look misattributed to this issuer. They should not be read as infrastructure history. On capacity, the pattern is rapid upward restatement. The fleet target of 240 PH/s (14906, 14887, 14871) became 294 PH/s within three weeks (14858). It was then overtaken by the 10,500-unit S-19 order (14842) and the 10,000-unit purchase (14762). Hardin held its headline 100 MW (14789) but was rephased, from a 30 MW Phase I (14790) to 70 MW plus 30 MW (14645, 14646), with no stated reason. Hardin and the Quebec site are now both cancelled. The excerpt does not show either cancellation being announced. Dates held where they were narrow and equipment-based: the mid-August 2020 deliveries (14873, 14872, 14854) and the March 2021 S-19 tranche (14762) all verify. Revisions went unflagged elsewhere. The SelectGreen purchase is marked withdrawn (15062), yet it is restated as completed in 14883 and 14851. The Quebec lease through November 2022 shows as abandoned under Quebec (15229) and as unverifiable under Granby (15219). That suggests one obligation was tracked twice and diverged. The Q4 2020 delivery of 1,000 S-19s (14855) was never resolved. Independent evidence is absent. Nothing in the record ties any claim to a permit, an interconnection queue position, or site reporting. Every project, operational or announced, carries a null capacity and cost. That includes MAT 1177, Long Ridge, Matagorda County, Coshocton, and Plano. Over the next 90 days, watch the Q3 2026 results and 10-Q for the first MW or cost figures on the announced sites. Then check them against ERCOT large-load filings for the Texas projects and PJM queue entries for the two Ohio sites. Also check whether the operational Texas, Dakota, and Nebraska sites get stated capacities that can be reconciled. Until that happens, the portfolio's delivery pace rests on company assertion alone.

The record supplied here effectively ends in May 2021, mid-entry at claim 14649. Everything after that is missing, including any capacity or timeline claims for the projects that matter now: Long Ridge, Matagorda County, MAT 1177, Coshocton and Plano. Those projects carry no capacity or cost figures in the project list, so this read covers only Marathon's early mining build-out. The 2013-2014 entries (claims 20898, 20897, 20896, 20895 and the patent-portfolio cluster around 21015-21035) describe the predecessor patent-licensing business, or appear misattributed, and say nothing about facility delivery. In the window that is visible, capacity grew in steps. The fleet hashrate target held at 240 PH/s across three statements (14906, 14887, 14871) and then rose to 294 PH/s (14858) once more S-19 Pro orders were added. Hardin is the instructive case. It was first framed as a 30 MW Phase I within a 100 MW total (14790, 14789). Seven months later it was restated as a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The total stayed at 100 MW, but the front-loaded share more than doubled, and the record does not explain why. Hardin now sits as cancelled, so the larger near-term commitment did not survive. Date discipline on hardware was good. Mid-August 2020 deliveries (14873, 14872, 14854) and the 10,000-unit S-19 Pro order due by March 2021 (14762) are all verified. The Q4 2020 tranche of 1,000 miners (14855) never received a verification status. Facility leases fared worse. The Quebec lease running to November 2022 is marked abandoned (15229), while a parallel claim tagged Granby is still unverifiable (15219). That suggests the exit was never cleanly disclosed. The SelectGreen purchase is likewise marked withdrawn (15062) but restated as completed in later filings (14883, 14851). That inconsistency deserves reconciliation. No independent evidence appears anywhere in the record: no permits, interconnection-queue entries or site reporting. The announced pace therefore rests entirely on company statements. Over the next 90 days, the Q3 2026 10-Q should be checked for capacity and energization dates on the Texas and Ohio sites. ERCOT large-load interconnection filings should be checked for Matagorda County and MAT 1177, and Ohio regulatory or local zoning dockets for Long Ridge and Coshocton. Any Plano filing should be examined for whether it is a real facility or an office. Above all, the claim history after May 2021 needs to be backfilled before conclusions are drawn about current delivery credibility.

The record supplied here ends in May 2021 at claim 14645. That leaves everything after the Hardin buildout unexamined, including the Texas, Ohio and Plano sites now listed as announced. Read this as a verdict on MARA's early mining pivot, not its current pipeline. The 2013 and 2014 entries (20897, 20898, 20896, 20895) describe a 26-country footprint and UN Women and EY partnerships, which do not match the predecessor patent-licensing business. They look like entity-resolution errors and should be purged before publication. On capacity, the trajectory runs steadily upward. The fleet target was 240 PH/s through mid-2020 (14906, 14887, 14871). It rose to 294 PH/s within weeks (14858). It then jumped by roughly an order of magnitude with the 10,500 and 10,000 unit S-19 Pro orders (14842, 14762). Hardin was framed as 100 MW in total with a 30 MW first phase (14789, 14790), and a 30 MW second phase followed (14646). Hardin now carries a cancelled status, and nothing in the visible claims announces or explains that reversal. That is the clearest case of a revision made without acknowledgement. Dates held where they were testable. The mid-August 2020 delivery of 660 miners verified (14873, 14854), and so did the 10,000-unit tranche due by March 2021 (14762). The fourth-quarter 2020 delivery of 1,000 units (14855) was never checked against an outcome. The Quebec lease shows two problems. It is marked abandoned under the Quebec project (15229) but remains not yet verifiable under Granby (15219), and Granby is still listed as announced, so the two project rows probably describe one site. The SelectGreen purchase was withdrawn (15062) and then restated as complete (14851, 14883) without explanation. No independent evidence appears anywhere in this record: no permits, no interconnection queue positions, no local reporting. Every capacity claim rests on company disclosure alone, and no project carries a recorded megawatt or cost figure. Over the next 90 days, the third-quarter filing should state energized megawatts and say plainly whether the announced Texas and Ohio sites are under construction, paused or reclassified. Check ERCOT large-load interconnection filings for MAT 1177, Matagorda County and Hansford County. Check Ohio utility and siting dockets for Long Ridge and Coshocton. Finally, re-ingest the claims after 14645 so that any revision since 2021 can be tested against what was originally promised.

The record supplied here ends mid-entry in May 2021, so this read covers Marathon's early build-out rather than its current portfolio. That gap is the most important finding. Every project now listed as operational or announced, from Granbury and McCamey to Long Ridge, Coshocton, Matagorda County and MAT 1177, carries no capacity or cost figure, and none of them appears in the visible claims. The pre-2019 entries also need cleanup before publication. The patent acquisitions (21016, 21020, 21031) reflect Marathon's patent-licensing era, but the 26-country footprint and the UN Women and EY partnerships (20897, 20896, 20895) do not read like this company and look misattributed.

Where capacity was stated, it grew rather than shrank. The fleet hashrate target of 240 PH/s (14906, 14887, 14871) became 294 PH/s within weeks (14858) as S-19 Pro orders expanded. At Hardin, the original plan of a 30 MW first phase toward 100 MW (14790, 14789) was restructured into a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The total stayed the same, but most of it moved into the first phase. Hardin is now cancelled, so that 100 MW commitment did not survive in its announced form.

Near-term equipment dates held. The mid-August 2020 miner deliveries (14873, 14872, 14854) and the 10,000-unit S-19 Pro order due by March 2021 (14762) are all verified. Two items show quiet revision. First, the Q4 2020 delivery of 1,000 S-19 Pros (14855) has no recorded resolution. Second, the SelectGreen purchase was marked withdrawn (15062) and then restated as completed (14851, 14883) without any reconciliation.

The Quebec records conflict with each other. The same lease running to November 2022 is tagged abandoned under Quebec (15229) and not yet verifiable under Granby (15219). Meanwhile Granby is still listed as announced, even though the operating partnerships there (15198, 14859) suggest it was running in 2019 and 2020. These are probably one facility and should be merged or reconciled.

No independent evidence, such as permits, interconnection filings or local reporting, appears anywhere in the record. Nothing external corroborates the announced pace for any site.

Over the next 90 days, there are three things to watch. The first is whether Marathon attaches megawatt figures and in-service dates to Long Ridge, Coshocton, Matagorda County, MAT 1177 and Plano. The second is whether large-load interconnection entries with ERCOT for the Texas sites, and utility or siting filings in Ohio, match those figures. The third is whether the full post-2021 claim set, once loaded, shows the same pattern seen here: equipment dates that hold, and facility plans that get reshaped or dropped without acknowledgement.

The supplied record breaks off at claim 14647 (May 2021), so this read covers Marathon's early build-out in depth and its later projects only through their status flags. The oldest entries (20897, 20898, 20895, 20896) describe a 26-country footprint and UN Women and EY partnerships. Those fit neither Marathon's patent-licensing past (21016 through 21035) nor its mining business, and look misattributed. They should be excluded from any capacity baseline.

On capacity, the mining story starts with rapid upward revision. The 240 PH/s fleet target was repeated across three filings (14906, 14887, 14871), then rose to 294 PH/s within weeks (14858). It was soon overtaken by orders for 10,500 and 10,000 S-19 Pro units (14842, 14762). Hardin shows the same pattern at the site level. The 100 MW total held (14789, 14645 plus 14646), but Phase I was quietly recut from 30 MW (14790) to 70 MW (14645) without any explanation in the record. Hardin is now marked cancelled, so the largest capacity commitment on file never became a lasting operating site under this record.

The timeline evidence is better than the capacity evidence. The dated hardware deliveries were verified (14872, 14873, 14854, 14762). The facility-level dates are weaker. The Quebec lease running to November 2022 is flagged abandoned (15229), and the matching Granby lease claim (15219) is still unverifiable. Even so, Granby carries an announced status while Quebec is cancelled, and that inconsistency should be resolved. The SelectGreen purchase was withdrawn (15062) and later restated as completed (14851, 14883) with no acknowledgement of the reversal.

The record contains no independent corroboration of the claimed pace: no permits, no ERCOT or PJM interconnection queue entries, and no on-the-ground reporting. Every one of the sixteen listed projects has null capacity and null cost. The operational sites, including Granbury, McCamey, Garden City, Ellendale, Jamestown, Kearney, Hansford County and Abu Dhabi, have their status asserted but no supporting claim history in the material provided.

Over the next 90 days, the third-quarter 10-Q due in early November is the main test. It should show energized megawatts by site and whether MAT 1177, Matagorda County, Plano, Long Ridge and Coshocton have moved beyond announced status. In parallel, check the ERCOT large-load interconnection queue for the Texas sites and Ohio PUC and PJM filings for Long Ridge and Coshocton. Also watch whether Granby is formally retired. Any site that stays announced with no queue position by year-end should be treated as aspirational.

The ledger supplied here ends mid-record at May 2021, so this read covers Marathon's early buildout and cannot assess the current Texas, Ohio or Plano pipeline. The 2013-2014 entries (20897, 20898, 20896, 20895) describe a multinational entrepreneurship footprint that does not fit Marathon Patent Group's history. They look like misattributed records and should be set aside. On capacity, the pattern is quiet re-scoping rather than outright shrinkage. Fleet hashrate guidance held at 240 PH/s across three filings (14906, 14887, 14871). It then rose to 294 PH/s a week later (14858) once the Compute North hosting arrangement and additional S-19 orders entered the picture, with no reconciliation of the two figures. Hardin is the clearer case. It was announced at 30 MW for Phase I within a 100 MW target (14790, 14789), then restated seven months later as a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The total stayed the same, but the phasing was inverted without explanation, and the project now sits as cancelled. Timing claims tied to hardware deliveries largely held: the mid-August 2020 S-19 Pro and MicroBT deliveries (14872, 14873, 14854) and the 10,000-unit order (14762) are all verified. The Quebec lease tells a different story. One record marks it abandoned (15229), while a parallel record, tagged to Granby, remains unverifiable (15219). The site appears in the project list as cancelled with no claim documenting the exit. The SelectGreen purchase is similarly muddled: it is marked withdrawn (15062), yet later filings describe the same 6,000-unit deal as completed (14883, 14851). Independent corroboration is absent throughout. No permit, interconnection-queue or on-the-ground evidence is attached to any site, and every project, including the seven listed as operational, carries a null capacity figure. That leaves Marathon's stated pace effectively self-reported in this dataset. Over the next 90 days, the priority is to backfill post-2021 claims so the announced sites (Long Ridge, Matagorda County, Coshocton, Plano, MAT 1177) can be tracked against firm megawatt and date commitments. Specific items to watch are capacity disclosures in the Q3 2026 10-Q, ERCOT large-load interconnection filings for the Texas sites, Ohio siting or utility filings for Coshocton and Long Ridge, and whether any announced site is reclassified or dropped without a corresponding claim, as Hardin and Quebec were.

The record supplied for Marathon is truncated at May 2021, so this read covers the Hardin, Quebec/Granby and early fleet-buildout era. It cannot speak to the sites the project list shows as operational or announced today, including Granbury, Ellendale, Abu Dhabi, Plano and Coshocton, none of which carry capacity or cost figures here. The 2013-2014 entries on country footprints and the UN Women and EY partnerships (20897, 20898, 20896, 20895) do not fit a mining or patent business. They look like misattributed records and should not be weighed. Capacity framing scaled up quickly and was reshaped quietly. The fleet hashrate target held at 240 PH/s across three statements (14906, 14887, 14871), then became 294 PH/s within days (14858) without a stated reconciliation. Hardin is the clearest case. The 100 MW total in 14789 survived, but the phase split moved from a 30 MW Phase I (14790) to a 70 MW Phase I plus 30 MW Phase II (14645, 14646). The record shows no acknowledgement of that change. Near-term equipment dates held up well. The mid-August 2020 deliveries (14873, 14854, 14872) and the 10,000-unit S-19 order due by March 2021 (14762) are verified. The Q4 2020 tranche of 1,000 S-19s (14855) has no recorded outcome. The SelectGreen purchase was marked withdrawn (15062) and then restated as completed (14851, 14883), a sequence worth flagging. Facility-level timelines are weaker. The Quebec lease running to November 2022 appears as both abandoned (15229) and not yet verifiable (15219). Granby is still listed as announced despite operating claims dating to 2019 (15198, 14859). Hardin is now cancelled with no dated cancellation claim in this excerpt. That makes the 100 MW narrative unresolved rather than visibly retired. No independent evidence appears anywhere in the record, including permits, utility filings, interconnection queue entries and site reporting. Every capacity figure here therefore rests on company statements alone. Over the next 90 days, the priority is ingesting post-2021 claims so the announced Texas and Ohio sites get dated capacity and timeline baselines. After that, check ERCOT large-load and PJM queue filings and county permitting for Matagorda, Hansford, Coshocton and Long Ridge against whatever megawatt figures Marathon attaches. Finally, look for a filing that explicitly reconciles the Hardin exit and the Quebec lease's end.

This record covers only part of Marathon's history. The supplied claims stop in May 2021 at claim 14650, so the current pipeline cannot be assessed from it: Long Ridge, MAT 1177, Matagorda County, Coshocton and Plano appear only as status labels with no capacity, cost or dated milestones. The earliest entries also need cleaning. Claims 20896 through 20898, covering UN Women and EY partnerships and operations across 26 countries, do not fit Marathon's patent-licensing predecessor and look misattributed to it. The patent-era acquisitions (21015 through 21035) are not facility claims at all.

In the mining era, announced capacity grew quickly. The fleet target went from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858), and then to a 10,500-unit S-19 Pro order (14842) that dwarfed both. Hardin shows a quieter kind of revision. The original 100 MW plan had a 30 MW Phase I (14789, 14790). By May 2021 it was described as a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The total was the same, but the phasing was reversed with no acknowledgement in the record, and the site is now listed as cancelled.

Near-term delivery promises held well. The mid-August 2020 shipments (14854, 14872, 14873) and the 10,000-unit March 2021 order (14762) are all verified. Facility commitments fared worse. The Quebec lease running to November 2022 is marked abandoned in 15229, yet the same lease is still not_yet_verifiable under Granby in 15219. That conflict needs to be reconciled before Granby's announced status means anything. The SelectGreen purchase was recorded as withdrawn (15062) and then restated as completed (14851, 14883).

The record contains no independent evidence: no permits, no interconnection queue positions, no site reporting. Nothing here supports or contradicts the announced pace of any active project.

Over the next 90 days, the first priority is to load the post-2021 claims so the open projects can be dated. After that, check ERCOT queue and county filings for the Texas sites and PJM queue and Ohio Power Siting Board filings for Long Ridge and Coshocton. Also watch the Q3 2026 filing for capacity or energization dates attached to any project still listed as announced.

Marathon's capacity record is really two companies stitched together, and the early layer should be read with caution. The 2013-2014 entries are patent-portfolio acquisitions and licensing deals from the Marathon Patent Group era. The multinational-footprint and partnership claims among them (20897, 20898, 20896, 20895) look like misattributed records that say nothing about mining infrastructure. The facility story starts in mid-2019 with the Quebec lease (15229, 15230, 15219). The record contradicts itself there: the same November 2022 lease end-date is tagged abandoned under the Quebec project and still not_yet_verifiable under Granby. Both labels cannot be right, and the abandonment was never accompanied by an explicit company statement.

Announced compute capacity drifted upward rather than shrinking. The fleet target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858) as S-19 Pro orders were layered on (14853, 14842). The near-term delivery promises in this window are the strongest part of the record: the August 2020 S-19 Pro and M31S+ shipments (14872, 14873, 14854) and the 10,000-unit order due by March 2021 (14762) are all marked verified. The SelectGreen S-9 purchase is murkier. It was withdrawn (15062), then restated as completed months later (14883, 14851) without the reversal being explained.

Hardin is where the trajectory breaks. In October 2020 it was framed as a 30 MW first phase toward 100 MW (14790, 14789). By May 2021 the same 100 MW was re-split into a 70 MW Phase I and a 30 MW Phase II with fresh cost figures (14645, 14646). That is a silent re-baselining, not a scale-up. The project now sits as cancelled, and the record contains no dated claim marking when or why.

Independent corroboration is thin across the board. The record holds no permit, interconnection-queue or site-level reporting for any project. The verified items are equipment deliveries, not energized megawatts.

The largest gap is forward-looking. MAT 1177, Long Ridge, Matagorda County, Coshocton, Plano and Granby are all listed as announced, with no capacity, cost or date attached in the record as supplied, which also appears truncated after claim 14650. Over the next 90 days, watch for three things. First, any MW or energization dates Marathon attaches to these sites. Second, matching ERCOT interconnection entries for the Texas projects and PJM queue activity for Long Ridge and Coshocton. Third, whether Granby's status is reconciled with the Quebec cancellation. Until filings corroborate a date, treat each of these as a declared intention rather than a delivery schedule.

Marathon's record is almost entirely about machines and megawatts announced in 2019 to 2021, and it says little about the sites now listed as operational or announced. The 2013 to 2014 entries (20897, 20898, 20896, 20895) describe a multinational with UN Women and EY tie-ups, which does not fit Marathon's patent-holding predecessor. They look like an entity-resolution error and should be excluded from any trajectory read. The input also ends at claim 14650 in May 2021, so this read covers only the early mining build-out.

The early capacity statements scaled up quickly. The fleet target was 240 PH/s through mid-2020 (14906, 14887, 14871), rose to 294 PH/s within weeks (14858), and then the company committed to 10,000-plus S-19 Pro units (14842, 14762). Hardin moved the other way in composition but not in size. In October 2020 it was a 100 MW target with a 30 MW Phase I (14789, 14790). By May 2021 Phase I was 70 MW and Phase II was 30 MW (14645, 14646). The total held, but the phasing was reversed with no stated reason. Hardin is now marked cancelled, and nothing in the supplied record explains why.

Near-term delivery dates mostly held. The August 2020 miner deliveries (14872, 14873, 14854) and the 10,000-unit order (14762) are verified. The Q4 2020 delivery of 1,000 units (14855) was never confirmed either way. The SelectGreen purchase was marked withdrawn in November 2019 (15062) and then restated as completed in 2020 (14883, 14851) without the reversal being explained. The Quebec lease shows up under two projects with conflicting statuses: abandoned under Quebec (15229) and not yet verifiable under Granby (15219), even though its November 2022 end date passed long ago.

The record contains no independent evidence, such as permits, interconnection queue positions, or local reporting, for any site. That matters most for MAT 1177, Matagorda County, Plano, Long Ridge and Coshocton, which are announced with no stated capacity or cost.

Over the next 90 days, watch whether the Q3 2026 filing attaches megawatt figures and energization dates to the announced Texas and Ohio sites. Check whether those sites appear in ERCOT large-load interconnection filings or Ohio siting and utility dockets. Look for any company statement that settles whether Granby is still active. Until claims after 2021 are loaded and checked against third-party records, the current pipeline should be treated as unverified.

The record supplied for Marathon is heavily weighted toward its pre-mining history and stops abruptly in May 2021 (claim 14650 is truncated mid-entry), so this read covers the 2019 to 2021 build-out and flags gaps rather than speculating about the current portfolio. The 2013 to 2014 patent acquisitions and partnerships (21015 through 21042, 20895, 20896) reflect the Marathon Patent Group era and say nothing about facility delivery. Claims 20897 and 20898, which describe operations in 26 countries and an African footprint, do not match this company and look like an entity-resolution error worth correcting upstream.

On capacity, the hashrate target grew modestly and transparently, from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858) within about six weeks in mid-2020, driven by incremental S-19 Pro orders. Hardin is the more telling case. It was announced at 100 MW total with a 30 MW Phase I (14789, 14790). By May 2021 that had become a 70 MW Phase I and a 30 MW Phase II (14645, 14646). The total held, but the phasing was inverted with no stated reason, and the project now carries a cancelled status that no claim in this record explains.

Dated promises mostly held. The mid-August 2020 miner deliveries (14854, 14872, 14873) and the 10,000-unit S-19 Pro order due by March 2021 (14762) are verified. The weak spot is the Quebec lease running to November 2022. It is marked abandoned under 15229 but still not_yet_verifiable under 15219, even though that date passed nearly four years ago. The SelectGreen S-9 purchase was withdrawn (15062) and then restated as completed in 2020 (14851, 14883) without any acknowledgement in the record of the reversal.

No independent evidence is attached to any claim. The record contains no permits, interconnection-queue entries, or site reporting, so the announced pace rests entirely on company statements and the delivery verifications noted above.

The Texas, Ohio, and North Dakota sites listed as operational or announced (MAT 1177, Matagorda County, Long Ridge, Coshocton, Plano) have no capacity, cost, or timeline claims in this extract. That is the most important gap. Over the next 90 days, three things matter: the Q3 2026 10-Q, expected in early November, for MW and timeline figures on those sites; ERCOT large-load interconnection filings covering the Matagorda and MAT 1177 sites; and Ohio siting or utility filings for Long Ridge and Coshocton. Closing out the stale Granby lease claim and explaining the Hardin cancellation would also clean up the historical baseline against which those newer promises should be judged.

The record supplied is truncated at claim 14647 and stops in May 2021. It contains no capacity or timeline claims for any of the projects now marked announced, including MAT 1177, Long Ridge, Matagorda County, Coshocton and Plano. This read therefore covers Marathon's early buildout, not its current pipeline. Several entries also look misattributed. The UN Women and EY partnerships (20896, 20895) and the 19- and 26-country footprint claims (20898, 20897) do not match either the patent-licensing predecessor or the mining company. They should be excluded before publication.

On capacity, the fleet-level hashrate target held at 240 PH/s across three filings (14906, 14887, 14871). It then rose to 294 PH/s in August 2020 (14858), which is consistent with the added S-19 Pro orders rather than an unexplained jump. Hardin is less clean. The October 2020 framing was a 100 MW site with a 30 MW Phase I (14789, 14790). By May 2021 Phase I was described as 70 MW and Phase II as 30 MW (14645, 14646). The total held, but the phase sizes were effectively inverted with no stated rationale. Hardin is now cancelled, so the 100 MW figure never reached completion.

On dates, the near-term hardware commitments are the strongest part of the record. The mid-August 2020 delivery of 660 S-19 Pros (14873, 14854) and the 10,000-unit order due by March 2021 (14762) are both verified. The fourth-quarter 2020 tranche of 1,000 units (14855) has no verification status. The Quebec lease is internally contradictory. The same November 2022 term appears as abandoned (15229) and as not yet verifiable under Granby (15219), and the Quebec project itself is cancelled. The SelectGreen miner purchase was marked withdrawn (15062) but restated as completed in later filings (14883, 14851). This record does not resolve whether the withdrawal or the restatements are correct.

The record contains no independent evidence, such as permits, interconnection queue positions or site reporting, for any facility. The verified items are equipment deliveries, not energized megawatts.

Over the next 90 days, the first priority is whether Marathon attaches megawatt and cost figures to the announced projects, all of which currently show none. Beyond that, ERCOT large-load queue filings for the Texas sites, PJM and Ohio regulatory activity for Long Ridge and Coshocton, and whether Plano is framed as AI or HPC capacity distinct from mining would each give the first external test of the current pipeline.

The supplied record cuts off mid-entry at claim 14647 (May 2021), so this read covers only Marathon's early mining build-out. The operational Texas, North Dakota, Nebraska and Abu Dhabi sites, and the announced MAT 1177, Long Ridge, Matagorda, Coshocton and Plano projects, have no capacity, cost or timeline claims in the visible data. Treat any judgment on the current portfolio as unsupported until those claims are loaded. The 2013-2014 entries (claims 21015 through 21035, 20895 through 20898) belong to the patent-licensing predecessor business and say nothing about facility delivery.

Within the mining era, announced scale moved up quickly and was sometimes reframed rather than reconciled. The fleet hashrate target held at 240 PH/s across three statements (14906, 14887, 14871), then became 294 PH/s eight days later (14858), with no bridging explanation in the record. Hardin was first presented as a 30 MW Phase I within a 100 MW total (14790, 14789). Seven months later it was described as a 70 MW Phase I and a 30 MW Phase II (14645, 14646), which reverses the phase sizing without acknowledgement. The project is now marked cancelled, and no closing claim explains the outcome.

Where dates were attached, Marathon largely delivered. The mid-August 2020 miner shipments (14872, 14873, 14854) and the 10,000 S-19 Pro units due by March 2021 (14762) are all verified. The fourth-quarter 2020 tranche of 1,000 S-19 Pros (14855) carries no hard date and remains only stated. The SelectGreen equipment purchase was marked withdrawn (15062), yet it reappears as a completed acquisition in later filings (14883, 14851), which needs reconciliation. The Quebec lease running to November 2022 appears twice: as abandoned under Quebec (15229) and as not yet verifiable under Granby (15219). That second status is stale now that the date has passed, and the duplicate inflates the open-promise count.

The record contains no independent corroboration: no permits, no interconnection queue entries, no site reporting. The announced pace therefore rests entirely on company statements and on the verified equipment deliveries.

Over the next 90 days, three things matter. First, ingest Marathon's post-2021 claims, especially the capacity and energization dates for the announced Texas and Ohio sites. Second, pull ERCOT and PJM queue positions for Matagorda, MAT 1177 and Long Ridge. Third, resolve the duplicated Quebec/Granby lease and the status conflict on the SelectGreen purchase before scoring Marathon's reliability.

The record supplied here ends partway through May 2021 (claim 14650 is cut off), so it says nothing about the Texas, Ohio and North Dakota sites that now make up most of Marathon's footprint. What it does show is an early pattern that is worth knowing before reading anything newer. The 2013-2014 entries describe the patent-licensing business that preceded the miner, and several of them (20895 through 20898, covering UN Women, EY and a 19- or 26-country presence) do not appear to belong to this company at all. They should be treated as entity-resolution errors, not as capacity history.

On scale, the mining-era targets ratcheted up quickly. The first stated fleet was 6,000 used S-9 units (15150, 15136). By mid-2020 the aggregate hashrate goal was 240 PH/s (14906, 14887, 14871). Within weeks that rose to 294 PH/s (14858), and orders followed for 10,000 to 10,500 S-19 Pro machines (14762, 14842). Hardin was framed as a 100 MW site (14789) built in 30 MW phases (14790, 14646). It is now marked cancelled, as is the original Quebec lease site (15229, 15230). Neither cancellation is reflected anywhere in the claim stream.

On dates, the record is weak in both directions. Delivery promises such as the mid-August 2020 S-19 shipment (14873, 14854) and the March 2021 completion of the 10,000-unit order (14762) remain unverified. The SelectGreen S-9 purchase was marked withdrawn (15062) and then restated as completed in later filings (14883, 14851) with no explanation of the reversal. Claim 14734 carries a promised-by date that precedes its own filing date, which looks like an extraction artifact rather than a slip. None of these claims is tied to a permit, an interconnection-queue position or on-the-ground reporting. The announced pace therefore rests entirely on company statements and supplier contracts.

In the next 90 days, the useful checks are mostly outside the company's own disclosures. Long Ridge, Coshocton, Matagorda County, MAT 1177 and Plano are all carried as announced with no stated capacity. ERCOT large-load and generation-interconnection entries would test the Texas sites, and PJM queue and Ohio Power Siting Board filings would test the Ohio pair. The Q3 2026 10-Q and earnings materials, expected in early November, are where announced megawatts should either gain dates and dollar figures or quietly stop appearing. The record should also be backfilled past May 2021 before any of the operational-site claims are scored.

Marathon's visible record covers only two of its fifteen listed sites, Quebec/Granby and Hardin, and it breaks off mid-2021. None of the fifteen projects carries a capacity or cost figure. The Texas, North Dakota, Nebraska, Ohio and Abu Dhabi sites appear only as names, so this read covers the early mining build-out, not the current portfolio. The 2013–2014 entries on African operations, UN Women and EY (20898, 20897, 20896, 20895) look misattributed to the patent-licensing predecessor or to an unrelated entity, and should not be weighed. On scale, the hashrate target moved up rather than down. Statements through mid-2020 anchored on roughly 240 PH/s (14906, 14887, 14871). By August 2020 that had become 294 PH/s (14858), and within days a 10,500-unit S-19 Pro order (14842) implied a much larger ambition. Hardin was framed as 30 MW in Phase I scaling to 100 MW (14790, 14789), but nothing in the excerpt confirms either figure was energized. Dates have held poorly. The mid-August 2020 S-19 Pro and MicroBT deliveries (14873, 14854, 14872) slipped. The 700-unit MicroBT purchase was dated to June 2020 and is also flagged late (14852). The 10,000-unit order promised by March 2021 (14762) slipped as well. The Hardin services agreement was referenced with an October 2020 date but was restated and re-papered through April and May 2021 (14734, 14738, 14649). The SelectGreen S-9 purchase is the clearest unacknowledged revision. It was marked withdrawn in November 2019 (15062), then reappeared as completed in the August 2020 filings (14851, 14883), with no reconciliation visible in the record. The record contains no independent evidence: no permits, interconnection-queue entries or site reporting. The Hardin capacity figures rest entirely on company statements tied to the Beowulf power agreement (14791, 14801), so the pace cannot be independently confirmed from this file. Over the next 90 days, three things are worth watching. First, whether post-2021 claims are ingested at all, since the truncation hides the period that matters. Second, whether any of the thirteen undocumented sites gains a stated MW figure, especially the Texas and North Dakota locations, where ERCOT and MISO queue filings could be checked against announcements. Third, whether Marathon's own disclosures ever reconcile the SelectGreen withdrawal and the Hardin phase targets with what was actually energized.

The record supplied here is truncated after claim 14649 (May 2021), so this read covers Marathon's early transition from patent licensor to bitcoin miner. It cannot speak to the later sites on the project list: MAT 1177, Long Ridge, Matagorda County, Abu Dhabi, Ellendale, Jamestown, Kearney, Coshocton, Garden City, McCamey, Granbury and Plano. Every one of those carries status=announced with no capacity or cost attached. A ledger in which fifteen sites remain at announced is itself a data-quality finding. Several of those locations should by now have either an operating figure or an explicit retirement. The 2013-2014 entries (20897, 20898, 20895, 20896) describe a 26-country footprint and African expansion that do not match Marathon's patent-monetisation business of that era, and they look like an entity-resolution error worth auditing before publication. On capacity, the trajectory runs upward but restates itself quickly. The fleet hashrate target sat at 240 PH/s across three filings (14906, 14887, 14871), then became 294 PH/s within weeks (14858). The new figure was built from component estimates of 73 and 56 PH/s (14857, 14856). Hardin entered at 100 MW with a 30 MW first phase (14789, 14790), and no later claim in view updates either number. Dates are where the pattern is weakest. The SelectGreen purchase of 6,000 S-9 units was due by September 2019 (15071), was recorded as withdrawn in November 2019 (15062), and then reappeared as a completed purchase in July and August 2020 (14883, 14851). Nothing on the record reconciles that reversal. The mid-August 2020 delivery of 660 S-19 Pros (14873, 14854), the MicroBT order (14852) and the 10,000-unit S-19 Pro batch due by March 2021 (14762) are all flagged delayed. The Hardin services agreement was re-papered at least three times (14756, 14738, 14734) before Beowulf was compensated in equity (14649), which suggests the counterparty terms moved more than the capacity did. The Quebec lease entries (15219, 15229) are tagged delayed, but a lease end date is not a delivery promise, and that tag should probably be cleared. No independent evidence is present in this record: no permits, no interconnection queue entries, no site reporting. The announced pace is therefore company-sourced only. Over the next 90 days, the priority is to restore the claim history after May 2021, check the Texas sites against ERCOT large-load interconnection filings, and check Long Ridge and Coshocton against Ohio utility and PUCO dockets. The third-quarter results should also show whether any announced site finally reports energized megawatts.

The record supplied here ends in May 2021, so this read covers Marathon's early build-out only. None of the fifteen listed projects carries a capacity or cost figure, and most (Abu Dhabi, Ellendale, Jamestown, Kearney, Coshocton, Granbury, Garden City, McCamey, Plano, Matagorda, Long Ridge) have no supporting claim in the excerpt. The 2013-2014 entries (20897, 20898, 20896, 20895) describe a 26-country business with an African footprint and UN Women and EY partnerships. Those look like an entity-resolution error rather than Marathon history and should be excluded before publication. The patent acquisitions (21015-21036) belong to the Marathon Patent Group predecessor and are not facility evidence.

On scale, the trajectory runs steadily upward. It starts at 1,400 deployed units in Quebec (15220), moves to a 6,000-unit S-9 purchase (15150, 15136), then to a 240 PH/s target (14906, 14887, 14871) quickly raised to 294 PH/s (14858). Orders followed for 10,500 and 10,000 S-19 Pros (14842, 14762), and Hardin was framed as a 30 MW first phase of 100 MW (14790, 14789). Each restatement enlarged the ambition. None shrank it explicitly.

Dates held poorly. The SelectGreen S-9 deal was due by September 2019 (15071), marked withdrawn (15062), then reappeared in 2020 filings as completed (14883, 14851) without a clear account of the reversal. S-19 Pro and MicroBT deliveries promised for mid-August 2020 slipped (14873, 14854, 14872), and the MicroBT date itself shifted between 14852 and 14872. The 10,000-unit batch due by March 2021 is flagged delayed (14762). The Hardin services agreement was re-papered three times (14756, 14738, 14734) before being restructured with $11.22 million in equity compensation to Beowulf and 2P1 (14649). That pattern points to renegotiation rather than smooth execution. The Quebec lease appears under both Granby and Quebec tags (15219, 15229), likely a duplicate.

The record contains no permits, interconnection queue entries, or on-site reporting for any project. The announced pace is therefore supported only by company statements.

Over the next 90 days, the Q3 2026 10-Q due in early November is the main checkpoint. It should be reconciled against the project list to confirm which sites are owned, hosted, or abandoned. ERCOT, MISO, SPP and PJM queue filings tied to the Texas, North Dakota, Nebraska and Ohio sites would provide the first independent signal. Any capacity figure finally attached to Abu Dhabi or Granbury should be compared with its earlier unquantified announcement.

Marathon's capacity story in this record is mainly about miners and hashrate, not about building sites. The 2019 plan centred on 6,000 used S-9 units for the Quebec lease (15150, 15141). By mid-2020 the target was 240 PH/s (14906, 14887, 14871). Within weeks it rose to 294 PH/s (14858), and days later came a 10,500-unit S-19 Pro order (14842). The first claim tied to a power-defined site is Hardin: 100 MW total with a 30 MW Phase I (14789, 14790). That marks the shift from leased space to dedicated capacity. Ambition clearly grew, but nothing here records commissioned megawatts against those figures. The 2013-2014 patent-era entries are a different business. Claims 20895-20898, covering 26 countries, African operations and UN Women and EY partnerships, do not fit any Marathon business line and look misattributed. They should be checked before publication.

Dates have not held well. Every equipment delivery date has slipped: the mid-August 2020 S-19 arrival (14873, 14854), the MicroBT batch (14872, 14852) and the 10,000-unit tranche promised by March 2021 (14762). The Hardin services agreement, dated October 2020, was signed and amended repeatedly through May 2021 (14734, 14738, 14649). Those amendments read as renegotiation, not progress. The SelectGreen purchase is marked withdrawn in 15062 but is restated as completed in 14851 and 14883, and the record never explains the difference. That is the clearest case of a revision made without acknowledgement.

The record contains no independent evidence: no permits, no ERCOT, PJM or MISO queue positions, and no on-site reporting. The project table lists all fifteen sites as announced with no capacity or cost. That includes Hardin and Granby, which Marathon is known to have operated and later left. The claims history is also cut off after May 2021. As a result, the newer sites (Ellendale, Jamestown, Kearney, Granbury, McCamey, Garden City, Coshocton, Long Ridge, Abu Dhabi, Plano) have no claims we can test.

Over the next 90 days, the Q3 2026 10-Q in early November is the main checkpoint for energized versus contracted megawatts at each site. For the Texas wind-adjacent and Granbury sites, ERCOT large-load interconnection filings would be the strongest independent signal. For Coshocton and Long Ridge, Ohio siting and PJM records serve the same role. The 15 project rows also need status corrections, and the claims after 2021 need to be backfilled before this trajectory can be fairly judged.

The record supplied here ends partway through 2021, so this read covers Marathon's early transition from patent holder to miner rather than its current footprint. That gap matters. All fifteen listed projects carry status=announced with no capacity or cost recorded, and the excerpt contains no claims at all for Garden City, McCamey, Ellendale, Jamestown, Kearney, Granbury, Coshocton, Long Ridge, Abu Dhabi or Plano. The 2013-2014 entries on operations in 26 countries, African expansion and the UN Women and EY partnerships (20897, 20898, 20896, 20895) do not fit Marathon's patent business and look like an entity-matching error worth auditing before publication.

Within the visible window, announced capacity grew quickly. The fleet hashrate target rose from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858), and a 10,000-plus S-19 Pro order followed (14842, 14762). Hardin held a 100 MW total (14789), but its phasing was relabeled. Phase I was 30 MW in October 2020 (14790) and became 70 MW, with a 30 MW Phase II, by May 2021 (14645, 14646). The record shows no explanation for the change.

Dates slipped often, and the slips were seldom acknowledged. The 700 MicroBT units were promised by mid-2020 (14852) and pushed to mid-August (14872). The 660 S-19 Pro delivery is flagged delayed (14854, 14873). The 10,000-unit order targeted March 2021 and is also flagged delayed (14762). The Hardin services agreement was dated to October 2020 but surfaced in April 2021 (14734). The SelectGreen purchase was withdrawn (15062) and then restated as completed (14851, 14883). The MicroBT model is variously given as M31S+, M30S+ and M305+ (14872, 14868, 14856), which reads as transcription drift rather than three separate orders.

The record holds no independent corroboration. There are no permits, no interconnection-queue entries and no site reporting for Hardin, Granby or any later site. Every capacity figure here is self-reported.

Over the next 90 days, the first priority is to load the full claim history so the later sites carry their own capacity and date claims. After that, check ERCOT queue and county filings for the Texas sites, North Dakota and Nebraska utility filings for Ellendale, Jamestown and Kearney, and Ohio siting records for Coshocton and Long Ridge. Any site that stays capacity-less with no permit trail should be read as aspirational, not delivered.

Marathon's facility record is better read as a series of pivots than as one build-out. The pre-2019 entries are patent acquisitions and licensing deals from the company's intellectual-property era (claims 21015 through 21036). The country-footprint and development-partnership items (20895 through 20898) do not fit Marathon's history and look misattributed, so they should not count toward any infrastructure trajectory. The mining story begins with the Quebec lease and the 26,700-square-foot Granby site (15230, 15219). It then moves through a short-lived hosted model with Compute North (14861) and settles on Hardin, Montana: a 100 MW target with a 30 MW first phase (14789, 14790) and a Beowulf power agreement later priced at 2.8 cents per kWh (14647). Announced scale has grown in step with purchase orders rather than physical sites. The fleet target rose from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858). It then jumped to orders of 10,000 to 10,500 S-19 Pro units (14762, 14842), which implies a far larger hashrate that was never restated as a consolidated figure in this record. Dates have been soft throughout. The SelectGreen purchase of 6,000 S-9 units missed its September 2019 close (15071, 14851). It was logged as withdrawn (15062), then reappeared as completed in July 2020 (14883) without a reconciling statement. The mid-August 2020 S-19 delivery (14873, 14854), the MicroBT order (14852, 14872), and the 10,000-unit order due by March 2021 (14762) all carry delayed flags. The Hardin services agreement, keyed to October 2020, was re-papered in April and May 2021 (14734, 14738, 14649), which suggests the operating structure was renegotiated rather than simply executed. The record contains no independent evidence for any of these milestones: no permits, no interconnection-queue entries, and no site reporting for Hardin or the fifteen listed projects. Every project, including Long Ridge, Matagorda, Ellendale, Abu Dhabi and Plano, carries no capacity or cost figure, so the current portfolio cannot be tested against stated pace. Over the next 90 days, watch for three things. First, any capacity or energization date attached to a listed project, especially the Texas and North Dakota sites. Second, ERCOT and MISO queue filings or county permits that corroborate them. Third, whether Marathon restates aggregate hashrate in a way that reconciles with its unit orders rather than replacing the old target silently.

The excerpt we received ends mid-record at a claim fragment, so this read covers only the visible claims, which run from 2013 to May 2021. It does not reach the newer project list (Ellendale, Kearney, Granbury, Garden City, Abu Dhabi, Coshocton and others), none of which has capacity or cost attached. The earliest entries also need entity review before publication. The 2014 patent acquisitions (21015 through 21036) belong to the Marathon Patent Group predecessor and have no bearing on facilities. The 2013 to 2014 claims about African operations, UN Women and EY (20895 through 20898) do not resemble anything in Marathon's history and look misattributed. Treat both groups as noise in a facility read.

On scale, the visible trajectory is upward and fast. Quebec started as a 26,700 square foot leased site (15230) with 1,400 deployed units (15220). By mid-2020 the hashrate target was 240 PH/s (14906, 14887, 14871), and weeks later it was 294 PH/s (14858). The 10,000 and 10,500 S-19 Pro orders (14762, 14842) followed, along with a Hardin plan of 30 MW in Phase I toward 100 MW total (14790, 14789). The targets were restated upward several times within a single quarter, and the record shows no capacity being retired.

Timelines held less well. The mid-August 2020 S-19 delivery (14854, 14873) and the MicroBT purchase (14852, 14872) are both flagged delayed, as is the 10,000-unit order due by March 2021 (14762). The SelectGreen purchase was withdrawn (15062) and then reappeared as completed against an older 2019 date (14851, 14883), with nothing in the record reconciling the two. The Hardin services agreement was restated with a backdated October 2020 obligation (14734). The Granby lease running through November 2022 (15219, 15229) is flagged delayed without any visible renewal or exit claim.

Independent corroboration is absent from what we were given. No permits, interconnection queue positions or site reporting are attached to any project, so every capacity figure here rests on company statements alone. That matters most for Hardin, where capacity depends on a third-party power plant and a Beowulf power purchase agreement (14791, 14801) rather than on assets Marathon owns.

Over the next 90 days, the priority is ingesting the claims cut off after 14650, since the current portfolio cannot be assessed without them. After that, look for ERCOT and SPP queue entries or county filings for the Texas, North Dakota and Nebraska sites. Also watch whether filings acknowledge the Quebec lease outcome or leave it unaddressed.

This read covers only part of Marathon's record. The claim list ends mid-entry in May 2021, so it says nothing evidence-based about the sites that define the company today: Granbury, McCamey, Garden City, Ellendale, Jamestown, Kearney, Coshocton, Abu Dhabi and MAT 1177. None of the fifteen project records carries a capacity or cost figure either, so the project-level view is empty. Several early entries also look misattributed. The 2013 claims of operations across 26 countries and an African presence in 19 (20897, 20898) and the UN Women and EY partnerships (20896, 20895) do not fit the patent-licensing shell Marathon was before it turned to bitcoin. They should be checked before they colour any reading of the company's history.

Within the visible window, announced scale grew in steps rather than shrinking. The 2019 base was 1,400 units in a 26,700 sq ft Quebec space (15220, 15230). The company then announced 6,000 S-9s (15150), a 240 PH/s target (14906, 14887, 14871) and a 294 PH/s target weeks later (14858). In October 2020 that became a 10,000-unit S-19 order (14762) and a Hardin, Montana plan of 30 MW for phase one within 100 MW total (14790, 14789). Each step arrived as a new target, with no statement of how much of the prior one had been delivered.

Timing slipped. The mid-August 2020 S-19 Pro delivery (14873, 14854), the 700 MicroBT units (14872, 14852) and the 10,000-unit order due by March 2021 (14762) are all marked delayed. The SelectGreen purchase is the clearest unacknowledged revision. It was recorded as withdrawn in November 2019 (15062), then restated as completed in August 2020 against its original September 2019 date (14851), with no visible explanation in between. The Hardin services agreement was also re-papered three times (14756, 14738, 14734), the last carrying a promise date that had already passed.

The record contains no independent evidence, such as permits, interconnection queue filings or site reporting, so none of these targets can be checked against the outside world. Over the next 90 days, three gaps matter most. Capacity and cost should be filled in on the fifteen project records. The full post-2021 claim history should be loaded, so the Texas and North Dakota build-out and the Abu Dhabi venture can be judged against their own promises. And interconnection and permit records for Ellendale, Coshocton and the Texas sites should be matched to claims, so stated megawatts can be checked against what the grid has actually approved.

The record supplied here ends partway through the 2021 claims, so this read covers Marathon's early mining build-out and cannot assess the newer Texas, North Dakota, Nebraska, Ohio and Abu Dhabi sites. Those sites appear on the project list with no capacity or cost attached. The 2013 and 2014 entries describe patent-licensing activity (21015 through 21036), and several, such as the 19-country African presence (20898), the UN Women partnership (20896) and the EY partnership (20895), look misattributed. None of them bears on facility delivery. On capacity, the trajectory has mostly been upward revision. The fleet hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858). At Hardin, the stated first phase grew from 30 MW of a 100 MW site (14790, 14789) to 70 MW (14645). Neither change came with an explanation of what drove it. Dates have held less well. The mid-August 2020 S-19 Pro delivery is flagged delayed in two filings (14873, 14854). So is the matching MicroBT order (14872, 14852). The 10,000-unit S-19 Pro order promised by March 2021 is also marked delayed (14762). The Hardin services agreement was dated to October 2020 but surfaces as delayed in April 2021 (14734). The SelectGreen purchase is the clearest case of silent revision. It was marked withdrawn (15062) and then reappeared as completed in later filings (14851, 14883) with no reconciling disclosure in the record. Equipment counts were repeated across filings almost verbatim (15150, 15141, 15101), which inflates the apparent volume of activity. Independent corroboration is absent from what was provided. No permits, interconnection queue positions, or site reporting are attached to Hardin, Granby or Quebec, so the announced pace rests entirely on company statements. In the next 90 days, the priority is to complete the claim history past 2021 and test the listed sites against outside records. Those records are ERCOT large-load interconnection filings for Granbury, McCamey, Garden City, Matagorda and Plano, SPP and MISO queues for Ellendale, Jamestown and Kearney, and PJM records for Coshocton and Long Ridge. The November quarterly filing should be checked for megawatt and cost figures on any of the fifteen sites still carried at announced status, since capacity that remains unquantified this long deserves skepticism.

The supplied claims log is cut off mid-record after claim 14650 (2021-05-09), so this read covers Marathon's trajectory only through spring 2021. It cannot speak to the later Texas, North Dakota, Nebraska, Ohio or Abu Dhabi sites listed on the project roster, none of which carries a capacity or cost figure. The 2013-2014 entries reflect the company's earlier patent-licensing identity, and claims 20896-20898 (UN Women, EY, African operations) look like a different 'Marathon' entity merged into this record. They should be excluded until verified.

Within the mining era, announced scale moved upward quickly and in steps. The fleet hashrate target sat at 240 PH/s across claims 14906, 14887 and 14871. It was restated as 294 PH/s within days (14858), then overtaken by orders of 10,500 and 10,000 S-19 units (14842, 14762). At Hardin, the framing went from a 30 MW first phase to a 100 MW total (14790, 14789). The Quebec footprint stayed modest at 26,700 square feet (15230).

Dates have held poorly, and some revisions passed without comment. Delivery of 660 S-19 Pro units by mid-August 2020 is flagged delayed in both 14854 and 14873, as is the 700-unit MicroBT order (14872, 14852). The SelectGreen purchase shows the clearest unacknowledged revision. It was marked withdrawn in November 2019 (15062), then reappeared as completed in 2020 filings with the original September 2019 target still attached (14851, 14883). The MicroBT model is variously recorded as M30S+, M31S+ and M305+ (14868, 14872, 14852), which weakens confidence in the unit-level figures. The Hardin services agreement slipped past its October 2020 target (14734), and its counterparty and compensation terms were restated several times (14756, 14738, 14649).

No independent corroboration appears in the record: no building permits, no interconnection queue positions, and no site reporting for Hardin, Granby or any later location. Every capacity figure here traces back to the company.

Over the next 90 days, the first priority is a complete export of post-2021 claims, so the newer sites can be judged on their own record. The second is whether the next quarterly filing attaches megawatt or energization figures to any of the fifteen 'announced' projects. The third is whether grid-operator queue data or local permitting for those sites, in Texas, North Dakota, Nebraska and Ohio, independently confirms the pace the company implies.

The record supplied for this read ends mid-entry in May 2021, so it covers only Marathon's early mining build-out, not the fifteen sites on file. Most of those sites, including Granbury, Garden City, McCamey, Ellendale, Kearney and Abu Dhabi, show no capacity or cost figures, and no dated claims visible here tie to them. The 2013–2014 entries describe patent-portfolio acquisitions (21015 through 21036), which fit the company's pre-mining history as a patent licensor. The multi-country and UN Women items (20896–20898) look like a different entity entirely and should be checked before anyone relies on them.

Within the visible window, announced capacity moved upward in small, quick steps. Management first cited a 240 PH/s deployment target (14906, 14887, 14871). Weeks later that became 294 PH/s (14858). A 10,500-unit S-19 Pro order (14842) was restated as 10,000 units two months afterward (14762), with no reconciliation between the two figures. At Hardin, the company described a 30 MW first phase inside a 100 MW envelope (14790, 14789). Neither figure is followed up with a dated energization claim in the record.

Delivery dates slipped repeatedly, and the company rarely said so directly. The mid-August 2020 arrivals of 660 S-19 Pros and 700 MicroBT units (14873, 14854, 14872) are marked delayed. So is the 10,000-unit batch due by March 2021 (14762). The SelectGreen S-9 purchase is the clearest example of drift. It was withdrawn (15062), then reappeared as completed (15071, 14851) against a September 2019 date it had already missed. The MicroBT model is also named three different ways (14868, 14872, 14856), which makes unit counts hard to reconcile. The Hardin services agreement (14734) points back to an October 2020 date that preceded its own execution.

No permits, interconnection-queue entries or site reporting appear in this record. That means none of the announced pace can be independently corroborated here.

Over the next 90 days, three things matter. First, whether the full claim history reloads and fills in capacity figures for the Texas, North Dakota, Ohio and Nebraska sites. Second, whether ERCOT and MISO queue positions line up with any megawatt figures that surface. Third, whether Hardin's 100 MW ambition is ever formally closed out or quietly dropped from later disclosures.

The claims record supplied here stops partway through 2021, so this read covers only Marathon's transition from patent licensing to Bitcoin mining. It says nothing about the fifteen named sites from Granbury and McCamey to Ellendale, Kearney, Coshocton and Abu Dhabi. Every one of those projects carries status=announced with no capacity or cost recorded, so the structured project layer offers no basis for judging current delivery. The 2013-2014 patent acquisitions (21015 through 21036) belong to the Marathon Patent Group era. The African-presence, UN Women and EY entries (20896-20898) look like entity-resolution errors and should be discarded.

Within the visible window, stated capacity grew quickly. The fleet hashrate target stood at 240 PH/s through mid-2020 (14906, 14887, 14871), rose to 294 PH/s within weeks (14858), and was then overtaken by orders for 10,500 and 10,000 S-19 Pro units (14842, 14762). Hardin followed the same pattern. It was announced as a 100 MW site with a 30 MW Phase I (14789, 14790), and by May 2021 Phase I was described as 70 MW at $23 million (14645). That more than doubles the first phase, and the filings in view do not reconcile the change with the original phasing.

Marathon's dates mostly slipped. The mid-August 2020 delivery of 660 S-19 Pros (14873, 14854), the MicroBT order (14852, 14872) and the 10,000-unit tranche due by March 2021 (14762) are all flagged delayed. The Hardin services agreement was promised for October 2020 but appears in April 2021 filings (14734). The SelectGreen purchase was marked withdrawn (15062), then restated as completed (14851, 14883), with no explanation in the record for the reversal.

The record contains no independent corroboration. There are no permits, no interconnection-queue entries and no site reporting, so the announced pace rests entirely on company statements. Hardin's Beowulf power arrangement (14647) is the most concrete input, but it is still self-reported.

Over the next 90 days, the priority is completing ingestion of claims after 2021 so that each named site has dated capacity statements. Once that is done, the third-quarter 2026 results should be checked for energized megawatts by site rather than fleet-level hashrate. The Texas sites should be matched against ERCOT large-load and generation queue filings, Ellendale and Jamestown against MISO or SPP records, and Coshocton and Hannibal against PJM. Any site still reported only as announced without a queue position or permit trail should be treated as aspirational until those records appear.

The record we hold for Marathon covers far less than the project list implies. Fifteen sites are listed as announced, and none carries a capacity or cost figure. The claim trail also stops in May 2021 (14647, 14648), so everything from Ellendale and Jamestown through Garden City, McCamey, Granbury, Long Ridge, Matagorda, Coshocton, Plano and Abu Dhabi has no dated capacity or timeline claim behind it. The earliest entries also need care. The patent acquisitions (21015 through 21036) fit Marathon's patent-licensing past. The claims about business in 19 and 26 countries and the UN Women and EY partnerships (20895 through 20898) do not fit this company and look misattributed. They should not inform any read on its facilities.

Within the window we can see, stated capacity went up quickly and delivery lagged. The fleet hashrate target was 240 PH/s in mid-2020 (14906, 14887, 14871). Weeks later it was 294 PH/s (14858), and then came orders for 10,500 and 10,000 S-19 units (14842, 14762). Hardin was framed as a 30 MW first phase toward 100 MW (14790, 14789), backed by an exclusive Beowulf power deal at 2.8 cents per kWh (14791, 14647).

Dates held poorly. The 6,000-unit SelectGreen purchase was promised for September 2019 and flagged delayed (15071, 14851). It was then marked withdrawn (15062) and restated as a completed purchase the next summer (14883), with no reconciliation on the record. Mid-August 2020 delivery of 660 S-19 Pro units slipped (14873, 14854). The 10,000-unit tranche due by March 2021 slipped (14762). The Hardin services agreement, dated to October 2020, surfaced in April 2021 (14734). The Quebec lease through November 2022 is also tagged delayed (15229, 15219).

There is no independent corroboration in this file. It contains no permits, no interconnection queue entries, and no site reporting. The Hardin megawatt figures and every later site rest only on company statements.

Over the next 90 days, the main task is to fill the gap from 2022 to 2026. Marathon's third-quarter 10-Q, due in early November, should give energized megawatts by site, and each figure should be checked against a stated target. For the Texas sites, look for ERCOT large-load interconnection filings. For Long Ridge and Coshocton, look for PJM or Ohio siting records. For Abu Dhabi, see whether any joint-venture capacity is disclosed separately from owned capacity. Until those entries exist, Marathon's announced pace should be read as a claim, not an observed trajectory.

The record traces Marathon's footprint from one leased building to a hosted fleet sized in tens of thousands of machines. Capacity has scaled up consistently, but the milestones attached to that growth have almost all slipped. The Quebec lease described a 26,700 square foot site (15230) holding roughly 1,400 units (15220). The first large jump was the 6,000-unit S-9 purchase (15150, 15136). The 2020 re-equipment cycle set a 240 PH/s target (14906, 14887, 14871), then raised it to 294 PH/s within weeks (14858). The Hardin announcement introduced megawatts as the unit: 100 MW total with a 30 MW Phase I (14789, 14790). Phase I was later restated at 70 MW with a 30 MW Phase II (14645, 14637, 14646), and the record contains no explanation for that change. Hosting ambitions peaked at 73,000 miners with Compute North (14591). The only deployment figure in the record is 30,000 machines (16620), which matches the joint-venture tranche (14348) rather than the headline 73,000. Dated promises have fared poorly. The SelectGreen purchase missed its September 2019 date (15071), was marked withdrawn (15062), and then reappeared as completed (14883, 14851) without any stated reconciliation. S-19 Pro deliveries promised for mid-August 2020 slipped (14873, 14854), as did the 10,000-unit order due by March 2021 (14762) and the Hardin services agreement (14734). The Compute North hosting date moved from March 2022 (14591) to June 2022 (14346), and that revision was not flagged as a slip. The 30,000-unit ASIC schedule tied to the same June 2022 date is also marked delayed (14523, 14344). The record contains no independent corroboration of any kind: no permits, no interconnection or large-load queue entries, and no on-site reporting. All fifteen projects, including Hardin, which the claims describe as operating, carry an announced status with blank capacity and cost fields. The pace cannot be verified from this dataset. The supplied claims also stop in March 2022, so the Texas, North Dakota, Ohio, Nebraska and Abu Dhabi sites have no dated promises to test. Over the next 90 days, three things matter most. First, fill in the project table's status and capacity fields from filings. Second, look for ERCOT large-load or interconnection entries for McCamey, Garden City and Granbury, and for utility or regulatory dockets covering Ellendale and Jamestown. Third, check whether current disclosures reconcile the unexplained Hardin Phase I resizing and the gap between 73,000 contracted and 30,000 deployed machines.

Marathon's facility record is a story of mining-fleet targets that grew steadily while delivery dates repeatedly slipped. This read covers the claims supplied, which break off in March 2022. Nothing after that point is assessed here. Announced capacity only moved up in this period. The fleet hashrate target was set at 240 PH/s across three consecutive filings (14906, 14887, 14871) and raised to 294 PH/s within weeks (14858). The company then moved to much larger hosting commitments: 73,000 miners with Compute North (14591), a 30,000-miner joint venture (14348), and an 1,800 percent hash rate increase framed as already under way (16621). Hardin is the one site where the structure changed without explanation. The 100 MW total held (14789, 14636), but Phase I went from 30 MW (14790) to 70 MW (14637, 14645), with a new 30 MW Phase II (14638, 14646). No filing in the record explains the change. Dates have not held well. The SelectGreen equipment purchase was due by September 2019 (15071), marked withdrawn (15062), and then reappeared as completed (14851, 14883). None of those filings acknowledges the earlier withdrawal. S19 Pro and MicroBT deliveries promised for mid-August 2020 slipped (14854, 14873, 14872). The Compute North hosting schedule moved from end of Q1 2022 (14592) to end of Q2 2022 (14346), and the Bitmain contract deliveries share that later deadline (14344, 14523). None of these revisions came with a public explanation. Independent corroboration is absent. The record contains no permits, interconnection queue entries, or site reporting for any of the fifteen named locations. Every project, from Granbury and McCamey to Ellendale, Kearney, Coshocton and Abu Dhabi, sits at announced status with no stated capacity or cost. The pace therefore rests entirely on company disclosure. Several of those sites depended on hosting partners, including Compute North (14861, 14590) and Beowulf (14791), so counterparty health matters as much as Marathon's own execution. Over the next 90 days, the most useful step is to extend the claims ledger past early 2022, so later filings can confirm whether the 2022 hosting deadlines were met or quietly dropped. The fifteen announced sites also need megawatt and cost figures. Where available, those figures should be checked against ERCOT and other queue filings for the Texas locations and against state utility dockets in Ohio, North Dakota and Nebraska. A site that stays at announced status with no capacity figure through another reporting cycle should be treated as unverified rather than in progress.

The claims supplied for this review end in March 2022, partway through claim 14331. None of the fourteen projects on record carries a stated capacity or cost. That includes the later sites at Granbury, Ellendale, Abu Dhabi and the Ohio locations. So this read covers the early Quebec, Hardin and Compute North phase, not Marathon's current footprint. Capacity framing grew fast but kept being reshaped. The fleet target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858). The equipment orders behind it also escalated, from 1,660 S-19 Pros (14853) to 10,500 (14842). At Hardin, the 100 MW envelope held (14789, 14636), but the split changed. A 30 MW Phase I (14790) became a 70 MW Phase I and a 30 MW Phase II (14645, 14646, 14637, 14638). That is a real change in near-term deliverable load, presented without comment. Dates have slipped more often than they have held. The 700 MicroBT units were dated 2020-08-15 in one filing (14872) and 2020-06-30 eight days later (14852), with no explanation. The Compute North hosting commitment for 73,000 miners was due 2022-03-31 (14591, 14592) and was later restated to 2022-06-30 (14346). A March 2022 filing still referenced the original date (14311). The SelectGreen S-9 purchase is the clearest continuity problem. It was marked withdrawn in November 2019 (15062), then reappeared in 2020 filings still carrying its 2019-09-30 promise date (14883, 14851). Against that, the February 2022 claim of 30,000 machines deployed and an 1,800 percent hash rate increase (16620, 16621) is the strongest delivery assertion in the record. However, no independent evidence is attached to test it. There are no permits, utility filings, interconnection-queue entries or site reporting. Compute North's September 2022 Chapter 11 filing is public record and bears directly on the 73,000-miner hosting plan, which the record never revisits. Over the next 90 days, watch whether filings attach megawatt figures and energization dates to the Texas, North Dakota, Ohio and Abu Dhabi sites. Check whether those figures can be matched to ERCOT, MISO, SPP or PJM interconnection records. Also check whether hosted-versus-owned capacity is disclosed separately, given how much of the early growth depended on third-party hosts.

The record supplied here stops in March 2022, partway through claim 14344. That limits this read to Marathon's early buildout, which is still instructive. None of the fourteen named sites has a capacity or cost field populated, including the later Texas, North Dakota, Nebraska, Ohio and Abu Dhabi locations. So there is no current baseline against which to measure delivery.

Within the visible window, announced scale grew quickly, and the stated figures moved more than once. The fleet target went from 1,400 units (15220) to 6,000 S-9s (15150), then to 240 PH/s (14906, 14887) and 294 PH/s (14858) within about two months. It then jumped to 10,500 S-19s (14842). At Hardin, Phase I was first described as 30 MW of a 100 MW total (14790, 14789). Seven months later it appeared as 70 MW, with Phase II at 30 MW (14645, 14637, 14638). No explanation was given for the change. The Compute North hosting commitment began at 73,000 miners (14591). By December 2021 it was expressed through a joint venture covering 30,000 (14348), even though the 73,000 figure was repeated alongside it (14346).

Dates have not held well. Nearly every claim with a promised-by date is flagged delayed. These include the mid-August 2020 S-19 deliveries (14873, 14854), the MicroBT order due by June 2020 (14852), the 10,000-unit tranche due by March 2021 (14762), and the Compute North deadline. That deadline moved from March 2022 (14592) to June 2022 (14346) without any acknowledgement in the claims themselves. The SelectGreen purchase was marked withdrawn (15062), then reappeared as completed (14883, 14851). That sequence deserves reconciliation against the underlying filings.

The record contains no independent corroboration of these figures. There are no permits, no interconnection-queue entries and no site-level reporting. The one hard milestone, 30,000 machines deployed (16620), is company-sourced. It should also be read against Compute North's September 2022 bankruptcy, which falls outside this dataset.

Over the next 90 days, the first test is whether the Q3 2026 10-Q, due in early November, attaches megawatts and energization dates to the Texas, Ohio and Plano sites. The second is whether matching ERCOT and PJM queue entries or county filings appear for Granbury, McCamey, Garden City, Matagorda, Long Ridge and Coshocton. Until the claim history is extended past 2022, the pattern on record is one of repeated upward restatement paired with routine slippage.

The record supplied here ends in March 2022, mid-claim, so this read covers Marathon's hosted-mining and Hardin era and says nothing about the later Texas, North Dakota, Nebraska, Ohio or Abu Dhabi sites. All fourteen projects on file sit at status announced with no capacity or cost attached, which is itself a gap worth closing before any of them is characterized. On scale, the fleet narrative ratcheted upward steadily. The deployed-hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858), then gave way to five-figure S-19 orders (14842, 14762), a 73,000-miner hosting commitment with Compute North (14591), and a claimed 30,000 machines deployed by February 2022 (16620). Hardin's 100 MW envelope held constant, but its internal split did not: Phase I was first 30 MW (14790), later restated as 70 MW with a 30 MW Phase II (14645, 14637, 14638), a reallocation that was not framed as a revision. Dates have a weaker record than capacity. The mid-August 2020 S-19 Pro delivery slipped (14873, 14854), and the MicroBT order carried a June 30 date in one filing (14852) and August 15 in another (14872). The Compute North hosting ramp moved from March 31, 2022 (14592) to June 30, 2022 (14346), and the 30,000-unit delivery schedule (14523) and Hardin power station maintenance (14285) are both flagged delayed. The SelectGreen S-9 purchase is the clearest inconsistency: marked withdrawn in November 2019 (15062), it reappears as a completed purchase in 2020 filings (14883, 14851) without any explanation of the reversal. Independent corroboration is absent from this record. There are no permits, interconnection-queue entries or site reporting attached to any project, so every capacity figure rests on company disclosure and, at Hardin, on a single counterparty power arrangement with Beowulf (14647, 14791). Over the next 90 days, the priority is completing the claim history past March 2022 and attaching capacity and cost to the announced sites. It is also worth checking whether the 2022-dated promises were ever explicitly closed out or quietly dropped. For the Texas sites, look for ERCOT large-load interconnection filings; for Ellendale and Jamestown, North Dakota PSC or MISO queue records; and for Hardin, Montana air-permit status. Those are the documents that would show whether the announced pace has a physical basis.

The visible record ends in March 2022, and what it shows is a company whose stated capacity grew fast while its dates slid repeatedly and without comment. None of the fourteen sites on file, from Granbury and McCamey to Ellendale, Kearney and Abu Dhabi, carries a capacity or cost figure, and all remain at announced status. Hardin is the only project with megawatt claims. The ledger therefore describes an equipment buildout more than a facilities one.

Capacity figures have been reshaped rather than simply scaled. The fleet target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858). The March 2022 figure of 240 PH/s (14286) matches the 2020 target, even though the company claimed 30,000 deployed machines and an 1,800 percent hash-rate increase (16620, 16621). That points to either a stale restatement or an extraction error, and it should be checked against the source filing. Hardin kept its 100 MW total (14789, 14636), but Phase I was first described as 30 MW (14790) and later as 70 MW, with 30 MW moved to Phase II (14645, 14637, 14638). No reason for the change appears in the record. The Compute North hosting commitment of 73,000 miners (14591) sits beside a joint venture sized at 30,000 (14348), so the 73,000 figure was never fully tied to a vehicle.

Dates have not held. The mid-August 2020 deliveries (14854, 14872, 14873), the 10,000-unit S-19 order due by March 2021 (14762), the Hardin services agreement (14734) and the Compute North program (14590, 14592) are all flagged delayed. The Compute North deadline moved from March to June 2022 (14346) without acknowledgement. The SelectGreen purchase was withdrawn (15062) and then reappeared as completed (14883, 14851).

The record contains no permits, interconnection-queue entries or on-the-ground reporting for any site. The announced pace therefore rests entirely on company statements. One external fact matters here: Compute North filed for Chapter 11 in September 2022, which undercuts the hosting claims above.

Over the next 90 days, the first priority is backfilling claims from 2022 to 2026 so that current sites are measured against current promises. For the Texas sites, ERCOT large-load interconnection filings are the key test. For Ellendale, Jamestown and Kearney, the relevant evidence is SPP or MISO queue positions. For Hardin and every other site, the question is whether energized megawatts, rather than nameplate targets, begin to appear in disclosures.

The record shows Marathon's announced capacity growing in steps, each step replacing the last before it was delivered. Early claims are small: 1,400 units (15220), then 6,000 S-9s (15150, 15136). By mid-2020 the target was 240 PH/s (14906, 14887, 14871), and weeks later it was 294 PH/s (14858). A 10,500-unit S-19 order followed (14842). Hardin introduced the first megawatt figure, a 100 MW total (14789, 14636), but the phase split changed from a 30 MW Phase I (14790) to a 70 MW Phase I with 30 MW moved to Phase II (14645, 14637). The total stayed the same while the near-term commitment more than doubled, and no reconciling statement appears. The Compute North hosting commitment of 73,000 miners (14591) was later matched by a 30,000-miner joint-venture vehicle (14348), which covers well under half of that headline.

Dated promises have not held. The SelectGreen purchase was due by September 2019 (15071), was marked withdrawn (15062), and then reappeared as completed and still delayed in 2020 (14883, 14851). The mid-August 2020 S-19 deliveries slipped (14854, 14873). The 10,000-unit batch due by March 2021 slipped (14762). The Compute North deadline moved from March 2022 (14590, 14592) to June 2022 (14346), and the 30,000 S-19j deliveries were also tied to June 2022 (14302, 14523). None of these revisions carries an explicit acknowledgement in the record. The new date simply replaces the old one.

The record contains no independent corroboration. There are no permits, no interconnection queue entries and no site reporting. The strongest delivery claim, 30,000 machines deployed and an 1,800 percent hashrate increase (16620, 16621), is self-reported. Ten of the fourteen listed projects, including Long Ridge, Coshocton, Ellendale, Kearney, the Texas sites and Abu Dhabi, have no capacity, cost or dated claim at all in the material provided. The input also appears to be cut off at claim 14317, so the later history is not assessed here.

Over the next 90 days, watch whether Marathon attaches megawatt figures and energization dates to the unquantified sites. Check those figures against ERCOT, SPP, MISO and PJM interconnection filings for the Texas, Dakota, Nebraska and Ohio locations. The key test is whether any new schedule is presented as a revision of a prior one rather than a silent replacement.

Marathon's capacity story in this record is one of rapid escalation measured in machines and hashrate rather than in sites. The 240 PH/s deployment target (14906, 14887, 14871) became 294 PH/s within weeks (14858). It was followed by orders of 10,500 and then 10,000 S-19 Pro units (14842, 14762), a 73,000-miner hosting commitment with Compute North (14591), and a claimed 1,800 percent hashrate increase (16621). Hardin is the only site with megawatt figures. Its 100 MW total held steady, but Phase I was restated from 30 MW (14790) to 70 MW (14645, 14637), which moved capacity forward in the schedule rather than adding any. The February 2022 claim of 30,000 machines deployed (16620) sits awkwardly beside the March 2022 purchase of 30,000 S-19j Pro units still due through June (14302). Readers should not treat ordered and energized fleets as interchangeable here.

Dates have slipped repeatedly, usually through restatement rather than acknowledgement. The 6,000-unit SelectGreen purchase was due by September 2019 (15071), was later marked withdrawn (15062), and reappears as completed in 2020 (14851, 14883) with no reconciliation. The S-19 Pro deliveries promised for mid-August 2020 (14854, 14873) were missed. The Compute North hosting target moved from March 2022 (14591) to June 2022 (14346) between filings without comment. That entire hosting channel was then exposed when Compute North entered bankruptcy later in 2022, a fact outside this claim set but material to reading 14346 and 14348.

The record contains no independent corroboration. There are no permits, no interconnection queue entries, and no site-level reporting. All fourteen listed projects remain at status announced with no capacity or cost attached. That includes Granby and Hardin, both of which the company's own claims describe as operating (15198, 14859, 14738). This is a gap in the evidence base, and it also suggests the project records need refreshing. The supplied claim history ends in March 2022, so the later sites (Granbury, McCamey, Garden City, Ellendale, Jamestown, Kearney, Abu Dhabi, the Ohio pair and Plano) have no dated claims to test here.

The first thing to watch in the next 90 days is whether monthly production updates report energized megawatts by site rather than fleet-wide exahash. The second is large-load interconnection filings in ERCOT for the Texas sites and in PJM for Long Ridge and Coshocton, which would give the first external check on announced pace. The third is any restatement of Hardin or Granby status that would let this record move past announced.

The capacity record for Marathon is mostly a story of fleet counts rather than facility megawatts, and the one site with firm power figures shows consistent totals but changing phases. Hardin was announced at 100 MW with a 30 MW Phase I (14789, 14790). Seven months later the same 100 MW was split 70/30 (14636, 14637, 14638), with Phase I costed at $23 million (14645). That tripled the near-term tranche without restating the ceiling, so the change reads as a quiet acceleration rather than an expansion. Hashrate guidance moved the same way. The 240 PH/s target was repeated three times (14906, 14887, 14871) before it became 294 PH/s (14858). Machine counts climbed from 6,000 S-9s to 10,000-plus S-19s (14842, 14762) and then to a 73,000-miner hosting plan with Compute North (14591), against 30,000 machines reported deployed in early 2022 (16620).

Dates have held poorly. The mid-August 2020 S-19 Pro delivery (14873, 14854), the June 2020 MicroBT purchase (14852), the 30,000-unit ASIC schedule (14523) and the Bitmain large contract due January 2022 (14343) are all flagged delayed. The Compute North target moved from March to June 2022 (14591 to 14346) without the earlier date being acknowledged as missed. The SelectGreen purchase shows a worse pattern. It was marked withdrawn (15062), then reappeared the following year as a completed acquisition (14883, 14851) with no explanation of the reversal.

None of the supplied claims are corroborated by independent evidence. The record contains no permits, interconnection queue positions or site reporting for any location. Most of the fourteen projects on file, from Ellendale and Jamestown to Garden City, McCamey, Granbury, Long Ridge and Abu Dhabi, carry no capacity or cost figures at all. The claim history also ends in March 2022, mid-record (14284), so any post-2022 build-out cannot be assessed from this data. That gap is itself the main finding.

Over the next 90 days, the third-quarter filing due in early November should show whether Marathon assigns megawatt figures to its Texas, North Dakota, Nebraska and Ohio sites or keeps reporting in exahash. The ERCOT, MISO, SPP and PJM interconnection queues are worth checking for entries matching Granbury, Garden City, Ellendale, Kearney and Long Ridge, since those would give the first third-party check on announced pace. Any fresh phase figures for Hardin-era sites should be compared against the 70/30 split to see whether phase sizes are again being reset without comment.

The dominant pattern in this record is restatement rather than progression. A single 6,000-unit S-9 purchase was announced or re-announced at least five times between August and November 2019 (15150, 15141, 15136, 15101, 15071), withdrawn once (15062), then re-presented as new in mid-2020 (14883) and still carried against its original September 2019 date in the retrospective filing (14851). The Hardin services agreement follows the same shape, surfacing four separate times across seven months (14756, 14738, 14734, 14640) with only 14734 acknowledging slippage from the October 2020 commitment. Readers counting announcements would materially overstate activity.

Capacity scaling is more equivocal than the headline numbers suggest. The 240 PH/s target was repeated three times in six weeks (14906, 14887, 14871) before being revised upward to 294 PH/s (14858) with no reconciliation of the earlier figure. At Hardin the total held at 100 MW (14789, 14636) while the phase split inverted: Phase I moved from 30 MW to 70 MW and Phase II absorbed the remainder (14790 versus 14645, 14646, 14637, 14638). That is relabeling of the same envelope, not expansion. Genuine scale arrives only with the Compute North commitments — 73,000 units against a Q1 2022 date (14590, 14591, 14592) narrowed to 30,000 in the joint venture (14347, 14348) — and the deployment confirmed in February 2022 (16620, 16621) landed at the narrower number while the 73,000 figure lapsed without comment (14346).

Independent corroboration is thinnest exactly where it matters. Hardin's PPA pricing and Granby's lease expiry are verifiable, but eleven of the fourteen sites on record carry no capacity, no cost, and no status beyond announced, and the claim series itself terminates in early 2022 — leaving the Texas, North Dakota, Ohio, Nebraska and Abu Dhabi sites unsupported by any capacity assertion at all.

Over the next ninety days, watch whether those sites acquire quantified capacity in filings or remain placeholders; whether ERCOT large-load approvals and MISO queue positions appear for Granbury, McCamey, Ellendale and Coshocton; and whether energized megawatts are reported separately from contracted megawatts. Counterparty concentration in hosting remains the single largest unpriced risk in this trajectory.

The dominant pattern here is redenomination rather than compounding. The unit of measure shifts every eighteen months or so — square footage at Granby (15230), then machine counts, then aggregate hashrate, then percentage growth (16621) — and each shift resets the baseline against which the prior promise would have been judged. The 2020 hashrate target is a clean example: 240 PH/s is stated three times across six weeks (14906, 14887, 14871) and then becomes 294 PH/s (14858) without the earlier figure being retired or reconciled, even though the component arithmetic (14856, 14857) implies a different fleet than the one originally described.

Dates have slipped in a specific way: they are absorbed into completion announcements. The SelectGreen 6,000-unit purchase was promised for 2019-09-30 (15071), withdrawn (15062), and then reported as purchased on 2020-08-13 (14851) still carrying the original date — an eleven-month slip presented as delivery. The same transaction also appears as four separate announcements (15150, 15141, 15136, 15101), which inflates apparent activity without adding capacity. Delivery-window claims behave similarly: mid-August 2020 for 660 S19 Pros (14873, 14854), June 2020 for the M30S+ tranche (14852), and the MicroBT order (14872) all lapsed. The larger ambitions — 30,000 units (14523), then 73,000 hosted with Compute North (14590, 14591, 14592, 14346) — were all anchored to the first half of 2022 and all missed, which matters because that hosting counterparty later failed.

Hardin shows quiet phase redefinition: Phase I is 30 MW against a 100 MW total in October 2020 (14790, 14789), then Phase I is 70 MW and Phase II 30 MW in May 2021 (14645, 14646, 14637, 14638). The headline total held; the near-term deliverable tripled on paper without comment.

On corroboration, the supplied record is entirely company-sourced. No permits, interconnection queue positions, or site reporting appear against any entry, and the eleven newer sites — Ellendale, Jamestown, Kearney, Coshocton, Plano, Garden City, McCamey, Granbury, Matagorda, Long Ridge, Abu Dhabi — carry no capacity, cost, or date at all. That absence is the finding.

Over the next ninety days, watch whether any of those sites acquires a megawatt figure and an energization date in the Q3 filing, whether MISO, SPP and PJM queue entries and county air permits appear at Ellendale, Coshocton and Long Ridge, and whether Abu Dhabi reports commissioned rather than contracted capacity.

The record supplied here terminates mid-entry at claim 14344 (March 2022), so what follows reads the 2018–2022 arc only; nothing should be inferred about the period since.

The scale trajectory is steep and almost entirely procurement-led. A 26,700 square foot leased room in Quebec with 1,400 machines (15230, 15220) becomes a 240 PH/s target (14906, 14887, 14871), then 294 PH/s (14858) within eight days of the prior figure, with no acknowledgement that the earlier number had been superseded. By mid-2021 the unit of account has shifted again to 73,000 hosted miners (14591). The growth is real in the sense that orders were placed; it is not evidence of energized capacity, because every step is denominated in units bought rather than megawatts running.

The date record is where the pattern hardens. The SelectGreen purchase is the clearest case: promised for September 2019 (15071), marked withdrawn in November 2019 (15062), then reintroduced in July 2020 (14883) and restated as completed in August 2020 (14851) against the original 2019 deadline. A cancellation was quietly reversed without narrative reconciliation. Smaller slips repeat the shape — mid-August 2020 deliveries (14854, 14873), a June 2020 MicroBT order (14852), and the 73,000-miner Compute North timeline that moved from March 2022 (14590, 14592) to mid-2022 (14346) without being flagged as a revision. The February 2022 disclosure of 30,000 machines deployed (16620) is roughly forty percent of what had been promised by then.

Independent corroboration is uneven. Hardin is the strongest case: an exclusive PPA at a stated rate, a land lease, and an O&M agreement against an existing generating asset (14791, 14802, 14801, 14639) are checkable instruments. But the Phase I boundary silently moved from 30 MW (14790) to 70 MW (14645, 14637) while the 100 MW total held, which makes any "Phase I complete" claim non-comparable across filings. The fourteen listed sites — Long Ridge, Matagorda, Abu Dhabi, Ellendale, Kearney and the rest — carry no capacity or cost at all, so no permit or queue check is even possible against them.

Over the next ninety days, watch whether any of those announced sites attaches a megawatt figure and an interconnection queue position, whether the Hardin phase definitions are restated again, and whether the 73,000 target is reconciled or simply retired.

The record divides cleanly into two eras with very different disclosure quality. Through 2019 and 2020 the company operated at a scale small enough to describe precisely — a single leased building in Granby and a few thousand machines — and the failure mode was slippage rather than vagueness. The SelectGreen S-9 purchase is the clearest example: announced or re-announced at least five times (15150, 15141, 15136, 15101, 15071), carried once as withdrawn (15062), and ultimately reported closed roughly eleven months after its stated date (14851). Delivery dates set for mid-August 2020 slipped as a group (14873, 14854, 14872), which suggests supplier-side constraint rather than site readiness.

Capacity language escalated without reconciliation. A 240 PH/s target was restated three times in six weeks (14906, 14887, 14871), then replaced by 294 PH/s (14858) with no note that the earlier figure had been superseded. The Hardin phasing shows the same habit in megawatts: the 100 MW total held constant from October 2020 to May 2021 (14789, 14636), but Phase I was quietly recut from 30 MW (14790) to 70 MW (14645, 14637), which changes the near-term energization story substantially while leaving the headline number untouched. The Compute North arrangement is the largest gap between announcement and execution — 73,000 hosted miners promised by Q1 2022 (14590, 14591, 14592), rescheduled to mid-2022 (14346), and materializing as a 30,000-machine joint venture at 82 percent (14347, 14348, 16620), roughly two-fifths of the announced figure.

Independent corroboration is the weakest part of this file. Nothing here carries a permit number, an interconnection queue position, or a site visit. Hardin is a particular blind spot, since power arrived through a PPA at an existing generating asset, so no new queue entry would ever appear to confirm or contradict the schedule. Meanwhile the site list has grown to fourteen locations, every one carried as announced with capacity and cost fields empty — footprint expanding faster than specificity.

Over the next ninety days, the useful tests are narrow: whether any of the newer Texas, Dakota, Ohio or Abu Dhabi sites acquires a stated megawatt figure, whether phase splits get restated again without acknowledgement, and whether ERCOT and MISO queue filings or state siting dockets appear under a Marathon-affiliated name. Absent those, capacity here remains self-reported.

The most durable pattern in this record is not the size of the numbers but the instability of the unit they are expressed in. The 2019 disclosures measure the business in square feet and machine counts (15230, 15220, 15150). By mid-2020 the metric is aggregate hashrate, with 240 PH/s repeated across three filings in under two months (14906, 14887, 14871) before being revised upward to 294 PH/s within a week (14858). By late 2020 the frame shifts again to megawatts at Hardin (14789, 14790), then in 2021 to raw fleet counts of 73,000 miners (14591, 14346), then in early 2022 to a percentage growth figure (16621) that has no denominator a reader can independently check. Each change of unit arrives roughly when the prior unit would have invited a comparison to what was actually delivered.

Dates have slipped persistently and, more importantly, have been re-dated without reconciliation. The SelectGreen purchase is the clearest case: promised by September 2019 (15071), marked withdrawn (15062), reasserted as a fresh purchase in July 2020 (14883), and finally booked as completed a year past its original deadline (14851) with no acknowledgement of the intervening history. The August 2020 delivery cohort (14873, 14854) and the Compute North hosting commitments converging on mid-2022 (14592, 14590, 14591, 14523, 14346) follow the same shape. A services agreement signed in April 2021 carries an October 2020 effective date (14734), which is a retroactive paper fix rather than a delivery.

Hardin is the exception and deserves credit for it: an exclusive PPA at a stated price, a land lease, and an O&M agreement (14791, 14801, 14802, 14639) constitute real contractual scaffolding that leaves traces in state and utility records. Nothing comparable exists for the eleven other sites on the roster, which carry no capacity, no cost, and no date at all. There is, in other words, nothing yet to test against permits or interconnection filings.

Over the next ninety days, watch whether any of those capacity-less sites acquires a first dated commitment, whether the Compute North joint venture hosting figures (14347, 14348) are restated rather than repeated, and whether the mid-2022 delivery cohort is reconciled explicitly or simply disappears. Note also that the supplied record terminates in March 2022 (14340); anything after that is outside what can be assessed here.

The most telling pattern in this record is not expansion or contraction but migration of the measuring unit. Capacity is first stated in square feet (15230), then in miner counts (15220, 14842), then in petahash (14906, 14871, 14858), then in megawatts (14789, 14637). Each shift resets the baseline and makes year-over-year comparison a matter of trust rather than arithmetic. The 1,800 percent figure at 16621 is the clearest artifact of that: a percentage against a base small enough that the number carries no information about delivered load.

Where a stable unit does persist, the revisions are quiet. Hardin's headline total holds at 100 MW from October 2020 through March 2022 (14789, 14636, 14303), but the phase split is re-cut underneath it — Phase I goes from 30 MW (14790) to 70 MW (14645, 14637) with no acknowledgement, and Phase II absorbs the remainder (14646, 14638). The total looks stable precisely because the internals moved. Similarly, the SelectGreen purchase is announced, withdrawn (15062), and re-announced across roughly fourteen months (15150, 15141, 15136, 15101, 15071, 14883, 14851) against a delivery date of 2019-09-30 that was still being described in the past tense in August 2020. Dates that have a specific anchor slip on schedule: mid-August 2020 twice (14873, 14854), and the Compute North hosting commitment moves from 2022-03-31 (14590, 14591, 14592) to 2022-06-30 (14346, 14523) while the vehicle actually formed covers 30,000 miners at 82 percent (14347, 14348), not 73,000.

Independent corroboration is strongest exactly where the economics are most specific — the Beowulf PPA at $0.028/kWh and $0.006/kWh O&M (14647, 14639, 14650) sit against a real generating asset with a real interconnection. It is weakest across the fourteen sites listed, every one of which carries status announced with capacity and cost fields empty. Nothing in this record substantiates Abu Dhabi, Long Ridge, Matagorda, Granbury, Plano, Kearney or the Dakota sites; the claim series itself terminates in March 2022.

Over the next ninety days, the checks worth running are ERCOT and MISO queue positions for the Texas and North Dakota locations, county-level permit and electrical inspection filings, and the next quarterly disclosure of energized versus contracted megawatts. Watch also for sites quietly leaving the announced list without a withdrawal notice — that has happened before here, and only once was it labeled.

Marathon's stated capacity has grown, but not in a straight line, and not always in the direction the headline numbers imply. The 2019 posture was a single leased Quebec footprint of roughly 26,700 square feet with leases running to November 2022 (15230, 15229, 15219) and equipment measured in single-digit thousands of units (15150, 15141, 15136, 15101). Through 2020 the company restated the same 240 PH/s target three separate times without change (14906, 14887, 14871) before nudging it to 294 PH/s on incremental purchases (14858), then jumped an order of magnitude with 10,500 and 10,000 unit orders (14842, 14762). The more telling movement is at Hardin, where total capacity has never moved off 100 MW (14789, 14636, 14303) while the phase split was quietly re-cut: Phase I was 30 MW in October 2020 (14790) and 70 MW by May 2021 (14645, 14637), with the 30 MW figure reassigned to Phase II (14646, 14638). That is a re-labeling, not an expansion, and it was not flagged as a revision.

Dates have slipped as a matter of routine rather than exception. Miner deliveries promised for mid-August 2020 slid (14873, 14854, 14872), the SelectGreen purchase carried a September 2019 date still being reported as outstanding a year later (14851), and the Compute North hosting program promised for the first half of 2022 (14590, 14591, 14592, 14523) remained a forward commitment into December 2021, by which point the contracted vehicle covered 30,000 miners (14348) while the 73,000 figure continued to be quoted (14346, 14287). A two-thirds gap between the contracted and the advertised number is the single most important discrepancy in this record.

On independent corroboration, the record is thin rather than contradictory: fourteen sites — Long Ridge, Matagorda, Granbury, McCamey, Garden City, Ellendale, Jamestown, Kearney, Coshocton, Plano, Abu Dhabi and others — all sit at announced status with no capacity or cost attached, and no permit, interconnection queue, or site-visit evidence is tied to any of them here. Treat announced pace as unverified.

Over the next 90 days, watch whether any of those fourteen converts to a status carrying a megawatt figure, whether Texas and Ohio sites surface in ERCOT or PJM interconnection filings, whether the 73,000 number is retired or reconciled, and whether hosting concentration is disclosed as a risk rather than a capacity claim.

The through-line here is a company that repeatedly changed the unit in which it measured itself, and each change reset the baseline before the prior one could be judged. The 2019 record is small and concrete: a leased Quebec footprint measured in square feet with a lease horizon (15230, 15229, 15219) and a deployment counted in individual machines (15220). By mid-2020 the metric had become aggregate hashrate, with a 240 PH/s target restated three times in five weeks (14906, 14887, 14871) and then revised to 294 PH/s (14858) without the earlier number being retired. By October 2020 the unit changed again to megawatts at Hardin, initially 30 MW in Phase I against a 100 MW total (14790, 14789). Seven months later the same 100 MW total was re-cut as 70 MW in Phase I and 30 MW in Phase II (14645, 14646, 14637, 14638). The headline held; the phasing was silently inverted, which matters because Phase I is the part that gets built.

Dates have slipped in a recognizable pattern. The SelectGreen 6,000-unit purchase carried a September 2019 completion (15071), was withdrawn in November 2019 (15062), reappeared as a live purchase in July 2020 (14883), and was still being described in the past tense in August 2020 against the original 2019 date (14851). The mid-August 2020 delivery windows (14872, 14873, 14854) and the Compute North hosting commitment (14590, 14591, 14592) followed the same shape, with the 73,000-miner figure retained (14346) even as the contracted tranche was narrowed to 30,000 in the joint venture (14348, 14347).

On corroboration, the record is thin in a specific way: every claim traces to a counterparty agreement or a company statement. There are no permits, interconnection queue entries, or independent site reporting in this dataset, and all fourteen listed projects, including Abu Dhabi, Long Ridge, Matagorda, Ellendale and Plano, sit at announced status with capacity and cost blank. The claim series itself also goes quiet after March 2022 (14344, 14318), so nothing here supports conclusions about the current portfolio.

Over the next ninety days, the tells are whether any announced site acquires a megawatt figure and a queue position rather than a press date, whether the Hardin phase split is reconciled, and whether the lapsed Quebec lease horizon is addressed rather than left to expire quietly.

The shape of this record is one of steady unit inflation paired with steady date erosion, and the two rarely appear in the same disclosure. The earliest phase measures itself in square feet and machine counts (15230, 15220) because there was no megawatt story to tell yet; the shift to MW framing arrives only with Hardin in late 2020 (14790, 14789). That transition matters, because capacity units changed faster than capacity did.

The clearest tell is repetition. A single 6,000-unit SelectGreen order is announced or re-announced across at least five filings (15150, 15141, 15136, 15101, 15071), is marked withdrawn once (15062), and then resurfaces as accomplished (14883, 14851) still measured against a 2019-09-30 date it missed by roughly ten months. Nothing in the sequence acknowledges the gap; the order simply reappears as news. The same mechanism recurs with Compute North, where 73,000 hosted miners slides from a Q1 2022 date (14590, 14591, 14592) to mid-2022 (14346) with the headline number untouched and only 30,000 actually vehicled (14347, 14348). Slippage is absorbed by re-dating rather than by resizing.

Hardin deserves separate attention. Total capacity is constant at 100 MW (14789, 14636), but the phase split is quietly re-cut from a 30 MW Phase I to a 70 MW Phase I between October 2020 and May 2021 (14790 versus 14645, 14646). A total that holds while the phase boundary moves usually means the boundary was redrawn around work already done.

Independent corroboration is thin, and structurally so. Hardin sits behind the meter on a Beowulf PPA (14647, 14639), which bypasses the interconnection queue and removes the external checkpoint an analyst would normally use. Verification has to come from county building permits, state air and siting filings, and plant-level generation data. Meanwhile nine of the fourteen listed sites carry no capacity or cost at all, so announcement volume is running well ahead of specification.

Note that this record terminates in March 2022 (14318, truncated), so the read stops there. Over the next ninety days the things to watch are whether any announced-only site acquires a disclosed MW figure and an energization date, whether hosted-miner counts are restated again without reference to prior targets, and whether energized MW is ever reported alongside contracted MW.

The arc here is less a story of growth than of a shifting yardstick. In mid-2019 Marathon described itself in square feet and machine counts: a 26,700 square foot leased site in Quebec (15230) and 1,400 units deployed (15220). By mid-2020 the unit of account had become hashrate, with 240 PH/s repeated across three announcements (14906, 14887, 14871) and then restated as 294 PH/s eight days later (14858). By late 2020 the unit changed again to megawatts at Hardin, and by 2021 to hosted fleet size, culminating in 73,000 miners placed with Compute North (14591, 14346). Each reframing resets the baseline, and the effect is that no two consecutive announcements are directly comparable.

The most instructive revision is Hardin Phase I, announced at 30 MW in October 2020 (14790) and restated at 70 MW by May 2021 (14645, 14637) with the 30 MW figure migrating to Phase II (14646, 14638). The total stayed at 100 MW (14789, 14636); only the internal sequencing moved, without acknowledgement. The Compute North timeline shows the same pattern: a March 2022 implementation date (14590, 14592, 14591) reappears as June 2022 in the December 2021 joint venture disclosure (14346), a quarter absorbed into a new announcement rather than flagged as slippage.

Equipment dates have held poorly and consistently so. Machines promised by September 2019 (15071) were still being characterized as newly purchased in August 2020 (14851); mid-August 2020 deliveries slipped (14873, 14854, 14872); the 10,000-unit order for March 2021 slipped (14762); the 30,000-unit schedule for June 2022 slipped (14523). One purchase was withdrawn outright (15062), and the earliest transaction on file unwound with a three million share break fee (15461, 15477).

Independent corroboration is thin because the record is almost entirely self-reported. Hardin is the exception, anchored by a power purchase agreement and land lease (14647, 14802), and the 2022 disclosures of power station upgrades and maintenance owed by year end (14339, 14285) suggest capacity was announced ahead of what the host plant could carry. Eleven of the fourteen listed sites still carry no capacity or cost figure at all.

For the next ninety days, the series as supplied terminates in March 2022 and needs refreshing before any forward read is reliable. When it is, watch whether the announced-only Texas and North Dakota sites acquire a stated megawatt number and a matching interconnection queue position, and whether reported capacity distinguishes energized from contracted.

The visible record for Marathon is heavily front-loaded: nearly every facility-capacity and timeline claim dates from 2019 through early 2022, and the list cuts off mid-claim in March 2022 (claim 14344). Ten of the fourteen projects on file, including Ellendale, Kearney, Granbury, Garden City, McCamey, Abu Dhabi and Long Ridge, have no attached capacity, cost or timeline claims at all, so any read of those sites has to wait for the record to be filled in.

Within the visible window, announced scale grew quickly and was then restated. The company went from a 26,700 square-foot Quebec lease (15230) and a 240 PH/s fleet target repeated three times (14906, 14887, 14871) to a 294 PH/s target within weeks (14858). Hardin was framed as 30 MW Phase I of a 100 MW site (14790, 14789), then reframed as 70 MW Phase I plus 30 MW Phase II (14645, 14637, 14646). The total held, but the phase split more than doubled without an explanatory claim. Hosting ambitions followed the same pattern: 73,000 miners with Compute North (14591), then a joint vehicle scoped at 30,000 (14348) while the 73,000 figure was carried forward (14346).

Dates have been the weaker element. Nearly every dated promise is flagged delayed. Miner deliveries set for mid-August 2020 slipped (14873, 14854), as did the 10,000-unit S-19 Pro order due by March 2021 (14762). The Compute North hosting deadline moved from 2022-03-31 to 2022-06-30 (14592, 14346), and the 30,000-unit delivery schedule missed its mid-2022 target (14523). Hardin power-station upgrades and maintenance (14339, 14285) pushed dates into late 2022. The SelectGreen purchase was withdrawn (15062) and then reappears as completed at the same dollar figure (14851), a sequence that deserves a reconciling source.

The record contains no independent corroboration: no permits, interconnection-queue entries or site reporting. The announced pace therefore rests entirely on company statements, and the delay flags suggest discounting those statements.

Over the next 90 days, the priority is to backfill 2022-2026 claims for the Texas, North Dakota, Nebraska, Ohio and Abu Dhabi sites so that current capacity can be compared against earlier framing. Watch the Q3 2026 results for energized-MW and hashrate figures stated by site, and check any ERCOT, SPP or PJM queue filings tied to Granbury, Garden City, McCamey, Kearney or Coshocton. Any new AI or HPC capacity claim for existing sites should be tested against the delivery record shown above.

The record supplied here covers claims only through March 2022, ending mid-entry at claim 14344. Fourteen sites are listed, but none has a capacity or cost field populated. That includes the later-named locations: Ellendale, Granbury, Garden City, McCamey, Kearney, Coshocton and Abu Dhabi. So this read covers the early mining-fleet era and Hardin. It does not cover the current portfolio.

Capacity figures mostly scaled up, with one notable reshuffle. The fleet hashrate target sat at 240 PH/s across three statements (14906, 14887, 14871), then rose to 294 PH/s (14858) as S-19 Pro orders expanded. At Hardin, the 100 MW total held steady (14789, 14636). The phasing did not. Phase I was 30 MW in October 2020 (14790) and became 70 MW by May 2021 (14645, 14637), with a new 30 MW Phase II (14638). The later filings do not explain the change. The Compute North arrangement is harder to reconcile. The letter of intent spoke of hosting 73,000 miners (14591), while the joint vehicle covered 30,000 (14348), and the record does not say where the other 43,000 were to go.

Dates have slipped repeatedly, and at least twice without acknowledgement. The 73,000-miner hosting deadline moved from March 2022 (14592) to June 2022 (14346) without explanation. The 700 MicroBT units were first tied to June 2020 (14852) and then to mid-August 2020 (14872). Both dates carry a delayed tag. The 660 S-19 Pro delivery for mid-August 2020 slipped (14873, 14854), as did the 30,000-unit ASIC schedule (14523) and the Hardin power station upgrades due by end-2022 (14339, 14285). The Hardin services agreement was due in October 2020 (14734) but appears to have been executed in April 2021. The SelectGreen purchase was also marked withdrawn (15062) and later described as completed (14851, 14883).

The record contains no independent corroboration of any kind: no permits, no interconnection queue entries and no site reporting. The entire trajectory rests on company statements alone.

Over the next 90 days, the priority is to backfill this ledger. Watch for filings that attach capacity and energization dates to the named Texas, North Dakota, Nebraska and Ohio sites. Check those against ERCOT, MISO, SPP and PJM queue records, and watch whether the Hardin and hosted-capacity figures are quietly restated in current disclosures.

The capacity record here is mostly a record of hashrate and machine counts, not megawatts, and it stops in March 2022. Treat it as an account of Marathon's hosted-miner era, not of the site portfolio listed above it. Within that window, the announced scale rose quickly. Deployment targets went from a 1,400-unit Quebec fleet (15220) and 6,000 S-9 purchases (15150, 15136) to a 240 PH/s target (14906, 14887), then 294 PH/s (14858), then 10,000 to 10,500 S-19 Pro orders (14762, 14842), and finally a 73,000-miner hosting commitment with Compute North (14591, 14346). Hardin is the one site with stated megawatts. It began as a 30 MW Phase I inside a 100 MW envelope (14790, 14789) and was restated seven months later as 70 MW Phase I plus 30 MW Phase II (14645, 14646, 14637, 14638). The total held at 100 MW while the phasing was pulled forward, and the record does not explain the change.

Dates have held poorly and have mostly moved without acknowledgement. Every dated delivery claim is tagged delayed. That includes the mid-August 2020 S-19 and M31S+ shipments (14873, 14872, 14854), the March 2021 S-19 order (14762), the 30,000-unit ASIC schedule to mid-2022 (14523), and the Compute North hosting deadline. That deadline was first set for March 2022 (14592, 14332) and then quietly restated as June 2022 (14346). The Hardin services agreement carries a promise date that had already passed when it was filed (14734). The SelectGreen purchase was withdrawn (15062) and then reappeared as a completed acquisition (14851, 14883). That sequence deserves a documentary check.

The claim set contains no independent corroboration for any site: no air or building permits, no interconnection queue positions, no utility filings, and no on-the-ground reporting. The pace in these claims is therefore self-reported. The record's own trajectory depends heavily on third parties, with Beowulf supplying Hardin's power and Compute North supplying hosting. That dependence is the main risk the data exposes.

Over the next 90 days, the priority is closing the gap between March 2022 and today. Granbury, Garden City, McCamey, Ellendale, Jamestown, Kearney, Coshocton, Long Ridge, Matagorda and Abu Dhabi all sit at announced status with no capacity or cost on record. Each one needs dated megawatt claims, ERCOT, SPP, MISO or PJM queue entries, and county permit records before its pace can be judged. Hardin needs a documented end state, and the Compute North commitment needs a documented outcome.

The first thing to note about Marathon's record is its shape. Fourteen projects are on file, all still marked announced, none with a stated capacity or cost. The claim history ends in March 2022 (14333). Only Hardin, Granby and Quebec are tied to any claims. Granbury, Kearney, Garden City, McCamey, Ellendale, Jamestown, Coshocton, Plano, Long Ridge, Matagorda and Abu Dhabi have no facility-level claims in this extract, so no trajectory can be read for them yet.

Where there is data, headline capacity mostly grew while its components were reshuffled. The fleet hashrate target of 240 PH/s (14906, 14887, 14871) rose to 294 PH/s within weeks (14858). The size of the order book grew even faster, from 1,660 S-19 Pros (14853) to 10,500 (14842). Hardin's total stayed at 100 MW (14789, 14636), but Phase I went from 30 MW (14790) to 70 MW (14645, 14637), and Phase II took the remaining 30 MW (14638). That is a pull-forward of phasing, not new capacity.

Dates have held poorly, and several were revised without saying so. The Compute North hosting commitment for 73,000 miners was first tied to 2022-03-31 (14591, 14592). It was later restated with 2022-06-30 (14346), yet 14332 in March 2022 still cites the March date. The 30,000-unit ASIC delivery schedule (14523), the mid-August 2020 S-19 Pro delivery (14873, 14854), the Hardin services agreement (14734) and Hardin power-station maintenance (14285) are all flagged delayed. The SelectGreen purchase is marked withdrawn in 14062 but reappears as completed in 14851, a contradiction the company's own filings should resolve.

The record contains no independent corroboration: no permits, interconnection queue positions or site reporting for any project. Outside the claim set, Compute North's Chapter 11 filing in September 2022 is the obvious test of the 73,000-miner commitment. It should be sourced and attached before any conclusion about how that capacity was re-homed.

Over the next 90 days, the priority is closing the gap between 2022 and today. The Q3 2026 10-Q, due in early November, should give site-level megawatts and energization dates for the owned Texas, Nebraska, North Dakota and Ohio sites. ERCOT and MISO/SPP queue entries can check whether announced Texas and Plains capacity has firm interconnection. Any disclosure on the Abu Dhabi arrangement would need to separate owned capacity from joint-venture capacity.

Marathon's record here is a story of escalating unit counts rather than growing facility footprint, and the ledger stops in March 2022 while the project list names fourteen sites. The last ten of those sites, from Granbury and the West Texas locations through Ellendale, Kearney, Abu Dhabi and Coshocton, carry no capacity, cost or dated claim at all. The analysable trajectory is therefore Hardin, Quebec/Granby and the Compute North hosting relationship.

On scale, the hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s (14858) within weeks. Orders compounded from 660 and 500 S19s (14869, 14870) to 1,660 (14853), then 10,500 (14842), a 30,000-unit delivery schedule (14523) and 73,000 hosted miners (14591, 14346). Hardin's total held at 100 MW (14789, 14636), but the phase split was redrawn from a 30 MW first phase (14790) to 70 MW (14637, 14645) with no stated explanation. That rebalancing pulls most of the capacity forward and deserves scrutiny.

On dates, nearly every dated commitment is flagged delayed. The mid-August 2020 S19 and M31S+ deliveries (14873, 14854, 14872) slipped, and so did the 10,000-unit order due by March 2021 (14762). The Compute North ramp is the clearest unacknowledged revision. It was promised by 2022-03-31 (14591, 14592), restated to 2022-06-30 (14346), yet the March 2022 filing still cites the March deadline (14332) weeks before it lapsed. The SelectGreen purchase shows a similar inconsistency. It was marked withdrawn (15062), then described as completed in later filings (14883, 14851), and nothing reconciles the two.

Independent corroboration is absent from this record. Hardin runs on Beowulf's own generation under an exclusive PPA (14647), so the relevant check is plant availability, and the delayed maintenance item (14285) is the only operational signal. The hosted capacity depends entirely on a partner's build-out, and no permits, queue positions or site reporting are attached.

Over the next 90 days, the priority is whether filings attach megawatt figures and energisation dates to the ten unclaimed sites. Those figures can then be checked against the relevant interconnection queues: ERCOT for the Texas sites, SPP and MISO for North Dakota and Nebraska, and PJM for Ohio. Also watch whether Marathon ever reconciles the 73,000-miner and 100 MW Hardin commitments against capacity that was actually energised.

The record on file is a narrow window, and it does not match the project list. Every facility and timeline claim runs from 2018 to early 2022, and the list cuts off at claim 14284. None of the eleven sites now carried as announced has a capacity, cost, or date claim attached. Long Ridge, Granbury, Matagorda County, Abu Dhabi and the others are all blank. So what follows reads the early trajectory, not the current portfolio. Readers should treat conclusions about the named sites as unsupported by this dataset.

Within that window, announced capacity mostly grew, and the targets moved quickly. The fleet hashrate target of 240 PH/s (14906, 14887, 14871) became 294 PH/s within weeks (14858). The Hardin, Montana build held its 100 MW headline (14789, 14636), but the phases were re-cut without stated explanation. Phase I was 30 MW in October 2020 (14790) and 70 MW by May 2021 (14637, 14645), with a 30 MW Phase II added (14638, 14646). Hosting ambitions followed the same path: a 73,000-miner Compute North arrangement (14591), later partly formalized as a 30,000-miner joint vehicle (14348, 14347).

Dates held less well than headline numbers. The mid-August 2020 S-19 Pro delivery (14873, 14854) is flagged delayed. So is the 700-unit MicroBT order (14872, 14852). Every major Bitmain contract restated in March 2022 carries a missed promise date, spanning March 2021 through June 2022 (14341, 14340, 14342, 14343, 14301, 14344). The Compute North hosting completion moved from 2022-03-31 (14591, 14592) to 2022-06-30 (14346). Neither revision came with an acknowledgement in the claim text. The pattern is consistent: order sizes and target totals are announced firmly, while delivery and energization dates drift one to two quarters.

The record contains no independent corroboration. There are no permits, no interconnection queue positions, and no site reporting for Hardin, the Compute North sites, or any currently listed project. The pace of announcements therefore cannot be checked against physical progress here.

The next 90 days should close that gap. First, check whether the upcoming quarterly filing states energized megawatts or hashrate by site for the eleven announced locations. Second, look for interconnection or large-load filings: ERCOT for the Texas sites, PJM or Ohio regulatory dockets for Long Ridge and Coshocton, and the relevant regional operators for the North Dakota and Nebraska sites. Third, watch whether the Abu Dhabi entry acquires any disclosed capacity or partner terms. Until site-level numbers enter the record, the listed projects remain names rather than measurable commitments.

The first thing to say about this record is what it does not contain. All eleven projects now on file for Marathon, from Long Ridge in Ohio to Matagorda County, Granbury, Abu Dhabi and the North Dakota and Nebraska sites, carry no stated capacity, cost or dated milestone. The claims ledger as supplied also ends in March 2022, partway through claim 14284. Any judgment on the current portfolio therefore rests on the company's earlier behavior, not on commitments tied to these sites.

That earlier behavior shows headline capacity that expanded mostly by reshuffling. At Hardin, the 100 MW envelope held from claim 14789 through 14636. Phase I, however, was first described as 30 MW (14790) and later restated as 70 MW, with Phase II at 30 MW (14645, 14637, 14638). The total stayed the same while the near-term portion more than doubled. The fleet hashrate target moved from 240 PH/s (14906, 14887, 14871) to 294 PH/s within weeks (14858). The Compute North hosting program was framed as 73,000 miners (14591) and later split into a 30,000-miner joint-venture tranche (14348) alongside the same 73,000 figure (14346). The record does not show a reconciliation of the two numbers.

Dated promises have held poorly. Nearly every timeline in the file is marked delayed. That includes the mid-August 2020 S-19 Pro delivery (14873, 14854), the 10,000-unit order due by March 2021 (14762), and each Bitmain contract tranche from March 2021 through June 2022 (14341, 14340, 14342, 14343, 14344), along with the 30,000-unit schedule (14523). The Compute North deadline moved from March 2022 (14592) to June 2022 (14346). The file contains no claim acknowledging that change. The 6,000 S-9 purchase promised for September 2019 (15071) was still being described in August 2020 (14851). Much of the slippage sits with suppliers and hosts, not Marathon's own construction, but the pattern of reliance on third parties is the point.

No independent corroboration appears in the record. There are no permits, interconnection queue positions or site reporting, so the announced pace cannot be tested against physical progress.

Over the next 90 days, the useful signals are whether any megawatt or energization dates are attached to the eleven named sites. For the Texas locations, check for matching ERCOT large-load or queue filings, and for Long Ridge and Coshocton, PJM or local permitting activity. Also watch whether Abu Dhabi figures are disclosed with a named counterparty. Given the history, new dates should be discounted until a filing independent of the company supports them.

The record as supplied stops partway through March 2022 (it breaks off at claim 14332), so this read covers Marathon's mining-era buildout rather than the eleven sites now listed as announced. That gap is the first finding. None of those projects, from Granbury and McCamey to Ellendale, Coshocton and Abu Dhabi, has a capacity, cost or dated claim attached. Hardin and Quebec, the two sites with real claim history, do not appear on the project list at all.

Capacity scaled quickly and was restated along the way. The Quebec footprint of 26,700 square feet and 1,400 units (15230, 15220) became 6,000 S-9s (15150), then a 240 PH/s target repeated three times in 2020 (14906, 14887, 14871). Within weeks that target became 294 PH/s (14858), and a 10,500-unit S-19 Pro order followed (14842). Hardin launched as 100 MW with a 30 MW Phase I (14789, 14790). By May 2021 the phases had inverted to 70 MW then 30 MW (14645, 14646, 14637, 14638), with no stated reason. A 240 PH/s total hashrate figure reappears in March 2022 (14286), which is implausible against the order book and most likely a unit error that should be checked against the filing.

Dates show quiet revision rather than acknowledged slippage. Compute North hosting for 73,000 miners moved from 2022-03-31 (14591, 14592) to 2022-06-30 (14346) without explanation. The MicroBT M31S+ lot carries both 2020-06-30 and 2020-08-15 dates (14852, 14872). An August 2020 Bitmain contract is scheduled to deliver after an October 2020 one (14340, 14341). A services agreement is effective months before it was announced (14734). The Hardin power-station upgrade deadline of 2022-12-31 (14339) has no later claim confirming completion.

The record contains no independent corroboration: no permits, interconnection queue positions or site reporting. Its heaviest hosting dependency, the Compute North joint venture (14347, 14348), ended when Compute North filed for Chapter 11 in September 2022, which is outside the record.

Over the next 90 days, the priority is closing the 2022-2026 gap in the claim history. The Q3 2026 10-Q should be checked for megawatt, energization or cost figures on any of the eleven announced sites. The Texas sites should be checked against ERCOT large-load interconnection filings, and Ohio and North Dakota against local permitting records. Until dated capacity claims exist for these projects, their announced status is an assertion without a measurable trajectory.

The record supplied here ends partway through Marathon's March 2022 10-K (claim 14304 is cut off). None of the eleven projects on file carry a capacity, cost, or date. So this read covers the 2019 to early 2022 fleet-and-hosting era, and cannot yet assess the current Texas, North Dakota, Nebraska, Ohio and Abu Dhabi portfolio. That gap is the most important finding. Also, the Granby partnership in claims 15068 and 14859 appears to be a Quebec facility, not the Granbury, Texas site on the project list, and the two should not be linked without confirmation.

On capacity, the headline figures grew mainly through equipment orders, not built megawatts. A 240 PH/s target (14906, 14887, 14871) rose to 294 PH/s within weeks (14858). It then reappears unchanged at 240 PH/s in the 2022 filing (14286), which looks like a stale restatement or an extraction error rather than a real reduction. Hardin held at 100 MW in total (14789, 14636, 14303). However, the phase split changed from 30 MW for Phase I (14790) to 70 MW for Phase I plus 30 MW for Phase II (14645, 14637, 14638), and no filing on record explains the change.

The dates show quiet drift. The 73,000-miner Compute North hosting program was promised by 2022-03-31 (14591, 14590). It appeared as 2022-06-30 in December 2021 (14346), then reverted to 2022-03-31 in the March 2022 filing (14284, 14332) without explaining the discrepancy. Miner delivery windows were extended in stages from 2021-03-31 (14762) to January 2022 (14301) and June 2022 (14302, 14523). None of these extensions is presented as a slip.

There is no independent evidence in the record, such as permits, interconnection queue positions, or site reporting, so the announced pace is unverified. The heavy dependence on a single hosting counterparty (14347, 14333) is a risk worth testing against Compute North's later insolvency.

Over the next 90 days, the priority is to attach verifiable megawatts and energization dates to the eleven listed sites. Useful sources are ERCOT large-load interconnection filings for Granbury, McCamey and Garden City, SPP and MISO queue entries for the North Dakota and Nebraska sites, and Ohio Power Siting Board or county records for Coshocton and Hannibal. Marathon's next 10-Q should be checked for whether it gives site-level energized versus contracted capacity, and whether earlier Hardin and Compute North targets are ever reconciled.

The record for Marathon is split in two, and the halves do not connect. The capacity and timeline claims describe the 2020 to 2022 fleet buildout: miner purchase orders, the Hardin, Montana site with Beowulf, and hosting with Compute North. The eleven projects now on record, from Long Ridge and Coshocton in Ohio to the Texas sites, the North Dakota and Nebraska locations, and Abu Dhabi, carry no stated capacity, cost or date. The claims list is also truncated at March 2022. Nothing here allows any of the current sites to be measured against a promise.

Within the historical record, targets moved quickly and not always consistently. The 240 PH/s fleet target (14906, 14887, 14871) became 294 PH/s within weeks (14858). The March 2022 filing still repeats 240 PH/s (14286), which is more likely stale language than a reversal. At Hardin, the 100 MW total held (14789, 14636, 14303), but the phasing did not. Phase I was 30 MW in October 2020 (14790) and 70 MW, with a 30 MW Phase II, by May 2021 (14645, 14637, 14638). None of the later claims acknowledges that change.

The dates are weaker still. The 73,000-miner Compute North hosting program was first due by March 31, 2022 (14591, 14592). A December 2021 claim moved it to June 30, 2022 (14346). Yet the March 2022 filing again cites March 31, 2022 (14284, 14304, 14332), weeks before that deadline, with no reconciliation. Miner deliveries under the December 2020 Bitmain contract ran through January 2022 (14301), well past the March 2021 date attached to 10,000 units (14762). Some claims carry target dates earlier than the claims themselves (14851, 14734), or dollar figures that look mis-scaled (14317, 14315 against 14302, 14288). They should be treated as extraction artifacts, not as commitments.

There is no independent evidence in the record for any site: no permits, no interconnection queue entries and no local reporting. The announced pace, historical or current, is therefore supported only by Marathon's own filings.

Over the next 90 days, the key test is whether the next quarterly filing puts megawatt figures or energization dates on any of the eleven announced sites. Queue entries would give the first outside check: ERCOT for the Texas sites, PJM for the Ohio sites, and the relevant operators for North Dakota and Nebraska. Any disclosure for Abu Dhabi, which has no detail at all on record, is also worth watching.

Marathon's facility record is less a single project trajectory than a sequence of hosting arrangements that were announced, re-scoped, and abandoned. The first anchor was the Compute North deal. It began as 73,000 hosted miners due by March 2022 (14591, 14592), and the deadline moved to June 2022 in a December 2021 restatement (14346). Yet the March 2022 annual filing still repeated the original March 2022 date (14304, 14284, 14332). The slip was therefore disclosed in one place and not reconciled in another. Only 30,000 of those miners were tied to the 82 percent joint-venture vehicle (14347, 14348), so the headline figure overstated what sat in a structure Marathon controlled. Hardin, Montana followed a similar arc. It was described as a 100 MW design (14303) and then as 110 MW (14094). Within a month came a plan to terminate the site agreements by August 2022 (14095) and redeploy the miners elsewhere by September 2022 (14096), so the stated capacity grew right up until the site was dropped.

The March 2023 filing moves to a distributed hosted footprint, and nearly every dated commitment in it carries a delayed flag. Ellendale's 57,000-unit energization (12982, 13095), Jamestown's 10,400 air-cooled units for Q1 and 768 immersion units for Q2 (13051, 13052), Kearney's 1,300 units (13073), and the Coshocton exit (13053) all fall into that group. Abu Dhabi is internally inconsistent within a single filing. One set of claims gives July 2023 for operations (13065, 12978) and another gives December 2023 (13058, 13076). The capital commitment appears as both 81.2 million (12977) and 406 million (13056), and the 250 MW figure (13077) has no dated milestone attached.

No permits, interconnection queue entries, or site reporting sit in this record to corroborate any of it. The eleven listed projects carry no capacity or cost fields. Long Ridge, Matagorda County, and Granbury have no claims at all. The claim history also stops in March 2023, roughly three and a half years ago. Any read of current pace therefore rests on company statements alone and is stale.

Over the next 90 days, the third-quarter 10-Q is the first thing to watch. It should show whether Abu Dhabi, Ellendale, and Jamestown are energized at their stated scale, or whether the counts have quietly changed. Hosting contracts at Plano, Ellendale, Garden City, McCamey, and Jamestown expire between June and December 2027 (13102, 13096, 13093, 13090, 13100), so renewal or exit language should start appearing. Filings should also finally put capacity, cost, and interconnection substance behind Granbury, Matagorda, and Long Ridge.

Marathon's record is best read as a company that repeatedly outsourced its capacity and then had to take it back. The early promises were counted in machines rather than megawatts. The company claimed 73,000 S19 miners hosted with Compute North by March 2022 (14284, 14304, 14332), with the same figure elsewhere pinned to June 2022 (14346). Hardin, Montana grew from a 100 MW design (14303) to 110 MW (14094) within weeks, and in the same filing cycle the company set a date to terminate the Hardin agreements (14095) and move those miners elsewhere by September 2022 (14096). That made the upgrade dates for December 2022 (14285, 14339) moot. By March 2023 the vocabulary had shifted to named sites and megawatts, headlined by a 250 MW Abu Dhabi joint venture (12952, 12973, 13064). The record does not show the step down from 73,000 hosted units, but it is implied by the pivot, and Compute North's publicly reported 2022 bankruptcy would explain it.

On timing, slippage is the norm and is increasingly acknowledged. Ellendale energization (12960, 12982), Jamestown's 10,400 air-cooled and 768 immersion units (12961, 12962, 13051, 13052), Kearney's 1,300-unit expansion (13073), and the Coshocton exit (13053) all carry delayed status. Abu Dhabi is the least coherent case. Its start of operations is dated July 2023 in some claims (12978, 13065) and December 2023 in another (13058), all from the same filing. The capital figures do reconcile: an $81.2 million commitment (12977) is 20 percent (13063) of $406 million (13056). The miner purchase contracts carrying 2021 and early-2022 delivery dates (14340, 14341, 14342, 14301) have no delivery confirmations on record.

The claim set contains no independent corroboration. There are no permits, no interconnection queue entries in ERCOT, SPP, MISO or PJM, and no site reporting. Every capacity figure is self-reported, and several are duplicated or carry unit errors (14311, 14315, 14317). All eleven projects on record show no stated capacity or cost. Long Ridge, Matagorda County, Granbury and Plano have no claims in the visible record. The excerpt ends in March 2023, so the more recent pivot toward owned sites and AI-adjacent power is not assessable here.

Over the next 90 days, the Q3 2026 10-Q should show whether Abu Dhabi, Ellendale and Jamestown are reported in megawatts actually energized rather than units deployed. Watch for ERCOT large-load filings tied to Matagorda County and Granbury, and PJM or Ohio siting records for Long Ridge. The key test is whether any of the eleven announced projects gains a disclosed capacity figure that outside records can confirm.

Marathon's facility record is less a trajectory than a single snapshot repeated many times. Apart from the April 2022 Hardin disclosures, nearly every capacity and timeline claim traces to one filing dated 2023-03-15. No capacity figure has been revised upward or downward since then because nothing later has been recorded. The Abu Dhabi figure sits at 250 MW across every restatement (12952, 13064, 13077, 13086). The US hosted-site targets are equally fixed: 57,000 units at Ellendale (13094), 10,400 air-cooled and 768 immersion units at Jamestown (12968, 12969), 4,000 at McCamey (12965), 19 MW at Garden City (12966), 1,300 at Kearney (12963), and a 200,000-unit fleet goal (12956). The mix of MW and unit counts makes an aggregate power figure impossible to derive from the record. Every project on file still carries announced status with no capacity or cost populated.

The dates are where the record is weakest. Within the same filing, Ellendale energization appears as both 2023-03-31 (12982, 13070, 13095) and 2023-06-30 (12959, 13050, 13080). Abu Dhabi operations appear as both 2023-07-01 (12978, 13065) and 2023-12-31 (13058, 13076). Coshocton's exit is given as 2023-06-01 (13101) and 2023-06-30 (13053). The Abu Dhabi JV formation date of 2023-01-27 (13063) had already passed when it was disclosed. The capital figures also conflict: 81.2 million (12977, 13074) against 406,000 thousand (13056), which is more likely total JV scale or an extraction unit error than a real revision. Nearly all dated 2023 milestones are flagged delayed, and none has a recorded completion or acknowledged re-baseline. Hardin is similar: a 110 MW site (14094) was slated for agreement termination by 2022-08-15 (14095) and redeployment by 2022-09-30 (14096), and neither has a closing claim.

The record contains no independent corroboration for any site: no permits, interconnection queue positions, or local reporting. The pace therefore rests entirely on company statements. Long Ridge, Matagorda County and Granbury carry no visible claims, and the supplied history cuts off at 12691.

Over the next 90 days, the Jamestown immersion contract expiry of 2026-08-01 (13099) has already passed, so the Q3 10-Q due in early November should show whether it was renewed or exited. The 2027 hosting expirations then cluster: Plano (13102), Ellendale (13096), Garden City (13093), McCamey (13090) and Jamestown air-cooled (13100). Early renewal or migration disclosures would be the first real evidence of how the hosted footprint is evolving. Watch too for any filing that finally closes out Hardin and Abu Dhabi.

The first thing to say about Marathon's facility record is that it is not a trajectory. Every capacity and timeline claim on file dates to a single disclosure on 2023-03-15, so there is no observable scaling or shrinkage over time. What looks like a large body of evidence is mostly the same statements extracted repeatedly. The Ellendale 57,000-unit figure alone appears in 13080, 12959, 13070, 13094, 12967 and 13050. The fleet-wide 200,000-unit energization target (12956) has no later restatement to compare against. Three listed projects, Long Ridge, Matagorda County and Granbury, have no capacity or timeline claims at all. They should be treated as names on a list rather than commitments.

Within that one filing, the dates are internally layered rather than contradictory in most cases. Ellendale carries a Q1 2023 energization start (12960, 13095, 12982) and a Q2 2023 completion (13080, 12959, 13050). Abu Dhabi pairs a 2023-07-01 commencement (13065, 12978) with a year-end 2023 operational date (13058, 13076). The Abu Dhabi dollar figures reconcile: the $81.2 million commitment (13074, 12977) is 20 percent of the $406 million venture (13056, 13063). The one genuine inconsistency is Coshocton, where the exit date is given as 2023-06-30 in 13053 and 2023-06-01 in 13101. Every dated milestone is now tagged delayed. That tag reflects the absence of any follow-up claim confirming delivery. It is not affirmative evidence that the milestones slipped, and Marathon has not issued an acknowledged revision on record.

No independent evidence is attached to any of these sites: no permits, interconnection queue positions or on-the-ground reporting. The announced pace, 250 MW in Abu Dhabi (12952) and tens of thousands of rigs deployed within one or two quarters, is therefore unverified in either direction.

Over the next 90 days, the most concrete item is Jamestown. Its immersion hosting contract was due to expire on 2026-08-01 (13099), and the next quarterly filing, expected in early November, should show whether it was renewed, extended or wound down. The same filing is the natural place to look for renewal language ahead of the mid-2027 hosting expirations at Plano (13102), Ellendale (13096), Garden City (13093) and McCamey (13090). It should also show whether the Abu Dhabi venture ever reached its stated 250 MW, and whether Long Ridge, Matagorda or Granbury acquire any dated capacity commitments. Ingesting Marathon's post-2023 filings is the precondition for any real trajectory read.

The first thing to say about Marathon's record is that it is not really a trajectory. Nearly every facility claim on file traces to a single disclosure dated 2023-03-15, extracted many times over. The Abu Dhabi 250 MW figure alone appears in at least six entries (12952, 12973, 13064, 13077, 13086, 13058), and Ellendale's 57,000 units, Jamestown's 10,400 air-cooled and 768 immersion units, McCamey's 4,000 rigs and Garden City's 19 MW each repeat similarly. Capacity has therefore neither scaled nor shrunk on the record. It has been stated once and never revisited. The 200,000-unit full-fleet energization target (12956) is the only aggregate figure, and nothing later tests it. Long Ridge, Matagorda County and Granbury are listed as announced projects, but the visible claims attach no capacity, cost or date to them. The claim set also appears truncated after the April 2023 Auradine entries, so later disclosures may simply not have been captured.

The delayed labels need careful reading. Most attach to near-term 2023 targets with no follow-up on file: Ellendale energization by Q1 2023 (12960, 12982), Jamestown air-cooled deployment by Q1 2023 (13051, 13097), Abu Dhabi operations by mid- or end-2023 (12978, 13076) and the Coshocton exit by mid-2023 (13053, 13101). That pattern reflects missing confirmation, not documented slippage, and nothing shows a date quietly moved. Some labels are extraction artifacts. The JV formation date of 2023-01-27 (13063) and the Auradine investment dated 2022-09-23 (12686) both precede the filing and describe completed events, not promises. Internal inconsistencies matter more: Ellendale carries both Q1 (13070) and Q2 (12959, 13080) targets for the same 57,000 units, and the Coshocton exit is dated both June 1 (13101) and June 30 (13053).

No permits, interconnection queue entries or site reporting are on record, so the announced pace cannot be independently corroborated for any site. That gap is the central finding.

Over the next 90 days, Marathon's Q3 2026 10-Q, expected in early November, is the natural checkpoint for the three uncharacterized sites. ERCOT queue filings for Matagorda County and Granbury and PJM activity tied to Long Ridge would be the first independent tests. The Jamestown immersion hosting contract was due to expire in August 2026 (13099), so its renewal or exit status should be confirmed now. The mid-2027 expirations at Plano, Garden City, Ellendale and McCamey (13102, 13093, 13096, 13090) make any renewal disclosures in that filing worth tracking.

Marathon's record here is less a trajectory than a single snapshot. Every facility, capacity and timeline claim was captured on 2023-03-15, apart from the Auradine investment disclosures of 2023-04-27 (12686, 12691). Nothing on file shows capacity scaling or shrinking over time. What the record does show is a hosted-fleet buildout: Ellendale at 57,000 units (12959), Jamestown at 10,400 air-cooled plus 768 immersion units (12961, 12962), McCamey at 4,000 (12965), Kearney at 1,300 (12963) and Garden City at 19 MW (12966). These roll up toward a 200,000-unit fleet energization target (12956), with a 250 MW Abu Dhabi joint venture layered on top (12952). The Abu Dhabi capital figures reconcile rather than conflict. A 20 percent stake (12955, 13063) in a roughly $406 million venture (13056) matches the $81.2 million commitment (12977); 13074 appears to drop the thousands multiplier.

The dates are internally inconsistent within the same filing, and that matters more than any slip. Ellendale energization is pinned to 2023-03-31 in 13095, 12960 and 12982, but to 2023-06-30 in 12959, 13050 and 13080. Abu Dhabi commencement appears as 2023-07-01 in 13065 and 12978, and as year-end 2023 in 13076 and 13058. Some of this is probably extraction noise across overlapping passages. Still, it means no single committed date can be scored as held or missed without a follow-up disclosure, and none is on record. Every project, including the Coshocton exit promised by mid-2023 (13101, 13053), still sits at status announced, three years on.

No independent evidence is attached. There are no permits, no interconnection queue entries and no site reporting, so the pace described in 2023 is unverified here. Granbury, Matagorda County and Long Ridge carry no claims at all. That gap likely reflects later owned-site strategy the pipeline has not yet ingested, and it should not be read as inactivity.

Over the next 90 days, the priority is the Jamestown immersion hosting contract, which was due to expire 2026-08-01 (13099) and has now passed. A renewal, migration or quiet lapse would be the first test of the hosted model's durability. The other hosting terms cluster between mid-2027 and end-2027 (13102, 13093, 13096, 13090, 13100), so the next quarterly filing should signal renewal intent. Watch too for whether current disclosures restate Abu Dhabi's capacity and ownership, and whether status updates for Granbury, Matagorda and Long Ridge arrive with verifiable interconnection or permitting detail.

Marathon's facility record is lopsided. Nearly all specific capacity and date commitments come from one filing on 2023-03-15, and the three projects now on the books, Long Ridge in Ohio, Matagorda County and Granbury in Texas, carry no disclosed capacity or cost. Capacity was never framed as a single figure. It grew through a stack of hosted-site unit counts toward a 200,000-unit fleet target (12956), and the Abu Dhabi joint venture was pegged at 250 MW (12952, 13086). The later record moves from counting hosted rigs to buying the sites themselves: the GC Data Center purchase (6401, 6379), the Hansford County wind farm (5116, 9274), and a 53.2 EH/s energized hash rate (5022, truncated in this record). That is a real shift toward owning infrastructure, but the dataset holds no updated per-site figures to show whether the 2023 targets were met or quietly dropped.

The dates are internally inconsistent even within that single filing, which weakens any test of whether they held. Ellendale energization appears as 2023-03-31 in some extractions (12960, 13070, 13095) and as a deployment completion of 2023-06-30 in others (12959, 13050). Abu Dhabi operations were due by 2023-07-01 (12978, 13065) and also by 2023-12-31 (13058, 13076). The JV capital figure ranges from 81.2 million (12977, 13074) to 406 million (13056). The larger number is more likely total project cost or a unit error than a revision, but it deserves a check against the source text. The GC Data Center price rose from 178.65 million (6401) to 183.8 million (6206) with no stated reason, plausibly ordinary closing adjustments. No later claim acknowledges a slip on any 2023 date, and none confirms delivery either. That silence is the main finding.

The record contains no independent evidence at all: no permits, no ERCOT or PJM queue entries, no local reporting. The pace of deployment cannot be corroborated from what is here, and the three named projects are effectively placeholders.

Over the next 90 days, the Q3 2026 10-Q, due around early November, is the main test. It should show whether Long Ridge, Matagorda and Granbury carry megawatt or cost figures. Useful external checks are ERCOT large-load interconnection filings for Matagorda, Ohio Power Siting Board or PUCO dockets for Long Ridge, and Hood County permitting and noise disputes at Granbury. Any restatement of energized hash rate against the 53.2 EH/s baseline also matters. Separately, the duplicated and conflicting 2023-03-15 extractions should be reconciled before these claims are scored.

Marathon's stated footprint has grown steadily, though mostly by acquiring and contracting power rather than by commissioning new capacity. The energy portfolio went from 1.7 GW in April 2025 (5021) to nearly 2 GW in February 2026 (11788), then to a 1.9 GW global data center footprint, 70 percent owned (3490, 3491). Hashrate rose from 53.2 EH/s (5022) to 66.4 EH/s (3508). The site-level figures are harder to square with that headline. Granbury at 300 MW (3499), Central Texas at 250 MW (3532), East Ohio at 225 MW (3533), Central Ohio at 150 MW (3535), North Texas wind at 180 MW (3534) and the Nebraska acquisition at 42 MW (3514) add up to roughly 1.15 GW. That leaves about 750 MW not broken out in this record. The Hansford County wind farm also appears three ways: 114 MW nameplate (3505), 240 MW interconnection (3504) and 180 MW (3534).

The forward AI/HPC story rests almost entirely on the Starwood arrangement. Its size is quoted as 1 GW near-term (11781), a 2.5 GW pathway (11782, 926) and a 1 to 2.5 GW range (3512). Marathon's economic interest is described as both 50 percent (3494, 3528) and 10 percent (3662), a gap that needs a filing-level answer. The only timeline is a 24-month development decision window (3664), which is not a delivery date. MPLX is framed as an integrated campus partnership (11789) but documented only as a letter of intent (3513, 3524).

Dates have not slipped because almost none were given. The dated items are financial: the hedge running to December 2027 (6208) and the third Exaion tranche due March 2027 to reach 75 percent (4521, 4522). One transaction value was revised without comment, the GC Data Center purchase moving from $178.65M (6401) to $183.8M (6206), most likely closing adjustments. The Long Ridge acquisition at $1.5B (9088) is the largest commitment, but the claim record for it is truncated (9413). All three tracked projects, including Matagorda County with no supporting claims, still carry no capacity or cost figures.

This record contains no independent evidence, such as permits, ERCOT or PJM queue entries, or local reporting, so the announced pace is unverified. Over the next 90 days, watch the Q3 2026 results for a site-level reconciliation of the 1.9 GW figure. Also watch for the first property Starwood designates for development and a definitive JV ownership percentage, any conversion of the MPLX letter into a binding agreement, and the Long Ridge closing status. Interconnection queue filings for Hannibal and Matagorda County would be the clearest early signal of real build activity.

Marathon's stated capacity has grown steadily, but the unit being measured has changed along the way. The April 2025 figure was a 1.7 GW energy portfolio (claim 5021). By February 2026 it was nearly 2 GW of secured power (11788), and a week later a 1.9 GW global data center footprint (3490, 3507, 3520). These are not the same metric, so the apparent growth partly reflects relabeling rather than new megawatts. Hashrate grew from 53.2 to 66.4 EH/s (5022, 3508), which is a cleaner operational signal. The regional figures (Granbury 300 MW in 3499, Central Texas 250 MW in 3532, North Texas wind 180 MW in 3534, East Ohio 225 MW in 3533, Central Ohio 150 MW in 3535, Nebraska 42 MW in 3506) add up to roughly 1.15 GW. That leaves much of the 1.9 GW headline unallocated in the record. The wind asset is also described three ways: 180 MW, 114 MW nameplate and 240 MW interconnection (3505, 3504). The Starwood figures deserve the most caution. A 1 GW near-term target and a 2.5 GW pathway (11781, 11782) sit beside conflicting ownership terms, 10 percent in one disclosure (3662) and 50 percent in several others (3494, 3528). A 24-month development decision window (3664) also means none of that capacity is firmly committed before early 2028.

Dated promises are scarce, which is itself the main finding on timelines. Apart from the Exaion third tranche set for March 30, 2027 (4521) and a hedge running through 2027 (6208), almost nothing carries a delivery date. Long Ridge readiness (9411) has none. Deals have mostly closed, though prices drifted: GC Data Center moved from $178.65M to $183.8M (6401, 6206), and Exaion closed about six months after signing (2945). The 2023 Abu Dhabi venture (12646) is never mentioned again. The MPLX arrangement remains a letter of intent (3513).

Every claim in the record is company-stated. Nothing here confirms the announced pace through permits, grid interconnection queues or site reporting. Long Ridge is still listed as announced with no capacity or cost, despite the $1.5B acquisition (9088) and a 505 MW nameplate increase (9413). Matagorda County has no supporting claims at all. The supplied record also appears truncated after claim 9411.

Over the next 90 days, watch whether third-quarter results give Long Ridge a service date and filings that back the 505 MW authorization. Also watch for the Starwood ownership split to be reconciled, the first Starwood site to be designated, and the MPLX letter of intent to become a definitive agreement. Finally, check whether Granbury or Matagorda show up in ERCOT interconnection or county permitting records.

Marathon Digital's infrastructure trajectory from late 2023 through mid-2026 reveals capacity ambition outpacing credible delivery evidence. The GC Data Center Equity Holdings acquisition, initially stated at $178.65 million in December (6402, 6401), was restated at $183.8 million in May (6206) without explanation for the $5.15 million delta or acknowledgment of the missed January 2024 closure target (6379). This restatement pattern—claims revised without transparent reasoning—recurs throughout the portfolio and signals either planning flux or disclosure opacity.

Capacity announcements escalated significantly in early 2026. Energy portfolio grew from 1.7 GW in April 2025 (5021) to 1.9 GW by March 2026 (3490), and mining hashrate increased from 53.2 EH/s to 66.4 EH/s (3522). Yet the headline shift came with the February 2026 Starwood partnership: rather than organic development, Marathon structured joint ventures for 1.0 GW near-term (11781) and 2.5 GW pathway capacity (11782), retaining 50 percent ownership (3494). Simultaneously, Granbury emerged as a flagship 300 MW facility (3499), anchoring a distributed platform with East Ohio (225 MW), North Texas wind (180 MW), and Central Ohio (150 MW) (3533, 3534, 3535). These figures convey portfolio maturity; grid interconnection filings or construction permits do not appear in the record.

The April 2026 Long Ridge acquisition represents the starkest inflection. A $1.5 billion commitment for a 485 MW gas plant (9088, 1869) suggested transformational capital deployment. Within four weeks, a $75 million termination fee surfaced (1872), implying deal deterioration or restructuring wholly absent from the claims narrative. This gap signals material execution risk.

All capacity and timeline claims remain marked 'stated' without third-party validation, grid queue entries, or permitting evidence. Over the next 90 days, track Long Ridge resolution, Starwood's promised IT capacity delivery with geographic and timeline specificity, and whether Granbury construction initiates. Capacity 'pathways' lack credibility without interconnection agreements or binding development schedules.

Marathon Digital's announced capacity has expanded substantially across its portfolio since late 2024. The company moved from stating 1.7 GW of energy portfolio capacity in April 2025 to claiming 2.2 GW of operational and development capacity by April 2026, a 29 percent increase in baseline holdings. This growth trajectory accelerated further with the February 2026 Starwood Capital partnership, which layered in an additional 1.0 to 2.5 GW of AI and HPC infrastructure through a 50-50 joint venture structure, though Marathon retains only a 10 percent interest in the pre-development phase (claim 3662). The company has also granularized capacity claims significantly: Granbury, Texas shows 300 MW (claim 3509), Long Ridge in Hannibal, Ohio claims a 600 MW expansion path atop 200 MW current operations (claim 9407, 9401), and Nebraska adds 42 MW. Hashrate metrics similarly climbed from 53.2 EH/s in April 2025 to 66.4 EH/s by March 2026 (claims 5022, 3522).

On timing, most recent announcements lack explicit service-readiness dates. The Long Ridge acquisition, valued at $1.5 billion in April 2026 (claim 9088), announced an expected closing without specifying when (claim 9412). The Starwood partnership, signed in February 2026, imposed a 24-month development-decision timeline (claim 3664), pointing toward February 2028 as a key inflection. Earlier commitments show mixed execution: the Hansford County wind farm acquisition moved from announcement in February 2025 through stated closing by mid-February (claim 5116, 9274), but numerous bitcoin acquisition announcements from late 2024 (claims 5426, 5350, 5346) lack subsequent confirmation. The Exaion majority-stake acquisition, announced in August 2025 and again in February 2026, includes a conditional March 30, 2027 secondary transaction (claim 4521), creating a staged timeline that has no clear public milestone through late summer 2026.

No independent evidence—permits, FERC interconnection queue position, local land records, utility interconnection studies—has been provided to validate the announced pace. Without corroborating documentation from Texas Railroad Commission, Public Utility Commission of Texas, or regional grid operators, the gap between claimed capacity and actual grid-integrated delivery remains opaque.

Over the next 90 days, watch for: (1) Long Ridge acquisition closing and any revised timeline or capacity restatement; (2) public filings detailing Granbury's power and cooling infrastructure buildout; (3) Starwood-MARA joint-venture project announcements or preliminary engineering reports; (4) any restatement of the March 2027 Exaion option exercise.

Marathon Digital's announced capacity trajectory shows significant growth punctuated by consolidating announcements that warrant scrutiny. The company's energy portfolio expanded from 1.7 GW (claim 5021, April 2025) to 1.9 GW (claim 3490, March 2026), then jumped to 2.2 GW operational and development capacity following the Long Ridge acquisition announcement (claim 9406, April 2026). This represents roughly 29% nominal growth over twelve months, though the composition shifted markedly toward infrastructure partnerships rather than owned assets. The owned-and-operated percentage dropped to 65-70% (claims 3491, 9405), meaning Marathon increasingly relies on third-party execution.

Promised timelines show early cracks. The Long Ridge nameplate capacity expansion of 505 MW carried a July 1, 2026 deadline (claim 1870) but is flagged as delayed, suggesting the company missed its initial integration window by at least two months. The Starwood Digital Ventures partnership (claims 3493, 3511, announced February 2026) targets 1.0 to 2.5 GW of AI/HPC capacity with a 24-month development window (claim 3664), placing critical milestones in early 2028—a timeline dependent on EDF Pulse Ventures' (claim 9271) continued cooperation and capital deployment.

The March and April 2026 announcement clusters reveal a shift toward portfolio expansion through acquisition and partnership rather than organic build. Exaion's closing structure (claims 4520, 4521, 4522) involves three tranches concluding in March 2027, the Hansford County wind farm (claims 3504, 3523) brought 114 MW nameplate but 240 MW of interconnection, and the Nebraska data center added only 42 MW (claims 3506, 3514). None of these claims reference permit status, queue positions, or equipment procurement milestones, limiting visibility into actual execution risk.

The immediate watch list centers on Long Ridge remediation efforts and Starwood's pre-development spending rate. The Long Ridge facility represents the largest single asset acquisition at $1.5 billion (claim 9088) and its delay suggests either operational complexity or integration challenges that could propagate to other assets. Exaion's March 2027 closing and the Starwood capex cycle will determine whether 2026's announced capacity translates to operational megawatts or remains subscribed but undeployed.

Marathon Digital's announced capacity trajectory shows significant growth but masks complexity in asset composition and execution timing. The company progressed from 1.7 GW stated energy capacity in April 2025 (5021) to 1.9 GW globally by March 2026 (3490), then layered the Long Ridge acquisition in April 2026, bringing stated operational and development capacity to 2.2 GW (9406). Bitcoin hashrate increased more dramatically over the same period, from 53.2 EH/s (5022) to 66.4 EH/s (3508), suggesting hardware efficiency gains outpaced power additions. The portfolio now includes owned assets (Hannibal 200 MW per 9407, Granbury 300 MW per 3499), Exaion's European footprint (controlling stake acquired February 2026 per 2945), and a 50-50 Starwood Digital Ventures joint venture (3528) with an ambitious 1-2.5 GW target (3512) but only 1.0 GW near-term capacity commitments, suggesting pre-final investment decision status. Long Ridge's integration reveals a red flag: the 505 MW CCGT nameplate capacity expansion was promised by July 1, 2026 (1870), but is marked delayed as of the April 2026 filing. With today's date at September 11, 2026, this expansion is now overdue by more than two months. Regional breakdowns in the March 2026 filings—Granbury 300 MW, Central Texas non-Granbury 250 MW, East Ohio 225 MW, Central Ohio 150 MW, North Texas wind 180 MW (3531-3535)—total approximately 1.1 GW in stated capacity but lack cross-reference to Texas Public Utility Commission interconnection filings or Ohio transmission queue data. The Hansford County wind acquisition claimed 240 MW interconnection capacity but only 114 MW nameplate (3504-3505), a distinction that matters for actual dispatch. The Starwood partnership announced 24 months for development decision (3664); at six months elapsed since announcement, no ground-breaking announcements have emerged. In the next ninety days, watch for Long Ridge expansion closure or revised timeline, Granbury facility operational metrics (MW served versus MW available), Exaion European regulatory approvals for any announced expansions, and Starwood pathway clarity toward the 1.0 GW or higher capacity threshold. Capacity growth claims have been substantial but increasingly depend on non-controlled assets or partnerships at earlier stages than press releases imply.

Marathon Digital's capacity profile has undergone rapid expansion and strategic reorientation since early 2025. Through April 2026, announced total capacity grew from 1.7 GW (claim 5021, April 2025) to 1.9 GW (claims 3490, 3507, 3520, March 2026) and subsequently to 2.2 GW when including planned development capacity (claim 9406, April 2026). However, this growth reflects acquisitions—Exaion majority stake (claims 3492, 3500, 3526, March 2026), Hansford County wind at 114 MW nameplate with 240 MW interconnection (claims 3504–3505, March 2026), and the 485 MW Long Ridge CCGT plant (claim 1869, April 2026)—rather than organic buildout. The company retained only 70 percent owned-and-operated portfolio (claims 3491, 3521, March 2026), with the Starwood joint venture representing 50 percent partnership equity targeting 1.0–2.5 GW AI/HPC capacity over 24 months from February 2026 (claims 926, 3664, February–March 2026).

Timeline adherence shows a mixed record. The GC Data Center acquisition closure was promised for January 2024 (claim 6379) but the May 2024 acquisition claim (6206) references completion retrospectively. The Long Ridge Energy acquisition, announced in April 2026 with a $1.5 billion price tag (claims 9088, 1868), listed no explicit closing date, though claim 9412 references an "expected closing" without specificity. Starwood's 24-month development mandate from February 2026 implies initial deployment by February 2028, but no intermediate milestones have been announced.

Independent verification remains limited. The Hansford wind farm shows granular detail—240 MW interconnection versus 114 MW nameplate (claims 3504–3505)—suggesting actual grid queue placement, though no permits or transmission studies are cited. Long Ridge's proposed 600 MW expansion beyond the current 485 MW nameplate (claim 9401) shows no environmental review data. The Nebraska acquisition (42 MW, claim 3506) and Granbury facility (300 MW, claims 3509, 3531) cite no third-party validation. Energized metrics rely entirely on company disclosure.

Over the next 90 days, watch for Long Ridge acquisition closing confirmation and initial gas plant operational status, Starwood joint venture's first AI facility deployment and location announcement, Exaion integration milestones including any capacity revisions, and any grid interconnection or permitting updates on the Long Ridge 600 MW expansion. The portfolio's growth trajectory depends heavily on partnership and acquisition assets energizing; without permitting or grid queue evidence, announced capacity targets remain unvalidated.

Marathon Digital's capacity narrative has undergone substantial revision across a rapidly expanding footprint. The company initially emphasized energy portfolio aggregation, announcing 1.7 GW in portfolio capacity by April 2025 (claim 5021), then layered on global data center operations at 1.9 GW by March 2026 (claim 3520). The April 2026 Long Ridge acquisition announcement introduced a new metric entirely—2200 MW of operational and development capacity (claim 9406)—which conflates existing infrastructure with speculative expansion potential. The Long Ridge facility itself exemplifies this ambiguity: the company cites a 1000 MW total potential (claim 9400) against a 485 MW existing CCGT nameplate (claim 1869) and 505 MW authorized capacity expansion (claim 9413), muddying what portion is energized versus planned.

Timeline discipline has been inconsistent. The Hansford County wind acquisition was announced in February 2025 with an immediate closing promised for February 2026 (claims 5116, 9274); no evidence of realized generation or grid connection has been disclosed. The Exaion transaction, split across primary (August 2025), secondary, and conditional tertiary tranches (March 2027, claim 4521), has experienced compression—the controlling stake was declared achieved by February 2026 (claim 2945) rather than trailing the full closing sequence. The Starwood Digital Ventures partnership, announced February 2026 with a 24-month development window (claim 3664), targets 1.0 GW of AI/HPC capacity (claim 3529) without disclosed buildout schedules or power procurement guarantees.

Independent corroboration is sparse. Long Ridge's AI/IT construction timeline and service readiness dates (claims 9410, 9411) remain unspecified in the submitted claims. No interconnection queue positions, PUC filings, or third-party engineering studies are referenced to substantiate the compression from potential to operational capacity. The shift from hash-rate metrics (53.2 EH/s in April 2025, claim 5022, to 66.4 EH/s by March 2026, claim 3522) suggests operational gains, but facility-by-facility validation is absent.

Over the next 90 days, monitor Long Ridge's formal AI/IT buildout commencement and any disclosure of power purchase agreements or MPLX LP execution (claims 9410, 3513). Watch for Hansford wind farm grid interconnection notices and Starwood development commitment timelines. Clarification on owned-versus-managed capacity mix (65 percent claimed as owned, claim 9405) and reconciliation of the 2200 MW figure against actual energized megawatts will signal whether capacity growth tracks announced projections.

Marathon Digital's announced capacity trajectory shows material expansion but unclear execution timing. In April 2025, the company stated an energy portfolio capacity of 1.7 GW and energized hash rate of 53.2 EH/s (5021, 5022). By March 2026, total global data center capacity had grown to 1.9 GW with hash rate reaching 66.4 EH/s (3507, 3522), a 12% capacity increase and 25% hash rate gain. The April 2026 Long Ridge acquisition announcement then jumped operational and development capacity to 2.2 GW (9406), representing a 15% increase within a single month—contingent on a deal whose termination fee was cited just days later (1872), signaling early execution risk.

Capacity announcements lack clarity on ownership stakes. The February 2026 Starwood Capital partnership promised 2.5 GW in joint AI/HPC capacity (926), but Marathon retained only 10% initial ownership (3662) and 50% post-closing stake in the joint ventures (3528). The same announcement placed Granbury at 300 MW capacity (3509), yet Granbury appeared in the projects list as "announced" with no prior capacity claim, suggesting either retroactive capacity assignment or timeline compression. Exaion's August 2025 acquisition and February 2026 controlling stake (2945) added capacity but no specific MW figures were disclosed for its contribution.

Promised dates have been sparse. The sole construction-phase timeline came February 2026: a 24-month development decision window (3664), placing major decisions at February 2028. Long Ridge buildout commencement and initial capacity service readiness were announced April 2026 (9410, 9411) but without target dates. The Long Ridge acquisition itself listed "expected closing" (9412) without specificity, while the termination fee mention (1872) within days of the announcement raises questions about deal conviction.

Independent verification is limited by available data. Hansford County wind farm interconnection capacity was stated at 240 MW (3504, 3516) versus 114 MW nameplate (3505, 3523), a 2.1x spread typical of future expansion rights rather than current operational capacity. Without access to ERCOT or SPP interconnection queues, FERC filings, or construction permits for Granbury and Long Ridge, the pace claim cannot be independently validated.

In the next 90 days, watch for Long Ridge acquisition closure confirmation, any Granbury capacity energization announcements, and updates to the 2.2 GW operational target as Starwood and Exaion contributions clarify.

Marathon Digital's capacity narrative shows substantial expansion ambitions paired with an emerging timeline slip. The company claimed a 1.9 GW global data center footprint in March 2026 (claims 3490, 3507, 3520), growing from the 1.7 GW energy portfolio capacity stated in February 2025 (claim 5021). The portfolio rebalancing reflects a shift from primarily bitcoin mining toward a mixed AI/HPC and mining platform: the March portfolio includes 70 percent owned-and-operated assets (claim 3491), with hashrate holding at 66.4 EH/s (claim 3522) despite the capacity reorientation.

The April 2026 Long Ridge acquisition for 1.5 billion USD (claim 9088) promised a material step function: a 505 MW nameplate capacity expansion scheduled for July 1, 2026 (claim 1870). That deadline passed sixty-plus days ago, and the claim is flagged as delayed—the company's first explicit delivery miss. Total claimed operational and development capacity reaches 2200 MW (claim 9406) with owned-operated penetration expanding to 65 percent (claim 9405), but without clear deployment timelines the feasibility remains opaque.

The Starwood Capital partnership launched February 26, 2026 (claim 7504) targets 2.5 GW AI/HPC capacity through a 50-50 joint venture with 1.0 GW initial target (claims 3529, 3528) and a 24-month development cycle (claim 3664)—placing material milestones at February 2028. Exaion, acquired for controlling stake in February 2026 (claim 2945), gates a third tranche of investment to March 30, 2027 (claim 4521). Neither partnership carries intermediate date markers against which to measure progress through year-end.

No claims reference specific interconnection queue positions, building permits, or independent site validation. All announcements carry [stated] classification without external corroboration. The Long Ridge expansion delay is the first tangible signal that announced timelines may compress under execution scrutiny.

Watch the next 90 days for: revised Long Ridge deployment guidance, monthly or quarterly Starwood partnership milestones, any restatement of the 2200 MW figure, and whether interconnection or permitting data enters the narrative. The distance between announced capacity and actual permitting pace will clarify execution credibility.

Marathon Digital's expansion narrative rests substantially on acquisitions rather than organic capacity build. Through 2025, the company claimed 1.7 GW in energy capacity and 53.2 EH/s mining hashrate (claims 5021, 5022); by April 2026, following the Long Ridge purchase and Exaion majority stake acquisition, it cited 2.2 GW operational and development capacity (claim 9406). The stated hashrate grew to 66.4 EH/s, driven by the Long Ridge 485 MW gas plant and partnership-layer capacity rather than company-owned data centers. Owned-and-operated capacity is positioned at 70 percent of the portfolio (claim 3521), suggesting roughly 1.3-1.4 GW is retained, with the balance in joint ventures or minority positions.

The company's timeline discipline shows strain. A Long Ridge nameplate capacity expansion of 505 MW was promised by July 1, 2026 (claim 1870, flagged delayed)—now past—without public acknowledgment of delay or revised delivery. Most other major commitments lack hard dates: the Starwood partnership, announced in February 2026 targeting 1-2.5 GW AI/HPC capacity (claims 7504, 926), carries no development timeline, only a 24-month decision window (claim 3664). The Exaion third transaction, scheduled for March 30, 2027 (claims 4521, 4522), remains the only major near-term deadline flagged.

Independent verification remains sparse in the record. Granbury consistently claims 300 MW capacity across multiple assertions (claims 3509, 3531, 3499); the Hansford County wind farm lists 240 MW interconnection capacity against 114 MW nameplate (claims 3504, 3505, 3516), a mismatch suggesting reserved but not yet energized transmission. No interconnection queue positions, executed interconnect agreements, or recent permitting for Long Ridge's expansion were cited to validate the July target or current status. The MPLX letter of intent from March (claims 3513, 3524) remains unsigned.

Over the next 90 days, attention should fix on three points: whether Long Ridge expansion is complete and energized or further delayed; whether Starwood's pre-development services have translatable capacity momentum; and what the company reports on the operational status of the 2.2 GW claimed in April. The distinction between operational and development capacity requires clarification—much of the announced base rests on announced, not energized, infrastructure.

Marathon Digital has pursued a capacity-expansion strategy heavily dependent on acquisitions and partnership structures rather than organic facility development, with announced capacity growing from 1.7 gigawatts in April 2025 to 2.2 gigawatts by April 2026. This 29 percent expansion over twelve months primarily reflects portfolio consolidations: the Exaion majority stake, the Long Ridge Energy acquisition for 1.5 billion dollars, and earlier wind and data center purchases (claims 5021, 9406, 3514, 3515, 3525, 3500, 9088). The company's three originally announced projects—Long Ridge in Hannibal Ohio, Matagorda County Texas, and Granbury Texas—remain largely capacity-neutral in public claims until folded into later acquisition announcements. Granbury subsequently appears as a 300-megawatt facility within the consolidated corporate footprint, though no independent evidence of permitting, construction progress, or interconnection queue status has been disclosed (claims 3499, 3509, 3531). A material timeline slip has emerged: the Long Ridge nameplate expansion of 505 megawatts carried a promised delivery of July 1, 2026. As of September 9, 2026, this target has slipped by at least 70 days without revised guidance (claim 1870, marked delayed). This represents Marathon's first concrete capacity deadline to miss in the available record. The Starwood Digital Ventures partnership, announced February 26, 2026, promised a 2.5-gigawatt AI infrastructure platform (claims 926, 7504). However, Marathon retains only 10 percent ownership initially, suggesting execution risk lies with Starwood; the structure involves 24-month development windows and conditional ownership escalation to 50 percent (claims 3662, 3664, 3501, 3528). A separate letter of intent with MPLX LP, announced March 1, 2026, remains at LOI status with no binding agreement disclosed (claims 3497, 3513). Exaion's acquisition carries forward complex multi-tranche conditions, with a third subscription contingent on closing by March 30, 2027, and post-closing ownership targets of 64 to 75 percent across tranches (claims 4520, 4521, 4522). Over the next 90 days, watch for: revised Long Ridge expansion timelines; MPLX LOI progression to binding agreement; Starwood platform construction commencement; and any restatement of the 2.2-gigawatt baseline, which currently rests almost entirely on announcement rather than third-party verification.

Marathon Digital's stated capacity accelerated from 1.7 gigawatts in April 2025 (claim 5021) to 2.2 gigawatts by late April 2026 (claim 9406), a 29 percent expansion funded by Exaion's acquisition (claim 3515), the Starwood Capital partnership targeting 2.5 gigawatts of AI infrastructure (claim 3660) with Marathon owning 50 percent of a 1.0 gigawatt joint venture (claims 3494, 3529), and Long Ridge Energy for $1.5 billion (claim 9088). Hashrate grew from 53.2 to 66.4 exahashes per second (claims 5022, 3508). These acquisitions cluster densely in February through April 2026, suggesting either aggressive opportunistic buying or capacity-guidance inflation timed to capital markets activity.

Timeline discipline has visibly fractured. Long Ridge's 505-megawatt CCGT expansion was promised by July 1, 2026 (claim 1870); as of September 9, it remains delayed, seventy days past deadline. This is material slippage at a cornerstone asset Marathon acquired specifically for near-term operational contribution. Granbury's 300-megawatt facility (claim 3509), central Texas expansions of 250 megawatts (claim 3532), and Hansford wind assets (114 megawatts nameplate, 240 megawatts interconnection, claims 3505, 3516) carry no specified delivery dates, blocking timeline verification from public claims.

The Starwood partnership disclosure contains ambiguity. Announcements assert 2.5 gigawatts of AI infrastructure capacity, yet the joint venture itself targets only 1.0 gigawatt (claim 3529). The remaining 1.5 gigawatts is unspecified—whether representing separate Marathon assets, future Starwood contributions, or aspirational planning cannot be determined from filings. Marathon retained rights at properties (claim 3666), implying shared control structures requiring visible construction or grid operator filings to verify.

Bitcoin acquisition claims (5426, 5350, 5346, 4665) lack transaction dates, settlement conditions, or funding mechanics, indicating these are contingent rather than committed. Exaion's ownership escalation to 75 percent by March 2027 (claim 4522) depends on EDF Pulse Ventures consent (claim 9271).

Over the next ninety days through early December, track Long Ridge's revised completion date and interconnection queue status. Monitor Starwood announcements for JV pipeline detail beyond 1.0 gigawatt. Watch for ground-breaking activity or equipment shipments at Granbury and central Texas sites. Any quarterly capacity guidance revision would signal execution-risk reassessment.

Marathon Digital's infrastructure roadmap from early 2025 through mid-2026 reveals a pattern of aggressive capacity stacking through acquisition rather than organic buildout, paired with significant guidance erosion on its partnerships. The company's stated capacity grew from 1.7 GW in April 2025 (claim 5021) to 1.9 GW by March 2026 (claim 3490), then to 2.2 GW by April 2026 (claim 9406). However, this growth masks a critical scaling-back: the Starwood Capital partnership, initially announced at 2.5 GW in capacity contribution (claims 926, 7504), was quietly reduced to 1.0 GW within weeks (claim 3529). That 60 percent revision occurred without explicit acknowledgement between February 26 and March 1, 2026, raising questions about the partnership's realistic scope. Meanwhile, the Long Ridge Energy & Power acquisition—a 1.5 billion dollar transaction announced in late April (claims 9088, 1868)—includes a nameplate expansion of 505 MW promised by July 1, 2026 (claim 1870). As of today, September 9, 2026, this deadline is past and tagged as delayed, suggesting execution friction. Most of Marathon's reported capacity growth stems from acquired assets: Exaion (February through August), Hansford County wind farm (240 MW interconnection; claims 3504, 3516), Nebraska data center (42 MW; claim 3514), and Long Ridge. Few claims reference new construction starts or engineering milestones. The Starwood arrangement includes a 24-month development decision timeline (claim 3664) and carries only a 50 percent joint venture ownership stake (claim 3528), implying Marathon controls just half of any eventual output. The MPLX LP letter of intent (claims 3513, 3524, 3503) remains non-binding as of the March 2026 filing date. Watch for Long Ridge expansion completion status over the next 90 days, actual breakground on Starwood-backed projects, and whether the March 30, 2027 Exaion third-tranche completion proceeds as planned (claims 4521, 4522). Any further capacity reductions or timeline slips would suggest the company is running ahead of realistic deployment velocity.

Marathon Digital's capacity trajectory has expanded dramatically but with increasing structural complexity that obscures underlying delivery progress. Through mid-2025, capacity announcements centered on discrete acquisitions: Exaion majority stake (claimed Feb 2026 at claim 2945), a 240 MW wind farm in Hansford County (announced and closing Feb 2025, claims 5116, 9274), and a 42 MW Nebraska data center (claim 3506). By April 2026, total operational and development capacity reached 2200 MW (claim 9406), representing near-doubling from the 1.9 GW stated in March (claims 3490, 3507). However, this growth masks a structural shift. The February 26 Starwood partnership announcement fundamentally reorganized the platform, converting Marathon's stated Exaion majority ownership into a 50 percent joint venture (claim 3528) with a 2.5 GW AI infrastructure target (claims 926, 7504) and a 24-month development timeline (claim 3664). This reallocation happened without explicit acknowledgement of why controlling ownership became a minority stake.

Promised delivery dates show early slippage. The Long Ridge Energy acquisition, announced April 1 at 1.5 billion USD (claim 9088) and confirmed April 29, included a 505 MW CCGT nameplate expansion promised by July 1, 2026 (claim 1870). That deadline is now past by over two months, marked in the record as delayed. The GC Data Center acquisition saw price revision from 178.65 million USD (claim 6401) in December 2023 to 183.8 million USD (claim 6206) by January 2024, without stated rationale. Wind farm acquisition and closing were announced simultaneously in mid-February 2025, suggesting either aggressive deal completion or pre-announcement timing that overstated near-term closure certainty.

Independent confirmation is sparse in the public record. Granbury facility is stated as announced with 300 MW capacity (claims 3499, 3509, 3531), but no permit filings or interconnection queue entries are cited. Hansford wind farm interconnection capacity (240 MW, claim 3504) versus nameplate output (114 MW, claims 3505, 3523) is listed without grid operator confirmation. The Starwood joint venture framework, with 24-month development windows starting February 2026, places major capacity delivery targets at February 2028. Over the next 90 days, track Long Ridge CCGT operational status and any revised commissioning timeline, Granbury site permitting progress, and whether Starwood platform announcements yield concrete grid interconnection filings.

Marathon Digital's announced capacity has scaled consistently upward over eighteen months, rising from 1.7 GW in April 2025 to 1.9 GW by March 2026 and 2.2 GW by late April 2026—a 29% increase in stated operational and development capacity. The growth reflects a deliberate portfolio diversification from bitcoin mining (which peaked at 66.4 EH/s in March 2026, claim 3522) toward AI/HPC infrastructure and renewable energy. The company disclosed specific facility allocations in March 2026: Granbury at 300 MW, Central Texas non-Granbury at 250 MW, East Ohio at 225 MW, North Texas wind at 180 MW, and Central Ohio at 150 MW, suggesting a 1.1 GW committed operational base with 1.1 GW in announced but less defined development status. Regarding timeline execution, Marathon faces a significant gap. The Long Ridge Energy & Power acquisition—a $1.5 billion CCGT plant purchase announced April 1, 2026 with 485 MW nameplate capacity—was stated as "expected closing" as of April 29, 2026 (claim 9412) without a firm deadline. Nearly five months later in September 2026, no closing confirmation appears in the claims record, raising questions about deal certainty. The Starwood Capital partnership announced February 26, 2026 includes a 24-month development decision timeline (to February 2028) for a 1.0 to 2.5 GW joint venture, but this is a partnership framework, not committed capacity. The Exaion European data center investment, with closing ownership stake established at 64% by late August 2025, contemplates a March 30, 2027 third transaction to reach 75% ownership—a conditional future milestone. Independent verification remains sparse in the claims data. The Hansford County wind acquisition (240 MW interconnection, 114 MW nameplate, claim 3516) and Nebraska data center (42 MW, claim 3514) are smaller bolt-on assets. Over the next ninety days, monitor whether Long Ridge officially closes and whether Marathon provides engineering updates on the 600 MW expansion pathway (claim 9401). Track Starwood joint venture site selection and environmental permitting, given the twenty-four-month development window. Verify Granbury facility construction status and whether the 300 MW allocation reflects actual power supply agreements. Any delays in Long Ridge closing or revised ownership percentages for Exaion would signal execution risk beneath the headline capacity numbers.

Marathon Digital's facility capacity announcements have expanded substantially but with increasingly complex ownership structures. The company reported 1.7 gigawatts of energy portfolio capacity in April 2025 (claim 5021), which grew to 1.9 gigawatts of global data center footprint by March 2026 (claim 3490). The major capacity additions, however, come through the February 2026 Starwood Capital Group partnership, which announced 2.5 gigawatts of combined AI/HPC infrastructure capacity (claims 926, 7504), though Marathon retains only 10–50 percent ownership stakes across various joint ventures (claims 3662, 3494). This represents a shift toward partnership-based growth rather than owned buildout.

Timeline performance shows early slippage. Marathon promised a 505 megawatt nameplate capacity expansion at Long Ridge by July 1, 2026 (claim 1870), but that deadline passed without public revision. The Long Ridge acquisition closed for 1.5 billion dollars in April 2026 (claims 1868, 927), yet promised July capacity additions have not materialized. The company acquired complementary assets including a 240 megawatt interconnected wind farm in Hansford County (claim 3516) and a 42 megawatt Nebraska data center (claim 3514), but integration timelines remain sparse in the record.

Mining hashrate increased from 53.2 exahashes per second in April 2025 to 66.4 by March 2026 (claims 5022, 3508), suggesting operational momentum. However, the Starwood partnership emphasizes AI/HPC compute rather than bitcoin mining, marking a potential strategic pivot. The joint venture includes a 24-month pre-development phase (claim 3664) running through February 2028, placing major capacity additions well into the future.

Watch whether Long Ridge delivers promised capacity in the coming weeks and whether Marathon provides revised timelines. The Starwood joint venture's 50 percent ownership stake (claim 3494) and 24-month development window (claim 3664) will determine whether the announced 1.0–2.5 gigawatt range is achievable. Independent confirmation through FERC interconnection queue updates and on-site construction progress would validate operational timelines.

Marathon Digital's announced capacity has scaled considerably over eighteen months, expanding from a 1.7 GW energy portfolio and 53.2 EH/s hash rate in April 2025 to a stated 1.9 GW global data center footprint with 70.7 EH/s by June 2026. The Long Ridge acquisition announced in April 2026 (claims 927, 1868) introduced a more ambitious infrastructure play: a 1000 MW development campus (claim 930) built around a 485 MW gas-fired plant with plans for 505 MW nameplate expansion. However, this expansion has already encountered timing friction. Claim 1870 promised Long Ridge's 505 MW nameplate capacity expansion by July 1, 2026; the claim is now marked as delayed, and with today's date at September 7, 2026, this represents a two-month slip without apparent public revision or rescheduling notice.

The Starwood Capital partnership announced in February 2026 (claim 3660) added a secondary capacity vector: 2.5 GW of AI/HPC infrastructure through a 50-75 percent joint venture structure with a 24-month development decision timeline (claim 3664). This suggests real capacity ambitions of 1-2.5 GW from the partnership alone (claim 3512), contingent on joint venture governance and energy access tied to an MPLX LP letter of intent (claims 3513, 3524). The specific Granbury facility shows 300 MW stated capacity (claims 3499, 3509, 3531), Central Texas additional capacity of 250 MW (claim 3532), and smaller asset acquisitions including a Nebraska data center and Hansford County wind farm (114 MW nameplate, 240 MW interconnection).

Independent verification is limited within the provided record. The Long Ridge delay suggests capacity sequencing is more constrained than announcements imply, particularly for converting nameplate or interconnection rights into energized capacity. The hashrate growth from 53.2 to 70.7 EH/s indicates real mining equipment deployment, though this outpaced the modest capacity increase from 1.7 to 1.9 GW, suggesting higher utilization density or temporary leased capacity arrangements. The Starwood deal's 24-month development window ends February 2028; watch for any extension, reduction in capacity targets, or failure to meet development milestones. More immediately, expect clarity on Long Ridge's revised expansion timeline within 90 days, as a second public slip would signal operational constraints more fundamental than initial messaging suggested.

Marathon Digital's headline capacity claims have expanded dramatically but rest increasingly on future partnerships rather than operational assets. Through mid-2025, the company reported a 1.7 GW energy portfolio (claim 5021) with 53.2 EH/s of energized hashrate (claim 5022). By March 2026, announced global data center capacity jumped to 1.9 GW (claims 3520, 3490), though this marked marginal growth despite multiple acquisitions: a 42 MW Nebraska facility (claim 3514), Hansford County wind with 240 MW interconnection (claim 3496), and 66.4 EH/s mining capacity (claim 3522). The material inflection came via the Starwood partnership launched in late February 2026, which committed to 2.5 GW of AI/HPC capacity (claims 926, 7504) split 50-50, meaning Marathon's share reaches only 1.25 GW. That venture carries a 24-month development timeline (claim 3664), placing substantial delivery into 2028.

The Long Ridge acquisition—announced at $1.5 billion in April 2026 (claim 1868)—was positioned to materially shift this trajectory. The facility's 485 MW nameplate capacity (claim 1869) promised expansion to 505 MW by July 1, 2026 (claim 1870), marked as delayed. Today is September 6, 2026. No public update has acknowledged this miss or provided a revised timeline. The 1000 MW "development campus" capacity cited in April (claim 930) lacks specificity on timeline or phase-gate criteria. Parallel claims about Exaion's majority acquisition (claim 3515, March 2026) and a planned third conditional transaction in March 2027 (claims 4521, 4522) suggest integration is incomplete and capital deployment remains staged.

This pattern reflects a portfolio increasingly weighted toward announced deals and future-dated platforms rather than energized infrastructure. The redundancy across March 2026 facility claims (Granbury, Central Texas, East Ohio figures repeated multiple times across claims 3499, 3509, 3531) suggests data consolidation rather than new capacity deployment. Independent indicators remain sparse: no recent interconnection queue updates, FERC filings, or construction permits have been cited for the Starwood platform or Long Ridge expansion phase. Watch for Long Ridge expansion completion documentation within 90 days, groundbreaking for Starwood development sites, and any revised guidance on the 2.5 GW partnership timeline—all have material bearing on whether capacity tracks as announced or faces further slippage.

Marathon Digital's infrastructure narrative has expanded materially but unevenly across 2025-2026. The company moved from 1.7 GW energy portfolio capacity and 53.2 EH/s hashrate in April 2025 (claims 5021, 5022) to stated 1.9 GW global data center footprint and 66.4 EH/s by March 2026 (claims 3490, 3522), representing modest organic growth before major acquisitions. The April 2026 Long Ridge Energy acquisition—valued at $1.5 billion—added a 485 MW nameplate gas plant with promised expansion to 505 MW by July 1, 2026 (claims 927, 1869, 1870). That target has now slipped: as of early September, the expansion remains marked delayed, roughly 65 days past promise. This is Marathon's only dated delivery commitment among dozens of announcements, and it has already failed.

The capacity picture depends heavily on paper rather than hardware. The Starwood Capital partnership (February 2026) promises 2.5 GW of AI/HPC infrastructure (claims 926, 7504), but this is structured as a joint venture with Marathon holding 50 percent ownership and Starwood as managing member (claims 3528, 3663), with a 24-month development timeline (claim 3664) that extends into 2028. Similarly, the MPLX LP letter of intent (March 2026) does not constitute binding power access (claims 3513, 3524). Marathon's owned-and-operated capacity represents only 70 percent of the stated 1.9 GW portfolio (claim 3521), meaning roughly 570 MW depends on third-party operators. Granbury, Texas is cited at 300 MW (claims 3499, 3509), but sits on the announced-facility list without status updates or operational proof.

No independent evidence appears in these filings—no permits, FERC interconnection queue standings, or equipment shipment milestones. The Exaion controlling stake acquisition (February 2026, claim 2945) adds European datacenters but has not yet translated into disclosed capacity contributions to Marathon's total count. A third Exaion share purchase is scheduled for March 30, 2027 (claims 4521, 4522), contingent on unspecified conditions.

Over the next 90 days, monitor Long Ridge expansion completion—whether the 20 MW expansion gap closes and by when. Track Starwood and MPLX announcements for binding agreements or LOI terminations. Watch for Q3 2026 earnings disclosures on actual energized capacity and hashrate. Any further delays in Long Ridge without revised timelines would signal execution risk beyond isolated project friction.

Marathon announced three projects with escalating capacity profiles. The company reported 1.7 GW of energy portfolio capacity in April 2025 (claim 5021), growing to 1.9 GW by March 2026 (claims 3490, 3507)—a 12% increase over eleven months. The Long Ridge acquisition announced April 29, 2026 (claim 1868, valued at $1.5B) committed 485 MW of immediately available nameplate capacity (claim 1869), with a development campus expansion target of 1000 MW (claim 930). Yet concurrent hashrate disclosures present a sharply different expansion trajectory: 53.2 EH/s in April 2025 (claim 5022) grew to 70.3 EH/s by Q2 2026 (claim 2950)—a 32% increase. This divergence between facility capacity claims and operational hashrate growth suggests either that announced capacity significantly lags deployed infrastructure, or that utilization rates per megawatt are expanding faster than facility additions.

Timeline slippage is evident. The Long Ridge nameplate expansion to 505 MW was promised by July 1, 2026 (claim 1870), now marked delayed; as of September 6, 2026, this represents a two-month slip at minimum. The Starwood Digital Ventures partnership announced February 26 (claim 3660) specifies a 24-month development timeline (claim 3664), deferring development decisions to early 2028 and material capacity contributions beyond the current planning window. The Exaion transaction sequence similarly defers control: majority ownership announced March 2026 (claims 3515, 3526) is superseded by a scheduled March 2027 conditional third-transaction subscription (claim 4521) required to reach 75% ownership (claim 4522)—a one-year lag in actual command authority.

The record contains no independent verification through permits, interconnection queue entries, or on-the-ground reporting. This prevents assessment of whether announced timelines reflect engineering or regulatory reality.

Within 90 days, watch: Long Ridge equipment energization status, Starwood partnership milestone announcements, Granbury facility 300 MW capacity (claim 3509) grid-connection progress, and Exaion's contribution to the stated 66.4 EH/s hashrate target (claim 3508). The March 2027 Exaion third-transaction deadline is the nearest hard external milestone. Long Ridge delays compound the acquisition's credibility implications.

Marathon Digital's facility capacity narrative has undergone dramatic revision within six months. In April 2026, the company positioned a 1.9 GW global footprint (claim 3490) anchored by Long Ridge's 505 MW nameplate expansion, promised for July delivery (claim 1870). By August, that framing dissolved entirely. The company announced acquisition of Matagorda County's 2.0 GW site (claim 8500) and recast its power portfolio expansion target at 4.8 GW (claim 8498)—nearly triple the April baseline. Capacity targets escalated from 1.7 GW in April 2025 (claim 5021) to approximately 4.8 GW today, a pattern that suggests strategy reset rather than measured scaling.

The Long Ridge slip is instructive. Claim 1870 flags the July 2026 deadline as delayed; the company has not publicly revised or acknowledged this miss in the claims record, instead overlaying new asset announcements. Claims 3531 and 3499 repeat Granbury's 300 MW capacity verbatim, as do claims 1869 and 929 for Long Ridge, signaling internal revisions presented as fresh statements rather than amendments. This pattern obscures whether prior targets remain viable or have been abandoned.

The Starwood Capital partnership (claim 3660, announced February 2026) explicitly defined a 24-month development decision timeline (claim 3664), placing material milestones for the 2.5 GW joint venture target between February and October 2027. With Starwood positioned as managing member (claim 3663), Marathon's reported owned-and-operated asset ratio of 70 percent (claims 3490–3491) becomes contingent on partnership cadence. A secondary Exaion transaction is scheduled for March 2027 (claims 4521–4522), potentially unlocking further controlling interest progression.

The Matagorda acquisition arrived with no prior disclosure visible in this claims set, suggesting either compressed decision-making or reporting lag. The company has not specified whether Matagorda holds confirmed grid interconnection rights, power purchase agreements, or transmission queue position—items critical to validating the 2.0 GW nameplate rating but absent from available claims.

Watch over the next ninety days for Long Ridge's actual energization sequence against the now-delayed 505 MW commitment; Matagorda's interconnection queue status and regulatory approvals; visible Starwood joint venture site selection or permitting; and clarification of whether the 2.5 GW Starwood target and 4.8 GW company total represent overlapping or distinct capacity buckets, since their relationship remains unexplained.

Marathon Digital's announced capacity has remained essentially flat since March 2026. After signaling 1.7 GW of energy portfolio capacity in April 2025 (claim 5021), the company stabilized at 1.9 GW global data center footprint by early March 2026 (claim 3490) and has repeated that 1.9 GW figure through August 2026 (claim 8257). Meanwhile, mining hashrate has grown consistently—from 53.2 EH/s in April 2025 (claim 5022) to 66.4 EH/s by March 2026 (claim 3522) to 70.3 EH/s by June 2026 (claim 2950)—suggesting the company is extracting more throughput from existing capacity rather than building new infrastructure. This pattern implies either slower-than-announced build-out or aggressive retrofitting of existing assets.

Promised timelines show material slippage on Long Ridge. The company announced a 505 MW nameplate capacity expansion at its Long Ridge facility (claim 1870) with a delivery target of July 1, 2026. As of September 5, 2026, this expansion is marked as delayed. The $1.5 billion Long Ridge acquisition itself closed (claims 927, 1868), but the stated expansion timeline has not held. The company promised a development campus capacity of 1000 MW at Long Ridge (claim 930), yet no subsequent updates clarify whether this target remains on track, scaled back, or merely illustrates pre-acquisition aspiration versus post-close reality.

The Starwood Capital partnership announced February 2026 created capacity claims of 2.5 GW (claim 926) versus a joint venture target of only 1.0 GW AI/HPC capacity (claim 3529). This gap suggests either marketing language around potential versus actual committed build-out, or that the 2.5 GW encompasses uncommitted corridor capacity. Independent verification of interconnection queues, FERC filings, or grid operator interconnection agreements for these properties has not surfaced in public records, and on-the-ground reporting from the three named projects—Long Ridge, Granbury, and Matagorda County—remains limited.

Watch for Long Ridge expansion completion status over the next 90 days. Monitor whether Starwood joint venture development decisions advance beyond letters of intent (claim 3513 references MPLX LP LOI). Track hashrate announcements as a leading indicator of utilization pressure, particularly if mining efficiency gains plateau and Marathon begins announcing new capacity midstream. Any updates on Matagorda County and Granbury status will clarify whether these remain active or have shifted to lower priority.

Marathon's announced capacity scaled from 1.7 GW in April 2025 (claim 5021) to 1.9 GW by March 2026 (claim 3490)—a 200 MW increase despite substantial acquisition activity through the Exaion deal and Hansford wind farm. Long Ridge, acquired in April 2026 at 485 MW nameplate (claim 1869), represents a parallel capacity vector with stated development potential reaching 1000 MW (claim 930). Stated capacity additions appear modest relative to deal volume, suggesting significant claimed capacity is not yet operational or represents partner assets.

Discrepancies between facility-level and aggregate claims suggest counting ambiguity. March 2026 granular announcements itemized Granbury at 300 MW (claim 3499), East Ohio at 225 MW (claim 3533), central Texas non-Granbury at 250 MW (claim 3532), and smaller sites, totaling approximately 1.2 GW. Against this the 1.9 GW global claim (claim 3490) indicates the gap reflects either partner-owned or future capacity. The Hansford wind farm exemplifies this pattern: 240 MW interconnection queue capacity (claim 3504) against 114 MW actual nameplate (claim 3505), indicating grid congestion typical of PJM and ERCOT regions.

Announced timelines have shifted. Long Ridge's promised 505 MW expansion for July 1, 2026 (claim 1870) is marked delayed as of early September. Earlier GC Data Center closure, stated for January 2024 (claim 6379), has no explicit completion record. Starwood Capital partnership's 24-month development decision period (claim 3664) extends to February 2028, providing no near-term capacity clarity.

Record contains no independent verification—no interconnection application status, permit filings, or third-party site observation. In next 90 days, monitor: Long Ridge 505 MW energization; Matagorda County 2.0 GW announcement (August 2026, claim 1216) permitting and queue position; Starwood JV operational timeline; hashrate sustainability (70.3 EH/s stated Q2 2026, claim 2950) against announced infrastructure constraints.

Marathon Digital announced expanding capacity through owned facilities and partnerships. Owned data center footprint grew from 1.7 gigawatts in May 2025 (claim 5021) to 1.9 gigawatts globally by March 2026 (claim 3520), with 70 percent owned-and-operated (claim 3521). Simultaneously, the February-March 2026 Starwood Capital partnership introduced 2.5 gigawatts under partnership structure (claim 7504) with a 1-gigawatt AI/HPC capacity target and 50 percent Marathon ownership (claims 3529, 3494). The April 2026 Long Ridge acquisition committed 1.5 billion dollars with a 1-gigawatt campus expansion target (claims 1868, 930). The August 2026 Matagorda County announcement introduced a 2-gigawatt facility with no prior disclosure, suggesting either delayed announcement or repositioned capacity targets.

Timeline discipline deteriorated visibly. The Long Ridge nameplate expansion to 505 megawatts carried a July 1, 2026 deadline now marked delayed (claim 1870). Five months past that miss, acquisition closure remains undisclosed despite operational references in Q2 2026 (claim 2950). The Starwood partnership announcement fragmented across February 26 and March 1 as distinct "partnership" and "strategic agreement" filings (claims 3660, 3511)—either scope revision or incomplete initial disclosure. Exaion ownership percentages shifted: 64 percent post-closing (August 2025, claim 4520) to 75 percent expected by March 30, 2027 (claim 4522), contingent on a third transaction still seven months distant (claim 4521).

Mining hashrate rose 32 percent over 15 months: 53.2 exahash-per-second (May 2025, claim 5022) to 70.3 EH/s (Q2 2026, claim 2950). Growth accelerated into March 2026 then decelerated in subsequent quarters—a velocity shift worth monitoring against announced capacity ramps.

Independent verification remains sparse. No FERC interconnection queue entries, permits, or construction timelines surface for the Starwood platform or Matagorda County 2-gigawatt facility. Hansford County's 240-megawatt grid interconnection and 114-megawatt wind nameplate (claims 3504, 3505) are testable if ERCOT generation records materialize.

Watch for: (1) Long Ridge facility energization and confirmed closing date; (2) Starwood platform groundbreaking and revised timeline; (3) Matagorda County permit activity; (4) Exaion third-transaction execution by March 30, 2027; (5) public interconnection queue entries, particularly for the 2-gigawatt Matagorda facility.

Marathon's capacity narrative has expanded substantially but with mixed timeline execution. Hash rate claims grew from 53.2 EH/s in April 2025 (5022) to 70.3 EH/s by June 2026 (2950), a 32% increase in thirteen months, while data center footprint expanded from 1.7 GW to 1.9 GW (5021, 3490). The additions track a deliberate multi-leg acquisition strategy: Exaion for European HPC capability, Hansford County wind for 240 MW interconnection capacity, and Long Ridge Energy for a 485 MW natural-gas asset acquired for $1.5B (1868, 3516, 2947). This mix tilts the portfolio toward owned-and-operated sites at 70% (3491).

However, timing has proven elastic. The GC Data Center acquisition, promised closed in January 2024 (6379), resurfaced in May 2024 with revised terms and a higher price tag of $183.8M (6206), suggesting the earlier commitment slipped without explicit public acknowledgement. Long Ridge's nameplate capacity expansion—promised by July 1, 2026—is now marked delayed (1870). The company's Starwood Digital Ventures partnership, announced February 2026 with ambitious 2.5 GW AI infrastructure rhetoric (926, 7504), scaled the near-term joint venture target down to 1.0 GW (3512), deferring full realization across a 24-month development window (3664).

Starwood's structure compounds this deferral: Marathon holds 50% of joint ventures with a 24-month development decision timeline and reported 64% post-closing ownership in Exaion as of August 2025 (4520), with a third equity transaction conditional on March 30, 2027 (4521). This cascades material capacity accretion into 2027 and beyond. Independent evidence—interconnection queue entries, permitting filings, or on-site construction reporting for Matagorda County's newly announced 2.0 GW facility (1216)—remains sparse in the public record, making the stated timeline difficult to validate.

The 90-day watch list centers on Long Ridge recovery: whether the delayed 505 MW nameplate capacity comes online by September/October 2026 and confirms Marathon's ability to execute major thermal asset transitions. Second, track Starwood progress on jointly-owned properties; if the 24-month timeline holds credibility, meaningful updates should surface by Q4 2026. Third, monitor whether Matagorda County's 2.0 GW claim (announced August 2026 with no delivery date) receives interconnection queue confirmation or permitting milestones. If these don't materialize, the company's growth narrative compresses into 2027.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claim #15461 and echoed in claim #15477, and on 2018-06-28 that directors resigned, cited as claim #15476. What has not been said is who received those shares and what they were worth on the termination date. It is also unclear why the merger collapsed and whether the director resignations were tied to it. Were the departing directors affiliated with Global Bit Ventures?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options, cited as claim #15389, split into 2.5 million issued and 2.5 million unvested with change-of-control acceleration, cited as claims #15385 and #15386, alongside a $350,000 base salary, cited as claim #15388. What has not been said is the strike price, the vesting schedule and the definition of a change-of-control trigger. Given the 10,000,000-share 2018 Equity Incentive Plan, cited as claim #15460, did a single executive receive half of the entire plan reserve, and were there any bonus or performance conditions attached?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459. What has not been said is any operating detail. How many machines, at what hashrate and power capacity, at which physical site, and under what hosting or power-purchase counterparty is this expansion being carried out?

The company disclosed on 2018-08-13 substantial doubt about its 12-month cash runway, cited as claim #15462, in the same filing that announced the mining expansion and a 10,000,000-share equity plan. What has not been said is the cash balance, the monthly burn rate and the specific financing intended to close the gap. How dilutive is that financing expected to be relative to the share reserves just authorized?

The company disclosed on 2018-07-12 a commitment to appoint two new independent directors by 2019-01-07, cited as claim #15465. On 2018-08-21 it disclosed only Michael Berg's appointment and regained NASDAQ compliance, cited as claims #15435 and #15438. What has not been said is whether a second independent director was ever seated. Was Fred Thiel, named Compensation Committee Chairman in claim #15437, the counterpart, and did the board review his independence given his later executive role?

The company disclosed on 2014-03-30 that patent revenue generation had commenced, cited as claim #21014, under licensing agreements with Zhone and Siemens Technology, cited as claims #21022 and #21018. In the same filing it named IP Navigation Group as a key supplier, cited as claim #21024. What has not been said is the dollar consideration under each license and the revenue share or fee paid to IP Navigation Group. The employee count, cited as claim #21026, was also left unquantified. How much of that patent revenue actually stayed with the company?

Marathon disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares (claim #15461), following the 2018-06-28 termination notice (claim #15477). The record does not say who received the shares, what they were valued at on issuance, how the fee was accounted for, or why the deal was terminated. What was the dollar cost of the break-up fee, and did any officer or director have a relationship with Global Bit Ventures or its holders?

The 2018-10-15 disclosures on Merrick Okamoto's appointment as CEO and Chairman (claims #15384, #15387) cite a 350,000 USD base salary (claim #15388). They also cite option grants that do not reconcile: 2,500,000 shares (claim #15385), 2,500,000 unvested shares with change-of-control acceleration (claim #15386) and 5,000,000 shares (claim #15389). The strike price, vesting schedule and performance conditions are absent, as is what share of the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460) these grants consume. Is the total grant 2.5 million or 5 million options, and on what terms does it vest and accelerate?

The company stated a cryptocurrency mining expansion on 2018-08-13 (claim #15459). In the same filing it flagged substantial doubt about a 12-month cash runway (claim #15462). It has not named a mining site address, installed or targeted hashrate, megawatts of power capacity, a power or hosting counterparty, or the financing amount and timing needed to fund the expansion and cure the going-concern uncertainty. Where are the machines, how much power is contracted and at what rate, and how much capital is required over what period?

The 2014-03-30 annual disclosures list licensing agreements with Zhone and Siemens Technology (claims #21022, #21018). They also list a key supplier relationship with IP Navigation Group (claim #21024) and a string of portfolio acquisitions from Siemens, Mosaid, Delphi, Intergraph, TeleCommunication Systems, CyberFone and TT IP (claims #21031, #21029, #21023, #21020, #21019, #21017, #21021). The purchase prices, revenue-share or contingency-fee splits owed to the sellers and to IP Navigation Group, and the license revenue attributable to each counterparty are not given. What did each portfolio cost, and what percentage of recoveries is paid away before Marathon books revenue?

Marathon disclosed director resignations on 2018-06-28 (claim #15476) and a Nasdaq deficiency with compliance-plan and cure deadlines (claims #15465, #15466, #15467, #15468). It then reported regained compliance on 2018-08-21 (claim #15438) after appointing Michael Berg and Fred Thiel to chair the audit and compensation committees (claims #15435, #15437). Thiel received 50,000 options (claim #15396). The record does not name the resigning directors, give their reasons, or say whether any disagreement with management was cited. Separately, the 2013-04-04 entries on a 26-country, Africa-focused Mara group led by the Thakkar family (claims #20897-#20903) appear to describe a different entity, so the record's attribution should itself be verified before anyone relies on it.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO with a $350,000 base salary and 5,000,000 stock options (claims #15387, #15388, #15389), yet the same date's record also shows 2.5 million options issued to the CEO (claim #15385). What has not been said is whether these are separate tranches, a vesting split or a correction, and what the strike price, vesting schedule and performance conditions are against a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460). Taken together, these grants appear to consume up to three-quarters of that pool.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a 3,000,000-share break-up fee (claim #15461), following the 2018-06-28 termination and director resignations (claims #15476, #15477). What has not been said is which party owed the fee, its dollar value at issuance, whether the shares were actually delivered, and whether the resigning directors had any relationship with Global Bit Ventures.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) alongside going-concern-style uncertainty about a 12-month cash runway (claim #15462). What has not been said is the hashrate or machine count being added, the hosting site and its power capacity and cost per kWh, and how that expansion would be financed given the runway warning.

The company disclosed on 2018-08-21 that it regained NASDAQ listing compliance after appointing Michael Berg as Audit Committee Chairman and Fred Thiel as Compensation Committee Chairman (claims #15435, #15437, #15438). The 2018-07-12 notices had set separate deadlines for two independent directors and for compensation committee compliance (claims #15465, #15468). What has not been said is whether both deadlines were fully satisfied. The record also does not say what compensation or prior relationships these directors had with management, given that the Compensation Committee chair set the CEO package approved two months later.

The company disclosed on 2014-03-30 licensing agreements with Zhone and Siemens Technology, a key supplier relationship with IP Navigation Group, and the commencement of patent revenue (claims #21018, #21022, #21024, #21014). What has not been said is the license fees or revenue concentration by licensee, and the contingency-fee share or other economics paid to IP Navigation Group. The employee-count claim (#21026) also carries no actual number.

The record attributes to this company 2013-2014 claims about operations in 26 countries, an African presence in 19 countries, the Mara Foundation, and executives Ashish Thakkar, Jagdish Thakkar and Prashant Manek (claims #20897 through #20903, #20895, #20896). Those claims sit alongside a 2013-02-04 rename to Marathon Patent Group (claim #21045). What has not been established is whether these claims describe Marathon at all or the unrelated Mara Group, and this should be resolved before any of them is used as a disclosure baseline.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is which party bore the fee, the per-share value and accounting treatment of those shares, and what specifically caused the deal to collapse two months after directors resigned (claim #15476). Did the termination leave Marathon owning any GBV mining equipment or obligations?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options alongside a $350,000 base salary, cited as claims #15389 and #15388, two months after reserving 10,000,000 shares under the 2018 Equity Incentive Plan (claim #15460); what has not been said is the strike price, vesting schedule, and any performance conditions on a single grant equal to half the plan pool. The interim chief executive named in claim #16644 also goes unidentified, so what was the leadership and compensation arrangement between April and October 2018?

The company disclosed on 2018-08-13 a cryptocurrency mining expansion, cited as claim #15459, in the same filing that flagged substantial doubt about a 12-month cash runway (claim #15462); what has not been said is the hosting site or physical location, megawatts of power secured, the power or hosting counterparty, machine count or hashrate targets, and how the expansion would be funded given the going-concern language.

The company disclosed on 2018-07-12 a NASDAQ requirement to add two independent directors by 2019-01-07 and to cure compensation committee composition by 2019-06-28, cited as claims #15465, #15467 and #15468, then reported regained compliance on 2018-08-21 with Michael Berg and Fred Thiel seated as committee chairs (claims #15438, #15435, #15437); what has not been said is whether the separate compensation committee deficiency was fully cured, and whether either new director had any prior financial or advisory relationship with management that would bear on independence.

The company disclosed on 2014-03-30 that IP Navigation Group was a key supplier and that patent revenue had commenced, cited as claims #21024 and #21014, alongside licensing agreements with Siemens and Zhone (claims #21018, #21022); what has not been said is the revenue share or contingency fee payable to IP Navigation, the identities of the licensees actually generating revenue, and how much of that revenue concentrated in any single settlement. The same filing references an employee count (claim #21026) without the number itself.

The company disclosed a 10-K filing deadline of 2018-04-15 under interim leadership, cited as claim #16645; what has not been said is whether the report was filed on time, whether an NT 10-K was used, and whether the auditor raised any material weakness tied to the leadership transition. Separately, claims #20896 through #20903 describe Mara Group, an unrelated African conglomerate, and appear to be misattributed to this ticker rather than reflecting any Marathon disclosure.

The company disclosed on 2018-08-13 that its terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claim #15461, following the 2018-06-28 termination notice in claim #15477; what has not been said is who received those shares, what they were worth at issuance, how the fee was recorded, and what caused the deal to collapse. Was the termination initiated by Marathon or by Global Bit Ventures, and was the share fee negotiated in exchange for any assets, mining equipment or hosting rights that stayed with the company?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options alongside a $350,000 base salary, cited as claims #15389 and #15388, two months after reserving 10,000,000 shares under the 2018 Equity Incentive Plan, cited as claim #15460; what has not been said is the strike price, vesting schedule, or any performance conditions on a grant equal to half the plan reserve, nor whether Okamoto had any prior financing, trading or advisory relationship with the company before his appointment. How much of the plan reserve remained after the October 2018 grants to Okamoto, Lieberman, Crawford, Rudolph and Thiel, cited as claims #15389 and #15392 through #15396?

The company disclosed on 2018-08-13 a cryptocurrency mining expansion, cited as claim #15459, in the same filing that flagged uncertainty over a 12-month cash runway, cited as claim #15462; what has not been said is the machine count, hashrate, power draw in megawatts, physical site location, or the hosting or power counterparty behind that expansion. How was an expansion to be funded when the company could not assure 12 months of liquidity, and what cash balance and monthly burn rate underlay that warning?

The company disclosed on 2018-06-28 that directors had resigned, cited as claim #15476, and on 2018-07-12 laid out Nasdaq deadlines to appoint two independent directors and restore audit and compensation committee compliance, cited as claims #15465 through #15468; what has not been said is which directors left, whether any resignation stemmed from a disagreement with management, and whether the departures were tied to the Global Bit Ventures termination. Did the 2018-08-21 appointments of Michael Berg and Fred Thiel, cited as claims #15435, #15437 and #15438, come from candidates with any prior ties to management or to the terminated merger counterparty?

The company disclosed on 2014-03-30 that IP Navigation Group was a key supplier, cited as claim #21024, alongside licensing agreements with Siemens Technology and Zhone, cited as claims #21018 and #21022, and the commencement of patent revenue, cited as claim #21014; what has not been said is the fee split or contingency percentage owed to IP Navigation Group on recoveries, the dollar value of the Siemens and Zhone licenses, or which unnamed licensees generated the initial revenue. What share of 2013 patent revenue came from a single licensee, and what were the upfront and contingent payment obligations on the roughly ten portfolios acquired, cited as claims #21015 through #21035?

The company disclosed on 2018-04-01 an Interim Chief Executive Officer and a 10-K due by 2018-04-15, cited as claims #16644 and #16645; what has not been said is who held the interim role, the compensation attached to it, and whether the 10-K was filed on time or under an extension. Why was an interim chief executive in place during the pivot to cryptocurrency mining, and when was the permanent appointment made?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee paid in 3,000,000 shares, cited as claim #15461 (with the termination itself first noted on 2018-06-28, claim #15477). What has not been said is why the deal collapsed, who the recipients of those shares are, what value and per-share price were recorded for the fee, and whether any Global Bit Ventures principals had prior relationships with Marathon insiders.

The company disclosed on 2018-10-15 a grant of 5,000,000 stock options to Merrick Okamoto alongside a $350,000 base salary, cited as claims #15389 and #15388, only two months after reserving 10,000,000 shares under the 2018 Equity Incentive Plan, cited as claim #15460. What has not been said is the strike price, vesting schedule, and any performance conditions on a grant equal to half the entire plan reserve, or how the board, whose compensation committee was being reconstituted under claims #15437 and #15468, justified that concentration in a single executive.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion, cited as claim #15459, in the same filing that flagged substantial doubt about a 12-month cash runway, cited as claim #15462. What has not been said is how many machines are deployed or on order, the target hashrate, where the hosting site is physically located, what power cost per kWh has been contracted, and how the expansion is to be financed given the stated liquidity doubt.

The company disclosed on 2018-06-28 that directors had resigned, cited as claim #15476, followed by a plan to appoint two independent directors and a NASDAQ compliance plan due 2018-08-24, cited as claims #15465 and #15466, and then said on 2018-08-21 that compliance was regained, cited as claim #15438. What has not been said is which directors left and why, whether any resignation letters cited disagreements with management, and whether the separate compensation committee deadline of 2019-06-28 in claim #15468 was ever met.

The company disclosed on 2018-04-01 that an interim chief executive was in place and that a 10-K deadline of 2018-04-15 applied, cited as claims #16644 and #16645. What has not been said is who held the interim role, what circumstances led to the prior CEO's departure, and whether the 10-K was filed on time or required an NT 10-K extension.

The company disclosed on 2014-03-30 that patent revenue generation had commenced, that IP Navigation Group was a key supplier, and that it was actively pursuing enforcement litigation, cited as claims #21014, #21024 and #21025, across more than a dozen acquired portfolios such as those from Siemens, Intergraph, Delphi and CyberFone (claims #21031, #21020, #21023, #21017). What has not been said is the dollar amount of licensing revenue, the share of proceeds paid to IP Navigation Group and to the original patent sellers, the number of active cases, and the purchase price paid for each portfolio. Separately, the 2013-04-04 and 2014 claims #20896 through #20903 describe Mara Group, an unrelated African conglomerate, and appear to be misattributed to this company.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares (claim #15461, following the 2018-06-28 termination notice in claim #15477); what has not been said is which party paid that fee, what the shares were worth at issuance, and whether any GBV mining equipment or hosting arrangements that motivated the deal transferred to the company anyway. The same 2018-06-28 filing reports that directors resigned (claim #15476), but the record does not name them or give their reasons. Did any of them cite disagreement over the GBV transaction?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options alongside a $350,000 base salary (claims #15389 and #15388), against a 2018 Equity Incentive Plan reserve of 10,000000 shares disclosed on 2018-08-13 (claim #15460). What has not been said is the strike price, the vesting schedule, and any performance conditions on a single grant that consumes half the plan. The record also does not explain why that grant is 25 times the 200,000 options given to the incoming CFO David Lieberman (claim #15392).

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that flagged uncertainty about a 12-month cash runway (claim #15462). What has not been said is where the machines are sited, who hosts them, how many units are deployed, what hashrate and megawatt load they represent, what the contracted power price is, and how the expansion is being funded if the runway itself is in doubt.

The company disclosed on 2014-03-30 that patent revenue generation had commenced (claim #21014) and that it had licensing agreements with Siemens Technology and Zhone (claims #21018 and #21022). That revenue was supported by a key supplier relationship with IP Navigation Group (claim #21024) and ongoing enforcement litigation (claim #21025). What has not been said is which licensees produced the revenue, how much each contributed, and what share of recoveries IP Navigation Group takes under its contingency arrangement. The record also does not disclose any upfront payments or revenue-share obligations owed to the sellers of the acquired portfolios, including Sampo, Mosaid, CyberFone, Intergraph, TeleCommunication Systems, Delphi and TT IP (claims #21015 through #21023).

The company disclosed an Interim Chief Executive Officer on 2018-04-01 (claim #16644), with a 10-K filing deadline of 2018-04-15 (claim #16645), but the record names neither the interim officer nor whether the 10-K was filed on time. It then disclosed on 2018-07-12 a set of NASDAQ deficiencies that carried staggered deadlines: a compliance plan by 2018-08-24, two independent directors by 2019-01-07, and compensation committee compliance by 2019-06-28 (claims #15465 through #15468). It reported compliance regained on 2018-08-21 with the appointments of Michael Berg and Fred Thiel (claims #15438, #15435 and #15437). What has not been said is which specific listing rules were breached and whether the separate compensation-committee deficiency was also cured at that time or remained open.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO with a $350,000 base salary and stock options described as both 2.5 million shares and 5 million shares, with 2.5 million unvested options carrying change-of-control acceleration, cited as claims #15384, #15385, #15386, #15388 and #15389. What has not been said is the reconciliation: is the grant 2.5 million or 5 million options in total, what are the strike price, vesting schedule and expiry, and what defines a change of control that would accelerate vesting?

The company disclosed on 2018-06-28 and 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee paid in 3,000,000 shares, cited as claims #15477 and #15461. What has not been said is who paid whom, what the shares were valued at for accounting purposes, what assets or rights (if any) changed hands, and whether any Marathon director or officer had a prior relationship with Global Bit Ventures or its principals.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that flagged uncertainty about a 12-month cash runway, cited as claim #15462. What has not been said is where the mining operation is physically located, which hosting or power counterparty supplies it, how many machines and how much hash rate are deployed, what the all-in power cost per kWh is, and what cash balance, monthly burn and financing plan underlie the going-concern language.

The company disclosed on 2014-03-30 that patent revenue generation had commenced and that it held licensing agreements with Siemens Technology and Zhone, cited as claims #21014, #21018 and #21022, alongside a key supplier relationship with IP Navigation Group, cited as claim #21024. What has not been said is how much revenue each license produced, whether those amounts were one-time settlements or running royalties, and what share of recoveries IP Navigation Group receives under its fee arrangement.

The company disclosed on 2018-04-01 that an interim Chief Executive Officer was in place and that its 10-K was due by 2018-04-15, cited as claims #16644 and #16645. What has not been said is who held the interim role, why the prior CEO departed, and whether the annual report was filed on time or under an extension. The same question carries into the 2018-06-28 director resignations, cited as claim #15476: which directors resigned, and did any cite a disagreement with management?

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477) and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461); what has not been said is why the deal collapsed, which party paid and received those shares, whether they were actually issued, and what dollar value the company attached to them.

The company disclosed a 2018 Equity Incentive Plan reserve of 10,000,000 shares on 2018-08-13 (claim #15460) and on 2018-10-15 a grant of 5,000,000 options to Merrick Okamoto (claim #15389), along with 2,500,000 unvested options carrying change-of-control acceleration (claim #15386); what has not been said is the strike price, vesting schedule and holder of the accelerating options, and whether there is any related-party relationship that explains half the plan going to one person, a period when it had a 12-month cash runway warning (claim #15462).

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459; what has not been said is where the machines are located, how many there are, the resulting hashrate or megawatt capacity, and who the power or hosting counterparty is.

The company disclosed on 2018-08-13 substantial doubt about covering the next 12 months of operations, cited as claim #15462; what has not been said is how large the funding shortfall is, which financing was planned to close it, and on what terms and timeline.

The company disclosed an Interim Chief Executive Officer on 2018-04-01 (claim #16644) and director resignations on 2018-06-28 (claim #15476), followed by Nasdaq governance deadlines (claims #15465, #15466, #15467, #15468) and a claim of regained compliance after Michael Berg and Fred Thiel joined (claims #15435, #15437, #15438); what has not been said is who the interim CEO and departing directors were, why they left, and whether the separate compensation committee deadline running to 2019-06-28 was ever met.

The company disclosed licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and a string of patent portfolio acquisitions on 2014-03-30 (claims #21016 through #21035), with IP Navigation Group named as a key supplier (claim #21024); what has not been said is the purchase price of each portfolio, the license revenue from each counterparty, and the share of recoveries owed to IP Navigation Group or to the original patent sellers.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claim #15461 (with the termination itself at claim #15477 on 2018-06-28). What has not been said is who received those shares, what they were worth at issuance, how the charge was booked, and whether any GBV principals had prior dealings with management or directors.

The company disclosed Merrick Okamoto's appointment as CEO and Chairman on 2018-10-15 with a $350,000 base salary and option grants of 5,000,000 shares, 2,500,000 shares, and 2,500,000 unvested shares with change-of-control acceleration, cited as claims #15384, #15388, #15389, #15385 and #15386. What has not been said is whether these figures overlap or add up to a single 5,000,000-share grant, and what the strike price, vesting schedule and acceleration trigger are. It also has not said whether Okamoto had any prior investor, lender or consulting relationship with the company that would be related-party context for the proxy.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same period it flagged doubt about a 12-month cash runway, cited as claim #15462. What has not been said is the miner count, hash rate, site location, hosting counterparty, power cost per kWh, and how the expansion was to be financed given the going-concern language. A first-year analyst would ask how many of the 10,000,000 shares reserved under the 2018 Equity Incentive Plan (claim #15460) were already committed.

The company disclosed a string of patent portfolio acquisitions on 2014-03-30, cited as claims #15016–#21023, #21029, #21031 and #21035, which drew on Mosaid, Intergraph, Delphi, TeleCommunication Systems, Siemens, Zhone and CyberFone. It also disclosed licensing agreements with Zhone and Siemens (claims #21022, #21018) and a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is the purchase consideration and contingent revenue-share terms for each portfolio, the dollar value of the licenses, and how much of patent revenue (claim #21014) was paid out to IP Navigation Group or the original patent owners.

The company disclosed an Interim Chief Executive Officer on 2018-04-01 and a 10-K filing deadline of 2018-04-15, cited as claims #16644 and #16645. What has not been said, at least in this record, is who the interim CEO was, what they were paid, and whether the 10-K was filed on time or under an NT 10-K extension. A related open item is whether the compensation committee compliance deadline of 2019-06-28 (claim #15468) was met once Fred Thiel and Michael Berg were appointed (claims #15437, #15435, #15438).

The record attributes to this company the claims of operations in 26 countries, a Mara Foundation in four African countries, and a leadership slate of Jagdish Thakkar, Ashish J. Thakkar, Prashant Manek and Marwan Abedin, cited as claims #20897–#20903, along with UN Women and EY partnerships (claims #20896, #20895). None of this matches Marathon Patent Group's own filings, and it appears to describe the unrelated Mara Group. What has not been established is whether any of these claims belong to MARA at all, and they should be verified before publication.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures had been terminated, cited as claim #15477, and on 2018-08-13 that the termination carried a 3,000,000-share break-up fee, cited as claim #15461. What has not been said is why the deal was terminated, which party triggered the fee, who received the shares, and what value the company assigned to them at issuance.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion, cited as claim #15459, alongside substantial doubt about having 12 months of cash runway, cited as claim #15462. What has not been said is the expansion's scale and location: no miner count, hash rate target, hosting site or address, power contract or cost per kWh, or the financing source expected to fund it given the runway warning.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options and a $350,000 base salary as Executive Chairman and CEO, cited as claims #15389 and #15388, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claim #15460. What has not been said is the strike price, the vesting schedule, why one executive received half the plan, and which holder owns the 2,500,000 unvested options carrying change-of-control acceleration, cited as claim #15386.

The company disclosed on 2018-06-28 that directors had resigned, cited as claim #15476, and on 2018-08-21 that Michael Berg and Fred Thiel had joined the board and that NASDAQ compliance had been regained, cited as claims #15435, #15437 and #15438. What has not been said is which directors resigned, whether any disagreement with management was involved, and whether the separate compensation-committee deadline of 2019-06-28, cited as claim #15468, was formally cleared or folded into the August 2018 compliance notice.

The company disclosed on 2014-03-30 that patent revenue had begun, cited as claim #21014, that it had licensing agreements with Siemens Technology and Zhone, cited as claims #21018 and #21022, and that IP Navigation Group was a key supplier, cited as claim #21024. What has not been said is the consideration paid for each acquired portfolio, the revenue-sharing terms owed to the original patent sellers and to IP Navigation Group, and how much of reported revenue those payments consumed.

The company disclosed on 2013-02-04 that it was exploring a sale of its real estate business, cited as claim #21044, and on 2018-04-01 that an interim CEO was in place ahead of a 2018-04-15 10-K deadline, cited as claims #16644 and #16645. What has not been said is whether the real estate business was ever sold and at what price, who the interim CEO was, and whether the 2018 annual report was filed on time or required an extension.

The company disclosed a key supplier relationship with IP Navigation Group on 2014-03-30, cited as claim #21024, alongside the start of patent revenue (claim #21014) and licensing agreements with Siemens Technology (claim #21018) and Zhone (claim #21022). What has not been said is how revenue is split. What percentage of licensing proceeds goes to IP Navigation Group, and what upfront or running royalty did each named licensee pay?

The company disclosed a run of patent portfolio acquisitions on 2014-03-30, cited as claims #21016/#21029 (Mosaid), #21019 (TeleCommunication Systems), #21020 (Intergraph), #21023 (Delphi), #21017 (CyberFone) and #21021 (TT IP). What has not been said is the price of each deal, whether it was paid in cash, stock or contingent seller revenue-share, and how many patents each portfolio contained. Which of these portfolios are still in force or under active enforcement (claim #21025)?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477, following director resignations on 2018-06-28 (claim #15476). What has not been said is who received the break-up shares and at what valuation, why the merger failed, and which directors resigned and for what stated reasons. Were the resignations connected to the termination?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459. What has not been said is the operating hashrate, the number and model of miners, the site location, and the power or hosting counterparty. What does the expansion amount to in terahashes per second, at which facility, and at what cost per kilowatt-hour?

The company disclosed on 2018-10-15 that Merrick Okamoto, CEO and Chairman, received 2.5 million stock options with change-of-control acceleration, cited as claims #15384, #15385 and #15386. What has not been said is the strike price, the vesting schedule, and the precise acceleration trigger, nor the identity of the interim CEO disclosed on 2018-04-01 (claim #16644). Who was the interim CEO, and how do these grants compare with the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460)?

The company disclosed uncertainty about its 12-month cash runway on 2018-08-13, cited as claim #15462, while also flagging a 10-K filing deadline of 2018-04-15 (claim #16645). What has not been said is the cash balance, the monthly burn rate, or the specific financing plan behind the going-concern language. Was the 10-K filed on time, and what capital was raised in the following twelve months to resolve the runway concern?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claim #15461, after first reporting the termination on 2018-06-28 (claim #15477) alongside unnamed director resignations (claim #15476). What has not been said is which side paid the fee, what the shares were worth on the termination date, how the issuance was accounted for, and which directors resigned and whether their departures were tied to the failed deal.

The company disclosed on 2018-10-15 option grants to Merrick Okamoto described as 2.5 million (claims #15385 and #15386) and as 5 million (claim #15389), against a 10,000,000-share 2018 Equity Incentive Plan reserve, cited as claim #15460. What has not been said is whether these are two 2.5 million tranches or one grant described inconsistently, what the strike price and vesting schedule are, and why roughly half of the entire plan reserve went to a single executive whose base salary was disclosed as $350,000 (claim #15388).

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459. What has not been said is where the miners are physically sited, who hosts them or supplies power, what the installed hashrate, machine count, or megawatt capacity is, and on what timetable the expansion is meant to be complete.

The company disclosed on 2018-08-13 substantial doubt about its 12-month cash runway, cited as claim #15462, in the same filing that announced the mining expansion (claim #15459). What has not been said is the monthly cash burn, the cash balance the runway is measured from, and what financing, whether equity, debt, or equipment leasing, is planned to fund both continued operations and the new mining capex.

The company disclosed on 2014-03-30 licensing agreements with Siemens Technology and Zhone (claims #21018 and #21022), a key supplier relationship with IP Navigation Group (claim #21024), and roughly a dozen patent portfolio acquisitions (claims #21016 through #21035), along with an employee count whose figure is not recorded (claim #21026). What has not been said is the purchase price or revenue-share terms for each portfolio, the license fees received from Siemens and Zhone, the contingency percentage owed to IP Navigation Group, and the actual headcount behind the litigation program.

The company disclosed on 2018-04-01 an unnamed Interim Chief Executive Officer (claim #16644) and a 10-K filing deadline (claim #16645), then on 2018-07-12 NASDAQ deficiencies requiring two new independent directors and a compliance plan (claims #15465 through #15468), and announced regained compliance on 2018-08-21 (claim #15438). What has not been said is who served as interim CEO and on what terms, whether the 10-K was filed on time, and whether NASDAQ's compliance letter covered the compensation committee deficiency (claim #15468) or only the board-independence requirement. Separately, claims #20897 through #20903, covering the Mara Foundation, Ashish Thakkar, and the 26-country African footprint, appear to describe the unrelated Mara Group and should not be read as MARA disclosures.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claim #15461, following the termination noted on 2018-06-28 in claim #15477, the same date several directors resigned (claim #15476). What has not been said is who paid the fee to whom, what the shares were worth on the date of issuance, why the deal collapsed, and whether the director departures were connected to the termination.

The company disclosed on 2018-10-15 a 5,000,000-share option grant to Merrick Okamoto alongside a $350,000 base salary, cited as claims #15389 and #15388, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares disclosed on 2018-08-13 in claim #15460. What has not been said is the strike price, the vesting schedule, and any performance conditions on a single grant that consumes half the plan. Also missing is any related-party context for Okamoto, such as prior shareholdings, loans to the company, or financing relationships, that would explain a grant of that size.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same period it flagged uncertainty about a 12-month cash runway (claim #15462). What has not been said is the scale of the expansion in machines or hashrate, the physical site and power arrangement including the location, the utility or host, and the cost per kWh, and how the expansion was to be funded given the going-concern language.

The company disclosed on 2018-07-12 a series of NASDAQ compliance deadlines and a commitment to add two independent directors, cited as claims #15465 through #15468, then announced on 2018-08-21 that compliance was regained with Michael Berg as audit chair and Fred Thiel as compensation committee chair (claims #15435, #15437, #15438). Separately, an interim CEO and a 10-K filing deadline were disclosed on 2018-04-01 (claims #16644 and #16645). What has not been said is the identity of the interim CEO and the reason for the filing pressure, how the new directors were sourced, and whether Thiel's 50,000-share option grant on 2018-10-15 (claim #15396) was set by the committee he chairs.

The company disclosed on 2014-03-30 a string of patent portfolio acquisitions from Siemens, Zhone, Mosaid, Delphi, Intergraph, TeleCommunication Systems, CyberFone and TT IP, the commencement of patent revenue, active enforcement litigation, and IP Navigation Group as a key supplier, cited as claims #21013 through #21035. What has not been said is the purchase price or revenue-share owed to each seller, the economics of the IP Navigation relationship, and which licensees or defendants actually generated the revenue. Separately, the real estate disposition exploration flagged on 2013-02-04 (claim #21044) was never reported as resolved.

Claims #20895 through #20903, covering Mara Group, the Thakkar family, the Mara Foundation, the UN Women and EY partnerships, and the African country footprint, appear to describe a different, unrelated entity and were not used as a basis for any gap above.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is which party bore the fee, what the shares were worth at issuance, and why a deal central to the cryptocurrency pivot was abandoned only weeks before the board turnover recorded in claim #15476.

The company disclosed on 2018-10-15 that Merrick Okamoto received stock options totaling 5,000,000 shares, including 2,500,000 unvested options that accelerate on a change of control, cited as claims #15385, #15386 and #15389; what has not been said is the strike price, the vesting schedule or performance hurdles for the unvested tranche, and why a single executive received half of the 10,000,000-share 2018 Equity Incentive Plan reserve disclosed in claim #15460.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion alongside substantial doubt about its 12-month cash runway, cited as claims #15459 and #15462; what has not been said is how the expansion would be funded, how many miners and how much hashrate were planned, and where they would be hosted, including the site, power capacity in megawatts, and power cost.

The company disclosed on 2018-07-12 a Nasdaq compliance plan and a commitment to add two independent directors by 2019-01-07, then on 2018-08-21 said compliance had been regained, with Fred Thiel named Compensation Committee Chairman and Michael Berg named Audit Committee Chairman, cited as claims #15465, #15466, #15438, #15437 and #15435; what has not been said is whether the separate compensation committee deadline of 2019-06-28 in claim #15468 was met, and whether the newly formed committee approved the Okamoto option grants of claim #15389 or merely ratified them.

The company disclosed on 2014-03-30 that it had begun generating patent revenue, held licensing agreements with Zhone and Siemens Technology, and relied on IP Navigation Group as a key supplier, cited as claims #21014, #21022, #21018 and #21024; what has not been said is the purchase price of each acquired portfolio, the revenue attributable to each licensee, and the fee or revenue split owed to IP Navigation Group on enforcement proceeds.

The company disclosed on 2018-04-01 an interim CEO and a 10-K filing deadline of 2018-04-15, cited as claims #16644 and #16645; what has not been said is who the interim CEO was, what that person was paid, and whether the annual report was filed on time or under an extension.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares (claim #15461), after first reporting the termination on 2018-06-28 (claim #15477). What has not been said is which side issued those shares to whom, what they were worth on the termination date, why the deal collapsed, and whether any Global Bit Ventures principals had prior or ongoing relationships with Marathon insiders.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options (claim #15389). In the same filing it described a 2,500,000-option grant to the CEO (claim #15385), against a 2018 Equity Incentive Plan reserve of only 10,000,000 shares (claim #15460). What has not been said is how the two option figures reconcile, what the strike price and vesting schedule are, and whether a single executive received a quarter to half of the entire plan reserve in one grant.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459). On the same date it flagged substantial doubt about its 12-month cash runway (claim #15462). What has not been said is the hash rate, miner count, or megawatt capacity of the expansion, the physical location and power arrangement of the mining site, and how an expansion is to be funded when the company itself questions whether it can operate for a year.

The company disclosed on 2018-06-28 that directors had resigned (claim #15476), which triggered NASDAQ board and compensation committee compliance deadlines (claims #15465, #15467, #15468). It then appointed Fred Thiel as Compensation Committee Chairman on 2018-08-21 (claim #15437), and Thiel received 50,000 options on 2018-10-15 (claim #15396). What has not been said is who resigned and why, and whether the chair of the committee approving the CEO's outsized grant (claim #15389) was granted options in the same action.

The company disclosed on 2014-03-30 licensing agreements with Siemens Technology and Zhone (claims #21018, #21022), alongside a key supplier relationship with IP Navigation Group (claim #21024) and active patent enforcement litigation (claim #21025). What has not been said is the economics of these arrangements: what share of recoveries IP Navigation Group retains, what upfront or contingent consideration was paid for each acquired portfolio (claims #21017 to #21035), and how many employees the company actually had (claim #21026 records the topic but no figure).

The company disclosed on 2018-04-01 an Interim Chief Executive Officer (claim #16644) and a 10-K filing deadline of 2018-04-15 (claim #16645). What has not been said is who the interim CEO was, why the prior CEO departed, and whether that 10-K was filed on time or late. Separately, claims #20895 through #20903 (Ashish Thakkar, the Mara Foundation, and operations in 26 countries) appear to describe the unrelated African conglomerate Mara Group rather than Marathon, and should not be read as Marathon disclosures.

Merrick Okamoto's October 15, 2018 option grants are disclosed inconsistently (claims #15385, #15386 and #15389). One claim gives 2.5 million options issued, another 2.5 million unvested options with change-of-control acceleration, and another 5 million options in total. Alongside the $350,000 base salary (claim #15388), what has not been said is the strike price, the vesting schedule, any performance conditions, or a bonus target. Is the 5 million figure one grant split into vested and unvested tranches, or two separate awards?

The 2018 Equity Incentive Plan reserved 10,000,000 shares (claim #15460). Okamoto's grant alone appears to use half of that pool, and David Lieberman received another 200,000 options (claim #15392). What has not been said is how many shares were outstanding at the time, so readers cannot gauge dilution or how much of the pool was left for other employees.

The Global Bit Ventures merger was terminated with a break-up fee paid in shares, given as 3,000,000 shares (claims #15477 and #15461). Directors resigned on the same date (claim #15476). What has not been said is why the deal ended, who received the 3 million shares and what they were worth, which directors left and whether their departures were linked to the termination, and whether any GBV mining equipment or hosting arrangements carried over into the company's own business.

The company announced a cryptocurrency mining expansion on August 13, 2018 (claim #15459). In the same filing it flagged uncertainty about its 12-month cash runway (claim #15462). What has not been said is how many machines it operated or planned to add, the hashrate targeted, the physical site and power contract, the price per kWh, and how much financing the expansion required. Without those figures, the runway warning cannot be connected to the growth plan.

Nasdaq compliance deadlines were set on July 12, 2018: a compliance plan by August 24, 2018, two new independent directors by January 7, 2019, and the compensation committee by June 28, 2019 (claims #15466, #15465, #15467 and #15468). Compliance was then declared regained on August 21, 2018, when Michael Berg and Fred Thiel were appointed (claims #15435, #15437 and #15438). What has not been said is which deficiency notice started the process, whether a compliance plan was ever formally submitted, and whether the compensation committee met the independence and size requirements or only got a chair.

In the patent-monetization era, the company named IP Navigation Group as a key supplier (claim #21024) and disclosed licensing agreements with Siemens and Zhone (claims #21018 and #21022). It also listed portfolio purchases from Mosaid, CyberFone, Delphi, Intergraph and others (claims #21016, #21017, #21023 and #21020), and said patent revenue had started (claim #21014). What has not been said is the fee or revenue-share terms with IP Navigation, the consideration under each license, any contingent payments owed to the original patent sellers, or how revenue was split by portfolio.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461); what has not been said is why the deal collapsed, which side paid the fee to whom, what the shares were valued at for accounting purposes, and whether any GBV mining equipment, hosting contracts or personnel transferred anyway.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459), alongside a going-concern style warning that cash may not last twelve months (claim #15462); what has not been said is where the mining fleet physically sits, how many machines and what hash rate are deployed, what power price and hosting or interconnection terms apply, and how the expansion will be financed given the stated runway doubt.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options and a $350,000 base salary (claims #15389 and #15388), against a 2018 Equity Incentive Plan reserve of 10,000,000 shares disclosed on 2018-08-13 (claim #15460); what has not been said is the strike price, vesting schedule and performance conditions on a grant consuming half the plan, and whether Okamoto or affiliated entities have any financing, consulting or trading relationship with the company. The interim CEO disclosed on 2018-04-01 (claim #16644) is also not named in the record.

The company disclosed licensing agreements with Siemens Technology and Zhone on 2014-03-30 (claims #21018 and #21022), the start of patent revenue (claim #21014), and a key supplier relationship with IP Navigation Group (claim #21024); what has not been said is the dollar value of each license, what share of recoveries is owed to IP Navigation Group or to the original patent sellers under contingent-payment terms, and how much revenue each acquired portfolio has actually produced against its purchase cost.

The company disclosed on 2018-07-12 a Nasdaq compliance plan requiring two new independent directors and a compensation committee fix by 2019-06-28 (claims #15465 and #15468), after unexplained director resignations on 2018-06-28 (claim #15476), and then said on 2018-08-21 that compliance was regained with Michael Berg and Fred Thiel appointed (claims #15435, #15437, #15438); what has not been said is why the prior directors resigned, whether any disagreement was cited, and whether the 2018 10-K deadline of 2018-04-15 (claim #16645) was actually met or required an extension.

The company disclosed on 2013-02-04 a 2-for-3 reverse split authorization and an exploration of disposing of its real estate business (claims #21046 and #21044); what has not been said is whether the split was executed and on what date, and whether the real estate assets were sold, to whom, and at what price relative to book value.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 that the termination carried a 3,000,000-share break-up fee (claim #15461). What has not been said is why the deal was terminated, which party received the shares and at what per-share value, and how large an expense or dilution charge was booked for a company that was flagging cash-runway doubts in the same filing (claim #15462).

The company disclosed on 2018-06-28 that directors had resigned (claim #15476), and on 2018-07-12 committed to appoint two new independent directors and submit a NASDAQ compliance plan (claims #15465, #15466, #15467). It then announced on 2018-08-21 that compliance was regained, with Michael Berg chairing Audit and Fred Thiel chairing Compensation (claims #15438, #15435, #15437). What has not been said is which directors resigned and why, whether the separate compensation-committee requirement due 2019-06-28 (claim #15468) was cured by the same appointments, and whether NASDAQ formally closed the matter.

The company disclosed on 2018-10-15 that CEO and Chairman Merrick Okamoto received 2.5 million options (claims #15384, #15385), CFO David Lieberman received 200,000 options and a $180,000 salary (claims #15390, #15391, #15392), and Compensation Committee chair Fred Thiel himself received 50,000 options (claim #15396). What has not been said is the strike price, vesting schedule and performance conditions on these grants, Okamoto's cash compensation, and how the committee handled Thiel approving a grant to himself. Together these grants consume roughly 29% of the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460) within months of its adoption.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that raised substantial doubt about a 12-month cash runway (claim #15462). What has not been said is the site location, power capacity in megawatts, utility or interconnection arrangement, miner count, hash-rate target, or how the expansion would be financed given the going-concern language.

The company disclosed an interim Chief Executive Officer on 2018-04-01 (claim #16644) alongside a 10-K deadline of 2018-04-15 (claim #16645), whose status is still not verifiable. What has not been said is the interim CEO's name and compensation, and whether the annual report was filed on time or under an NT 10-K extension.

The company disclosed on 2014-03-30 that patent revenue had commenced (claim #21014), with licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and IP Navigation Group named as a key supplier (claim #21024). What has not been said is the dollar value of each license, which licensees accounted for the reported revenue, and the contingency-fee split owed to IP Navigation Group on monetization proceeds.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is who triggered the termination, what dollar value and accounting treatment the company assigned to those shares, and whether they were issued by Marathon or paid to it. With 10,000,000 shares reserved under the 2018 Equity Incentive Plan (claim #15460), how material was this fee relative to the share count?

The company disclosed on 2018-10-15 that CEO and Chairman Merrick Okamoto received 2.5 million stock options and that CFO David Lieberman received 200,000 options plus a $180,000 base salary, cited as claims #15384, #15385, #15390, #15391 and #15392; what has not been said is the strike price, vesting schedule and performance conditions on these grants, or the CEO's cash compensation. The CEO grant alone uses a quarter of the 10,000,000-share plan reserve (claim #15460). Why was it sized that way, and did the compensation committee chaired by Fred Thiel (claim #15437), who also received options (claim #15396), approve it independently?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, alongside uncertainty about its 12-month cash runway (claim #15462); what has not been said is the expansion's capacity in any unit (miners, megawatts or hashrate), the site location, the power contract or interconnection terms, and how the expansion would be financed given the going-concern language. What capital was committed, and on what timeline?

The company disclosed on 2014-03-30 that patent revenue had commenced, with licensing agreements with Siemens Technology and Zhone and a key supplier relationship with IP Navigation Group, cited as claims #21014, #21018, #21022 and #21024; what has not been said is the license fees or royalty terms, which portfolio produced the revenue, and the contingency-fee or revenue-share economics owed to IP Navigation Group. Across the many portfolio acquisitions (claims #21016 through #21035), what purchase prices and earn-outs were paid to sellers such as Mosaid, Intergraph, TeleCommunication Systems and Delphi?

The company disclosed NASDAQ deficiency deadlines on 2018-07-12, including two new independent directors and compensation committee compliance by 2019-06-28, cited as claims #15465, #15466, #15467 and #15468, and then reported regained compliance on 2018-08-21 (claim #15438); what has not been said is which specific listing rule was breached and whether the compensation committee requirement was separately cured. Relatedly, the 2018-04-01 disclosure of an unnamed interim CEO and a 10-K deadline of 2018-04-15 (claims #16644 and #16645) leaves open who held the role and whether the annual report was filed on time. Was the 2-for-3 reverse split authorized on 2013-02-04 (claim #21046) ever actually effected, and at what ratio?

Claims #20895 through #20903 (Ashish Thakkar, Mara Foundation, operations in 26 countries, UN Women and EY partnerships) appear to describe Mara Group, an unrelated African conglomerate, rather than Marathon. They should be verified or removed before any gap analysis relies on them.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is which party paid the fee to whom, what per-share value and total dollar amount were booked, and whether any Marathon insiders held interests in Global Bit Ventures that would make this a related-party settlement.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options alongside a $350,000 base salary, cited as claims #15389 and #15388, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares disclosed on 2018-08-13, cited as claim #15460; what has not been said is the strike price, vesting schedule and any performance conditions on a single grant that consumes half the plan, or how it compares with the 200,000, 100,000 and 50,000-share grants to Lieberman, Crawford, Thiel and Rudolph (claims #15392 through #15396).

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459; what has not been said is the hosting site's location, the number and model of machines, the targeted hashrate in specific units, the power counterparty and contracted price per kWh, or the timeline for deploying the capacity.

The company disclosed on 2018-08-13 substantial doubt about having 12 months of cash runway through 2019-08-13, cited as claim #15462, in the same filing that announced the mining expansion (claim #15459); what has not been said is the cash balance at that date, the monthly burn rate, and which specific equity or debt financing, with what size and timing, was expected to fund both operations and the expansion.

The company disclosed licensing agreements with Siemens Technology and Zhone and the start of patent revenue on 2014-03-30, cited as claims #21018, #21022 and #21014, along with a key supplier relationship with IP Navigation Group, cited as claim #21024; what has not been said is the dollar value of each license, the share of revenue paid to IP Navigation Group or to the original patent sellers under revenue-sharing terms, and how much of the acquired portfolios (claims #21016 through #21035) had generated any licensing income at all.

The company disclosed a 10-K filing deadline of 2018-04-15 and an interim CEO on 2018-04-01, cited as claims #16645 and #16644; what has not been said is whether that 10-K was filed on time or late, who the interim CEO was and on what compensation terms, and why it was necessary. Note also that claims #20895 through #20903, covering the Mara Foundation, the Thakkar family and operations in 19 to 26 countries, appear to describe the unrelated Mara Group rather than Marathon and were not used to support any gap above.

The company disclosed a string of patent portfolio acquisitions on 2014-03-30, from CyberFone Systems, TeleCommunication Systems, Intergraph, Delphi Technologies, Mosaid and TT IP, cited as claims #21017, #21019, #21020, #21023, #21016 and #21021. What has not been said is what was paid for each portfolio, or whether the sellers keep revenue-share or back-end participation rights in later licensing or litigation proceeds. What did each portfolio cost, and how much of any recovery goes back to the original owner?

The company disclosed licensing agreements with Siemens Technology and Zhone and the start of patent revenue generation on 2014-03-30, cited as claims #21018, #21022 and #21014. What has not been said is the dollar value, the term, or whether each license was a lump sum or recurring royalty. Nor has it said how concentrated early revenue was in these two counterparties. How much did each license contribute, and is any of it recurring?

The company named IP Navigation Group as a key supplier on 2014-03-30, cited as claim #21024, while also disclosing active enforcement litigation (claim #21025). What has not been said is the fee structure, such as the contingency percentage or fixed retainer, or how many active cases depend on that one relationship. What share of gross recoveries goes to IP Navigation, and what happens to the pipeline if the relationship ends?

The company disclosed that the Global Bit Ventures merger was terminated on 2018-06-28, alongside director resignations, then a 3,000,000-share break-up fee on 2018-08-13, cited as claims #15477, #15476 and #15461. What has not been said is which party triggered the termination and why, which directors resigned and whether they cited disagreements, or what the fee shares were worth on issuance. Who received the 3 million shares, at what implied value, and were the resignations tied to the failed deal?

The company disclosed CEO option grants to Merrick Okamoto on 2018-10-15 as both 2,500,000 and 5,000,000 shares, with change-of-control acceleration on 2.5 million unvested options, cited as claims #15385, #15389 and #15386, against a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460). What has not been said is how the two grant figures reconcile, along with the strike price, the vesting schedule and any performance conditions. How much of the plan reserve went to a single executive, at what exercise price, and what event triggers acceleration?

The company disclosed a cryptocurrency mining expansion on 2018-08-13 while flagging uncertainty about a 12-month cash runway, cited as claims #15459 and #15462. What has not been said is the number of machines, hash rate, site location, power contract or cost per kWh, or how the expansion would be funded. Where are the miners hosted, what is the all-in cost to mine a coin, and how will the expansion be financed if the runway is uncertain?

On 2018-10-15 the company disclosed option grants to Merrick Okamoto, cited as claims #15385, #15386 and #15389. One records 2.5 million options, another 2.5 million unvested options with change-of-control acceleration, and a third 5 million options. What has not been said is how these figures reconcile, what the strike prices and vesting schedules are, and what share of the 10 million-share 2018 Equity Incentive Plan reserve (claim #15460) went to a single executive within two months of the plan's creation.

The company disclosed on 2018-08-21 that Fred Thiel would chair the Compensation Committee (claim #15437), and on 2018-10-15 that Thiel received 50,000 options (claim #15396) in the same grant cycle as the CEO's award. What has not been said is whether the committee used an independent compensation consultant, what peer benchmarking supported the CEO grant, and whether the director grants were approved by the same committee that benefited from them.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated and that directors resigned (claims #15477 and #15476), and on 2018-08-13 that the termination carried a 3,000,000-share break-up fee (claim #15461). What has not been said is which directors resigned and whether any were affiliated with Global Bit Ventures, the fair value and accounting charge for the fee shares, why a terminating party paid in equity rather than cash, and whether any mining equipment or hosting arrangements changed hands with Global Bit Ventures despite the termination.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that flagged uncertainty over a 12-month cash runway (claim #15462). What has not been said is the installed machine count and hashrate, the physical location of the mining facility, the power contract terms and cost per kWh, and how the expansion would be financed given the going-concern language. A first-year analyst would also ask whether the runway concern was resolved by the 2019-08-13 horizon or carried into later filings.

The company disclosed Nasdaq compliance deadlines on 2018-07-12 for board independence, a compliance plan, and compensation committee composition (claims #15465 through #15468), then said on 2018-08-21 that compliance was regained (claim #15438) after appointing Michael Berg as Audit Committee Chair (claim #15435). What has not been said is whether the separate 2019-06-28 compensation-committee deadline (claim #15468) was satisfied on time or required a further cure period. Earlier, on 2018-04-01, the company disclosed an interim CEO (claim #16644) alongside an April 15 10-K deadline (claim #16645). What has not been said is the interim CEO's identity and compensation, or whether that 10-K was filed late.

In its 2014-03-30 annual filing the company disclosed licensing agreements with Siemens and Zhone (claims #21018 and #21022), the commencement of patent revenue (claim #21014), and a key supplier relationship with IP Navigation Group (claim #21024). It also disclosed portfolio acquisitions from Mosaid, CyberFone, TeleCommunication Systems, Delphi and Intergraph (claims #21016, #21017, #21019, #21023 and #21020). What has not been said is the dollar value of each license, the purchase price or contingent revenue-share owed to each seller, and the fee split retained by IP Navigation Group. Without those figures a reader cannot assess whether the monetization model was profitable before the pivot to mining.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 that the termination carried a break-up fee of 3,000,000 shares (claim #15461). What has not been said is who received those shares and why, what they were worth on the issue date, why the deal was terminated, and whether any director or officer had an interest in Global Bit Ventures.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO (claims #15387, #15384) and received stock options, but the records conflict. Claim #15385 says 2.5 million options, claim #15389 says 5,000,000, and claim #15386 describes 2.5 million unvested options with change-of-control acceleration. What has not been said is the CEO's actual total grant, its exercise price, vesting schedule and acceleration triggers. It is also unstated how a grant of that size fits within the 10,000,000-share reserve of the 2018 Equity Incentive Plan (claim #15460). Taken at face value, one executive would hold 25-50% of the pool.

The same 2018-10-15 filing shows 50,000-share option grants to Fred Thiel (claim #15396) and Michael Rudolph (claim #15394). Thiel had been named Compensation Committee Chairman on 2018-08-21 (claim #15437). What has not been said is who approved these grants, including the CEO's, and whether the compensation committee chair abstained from approving awards to himself. The company also has not said how the committee met the compliance deadline it was under (claim #15468).

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459). In the same filing it flagged uncertainty about having 12 months of cash runway (claim #15462). What has not been said is where the mining would physically take place, how much capacity it would have (machine count, hash rate or megawatts), and who supplies the power and on what terms. Above all, it is unstated how a company doubting its own one-year runway intended to fund the expansion.

The company disclosed on 2018-04-01 that it had an Interim Chief Executive Officer (claim #16644) and faced a 10-K deadline of 2018-04-15 (claim #16645). It then disclosed on 2018-06-28 that directors had resigned (claim #15476). What has not been said is who the interim CEO and the departing directors were, whether the 10-K was filed on time, and whether any of the resignations involved a disagreement with management. It is also unstated whether the departures caused the NASDAQ deficiencies behind claims #15465-#15467.

In its 2014-03-30 filing the company disclosed licensing agreements with Siemens Technology (claim #21018) and Zhone (claim #21022), and named IP Navigation Group as a key supplier (claim #21024). It also reported an employee count (claim #21026) without a figure. What has not been said is how much revenue each license produced, and what share of patent recoveries went to IP Navigation Group under that relationship. The actual headcount behind a business running more than ten portfolio acquisitions (claims #21015-#21035) is also missing.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), the same day directors resigned (claim #15476), and on 2018-08-13 described a share-based break-up fee of 3,000,000 shares (claim #15461). What has not been said is which party paid the fee, how the shares were valued and accounted for, why the deal collapsed, and whether the board departures were connected to the termination. What did the break-up cost shareholders in dilution and dollars?

The company disclosed on 2018-10-15 that Merrick Okamoto became Executive Chairman and CEO at a $350,000 base salary with 5,000,000 stock options (claims #15387, #15388, #15389). It split these into 2,500,000 issued options and 2,500,000 unvested options with change-of-control acceleration (claims #15385, #15386). What has not been said is the strike price, the vesting schedule, and the performance conditions. The record also omits why roughly half of the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460) went to a single executive, and whether Mr. Okamoto had any prior financing or related-party relationship with the company. Was the board's compensation committee, chaired by Fred Thiel from 2018-08-21 (claim #15437), independent in approving this package?

The company disclosed on 2018-08-13 a cryptocurrency mining expansion (claim #15459), and in the same filing flagged uncertainty about a 12-month cash runway (claim #15462). What has not been said is where the miners are physically sited, the power capacity in megawatts, the number and model of machines, any hosting counterparty, and how the expansion would be financed given the going-concern-style warning. What were the hash rate target, the power contract terms, and the funding source?

The company disclosed on 2018-07-12 that it needed to submit a NASDAQ compliance plan by 2018-08-24 (claim #15466) and appoint two new independent directors by 2019-01-07 (claim #15465). It also faced a compensation committee deadline of 2019-06-28 (claim #15468), and it announced on 2018-08-21 that compliance had been regained (claim #15438). The record shows only Michael Berg as a new board appointee (claim #15435). What has not been said is who the second new independent director was, and whether the compensation committee requirement was separately satisfied by its deadline.

The company disclosed on 2014-03-30 that IP Navigation Group was a key supplier (claim #15024 is not in the record; see claim #21024), alongside licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and the commencement of patent revenue (claim #21014). What has not been said is the economic split owed to IP Navigation on enforcement recoveries, the dollar value and duration of the Siemens and Zhone licenses, and how much of total revenue each counterparty represented. The disclosed employee count (claim #21026) carries no figure in the record. How concentrated was licensing revenue, and what did the company keep after contingency fees?

The company disclosed on 2018-04-01 an Interim Chief Executive Officer (claim #16644) and a 10-K filing deadline of 2018-04-15 (claim #16645). What has not been said is who the interim CEO was, why the prior CEO departed, and whether the annual report was filed on time or under an extension. Was there any auditor or internal-control issue behind the timing?

The claims dated 2013-04-04 (claims #20897 through #20903, including Ashish J. Thakkar as founder, the Mara Foundation, and operations in 26 countries) and the 2014 UN Women and EY partnerships (claims #20895, #20896) appear to describe the African conglomerate Mara Group rather than Marathon Patent Group or Marathon Digital. They should be reviewed for misattribution before any gap analysis relies on them. The claim list provided was also truncated after claim #15390, so later disclosures may close or add gaps not reflected here.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares (claim #15461), following the termination and director resignations announced on 2018-06-28 (claims #15477, #15476). What has not been said is who received those shares, what value the company assigned to them at issuance, why the merger was terminated, and whether the departing directors had any connection to Global Bit Ventures.

The company disclosed CEO Merrick Okamoto's option awards on 2018-10-15 inconsistently: 2.5 million options issued (claim #15385), 2.5 million unvested options with change-of-control acceleration (claim #15386), and a 5,000,000-option total (claim #15389). That total is half of the 10,000,000-share 2018 Equity Incentive Plan reserve disclosed on 2018-08-13 (claim #15460). What has not been said is the strike price, the vesting schedule for the unvested tranche, whether any vesting depends on performance, and why half the plan went to one executive two months after the plan was adopted.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that flagged uncertainty about its 12-month cash runway (claim #15462). What has not been said is how many machines were deployed or ordered, at what hashrate, at which physical site, at what contracted power cost, and how the expansion was to be funded given the going-concern language.

The company disclosed on 2014-03-30 licensing agreements with Zhone and Siemens Technology (claims #21022, #21018), the start of patent revenue generation (claim #21014), and IP Navigation Group as a key supplier (claim #21024). What has not been said is the dollar value of each license, how much of patent revenue came from those two counterparties versus litigation settlements (claim #21025), and what share of recoveries went to IP Navigation Group.

The company disclosed on 2018-07-12 a series of NASDAQ compliance deadlines, including a plan submission due 2018-08-24 and board and compensation-committee cures (claims #15466, #15467, #15468, #15465). It said on 2018-08-21 that compliance had been regained after appointing Michael Berg and Fred Thiel (claims #15438, #15435, #15437). What has not been said is which specific listing rule was breached, how the two new directors were sourced, and whether either had prior business ties to management. The interim CEO disclosed on 2018-04-01 (claim #16644) was also never named on the record, and there is no confirmation that the 10-K due 2018-04-15 (claim #16645) was filed on time.

The company disclosed on 2013-02-04 a 2-for-3 reverse split authorization and an exploration of disposing of its real estate business (claims #21046, #21044). What has not been said is whether the split was ever effected at that unusual ratio, and whether the real estate assets were sold, to whom, and at what price or gain/loss.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461). What has not been said is which side paid the fee, what per-share value and total dollar charge were booked, whether the shares came with registration rights or lock-ups, and why a deal walked away from by the counterparty would still require Marathon to issue equity.

The company disclosed on 2018-10-15 a grant of 5,000,000 stock options to Merrick Okamoto (claim #15389), against a 2018 Equity Incentive Plan reserve of 10,000,000 shares disclosed on 2018-08-13 (claim #15460), plus 2,500,000 unvested options carrying change-of-control acceleration (claim #15386). What has not been said is the strike price, vesting schedule and grant-date fair value of the Okamoto award, whose options the 2.5 million accelerating options are, and how the board justified committing half the plan reserve to one executive while warning of a 12-month cash runway uncertainty (claim #15462).

The company disclosed on 2018-08-13 an expansion of cryptocurrency mining (claim #15459) in the same filing that flagged doubt about funding operations through 2019-08-13 (claim #15462). What has not been said is the number and model of machines, deployed hashrate, the physical hosting site and its power contract or cost per kWh, and what financing, if any, was lined up to pay for the expansion.

The company disclosed on 2018-06-28 that directors had resigned (claim #15476), then on 2018-07-12 committed to appoint two new independent directors by 2019-01-07 (claim #15465) and to meet a compensation-committee compliance deadline of 2019-06-28 (claim #15468), before announcing on 2018-08-21 that NASDAQ compliance was regained with Michael Berg as Audit Chair and Fred Thiel as Compensation Chair (claims #15435, #15437, #15438). What has not been said is why the directors resigned, whether they had disagreements with management, and how independent the compensation committee is when its chair received 50,000 options in the same October 2018 grant cycle (claim #15396) that awarded the CEO 5,000,000.

The company disclosed on 2014-03-30 that IP Navigation Group was a key supplier (claim #21024) and that patent revenue had begun (claim #21014) from portfolios bought from Mosaid, Siemens, Intergraph, TeleCommunication Systems, Delphi and others (claims #21016, #21031, #21020, #21019, #21023). What has not been said is what share of gross licensing and settlement proceeds went to IP Navigation and to the original patent sellers under contingent-payment arrangements, and therefore what Marathon actually kept per dollar of headline revenue.

The company disclosed on 2018-04-01 an Interim Chief Executive Officer (claim #16644) alongside a 10-K deadline of 2018-04-15 (claim #16645), which remains unverified. What has not been said is whether that 10-K was filed on time or under an NT 10-K extension, who the interim CEO was and why the permanent CEO left, and whether the leadership change was tied to the Global Bit Ventures deal later abandoned.

The company disclosed on 2018-10-15 that Merrick Okamoto was named Executive Chairman and CEO with a $350,000 base salary, cited as claims #15387 and #15388, alongside option grants recorded as 2.5 million shares (claim #15385), 5 million shares (claim #15389) and 2.5 million unvested options with change-of-control acceleration (claim #15386). What has not been said is how these figures reconcile: is the total grant 5 million options split into vested and unvested tranches, and what are the strike price, vesting schedule and change-of-control trigger definition?

The company disclosed on 2018-06-28 and 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15477 and #15461, and that directors resigned at the same time, cited as claim #15476. What has not been said is why the merger was terminated, which side triggered the fee, what the 3 million shares were worth when issued, whether they were registered or restricted, and whether the director resignations were connected to the termination.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that flagged substantial doubt about a 12-month cash runway, cited as claim #15462. What has not been said is the scale and location of that expansion: how many machines, what hash rate, which hosting site or power counterparty, what cost per kWh, and how the expansion would be funded given the stated going-concern risk.

The company disclosed on 2018-07-12 a NASDAQ compliance plan with deadlines for two new independent directors and a compensation committee, cited as claims #15465, #15466, #15467 and #15468, and on 2018-08-21 that compliance had been regained with Michael Berg and Fred Thiel appointed as committee chairs, cited as claims #15435, #15437 and #15438. What has not been said is whether the compensation-committee requirement tracked separately to 2019-06-28 was also satisfied by those appointments, and whether the committee that approved Okamoto's October 2018 package (claims #15385 to #15389) was fully independent at the time it acted.

The company disclosed on 2014-03-30 licensing agreements with Siemens Technology and Zhone, cited as claims #21018 and #21022, the start of patent revenue generation, cited as claim #21014, and a key supplier relationship with IP Navigation Group, cited as claim #21024. What has not been said is the dollar value of each license, how much of reported revenue each counterparty contributed, and the fee or contingency split paid to IP Navigation Group out of litigation and licensing proceeds. The same filing references an employee count, cited as claim #21026, without the number appearing in the record.

The company disclosed on 2018-04-01 the appointment of an interim Chief Executive Officer and a 10-K filing deadline of 2018-04-15, cited as claims #16644 and #16645. What has not been said is who the interim CEO was, what precipitated the leadership change, and whether the annual report was filed on time or under an extension.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO at a $350,000 base salary with 5,000,000 stock options, cited as claims #15387, #15388 and #15389. It also disclosed a 2,500,000-option tranche, claim #15385, and 2,500,000 unvested options with change-of-control acceleration, claim #15386. What has not been said is the strike price, the vesting schedule or performance hurdles for the unvested half, any bonus target, and the change-of-control definition that triggers acceleration. Those terms decide whether the grant rewards performance or simply rewards a sale.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477. What has not been said is which party received the shares, what they were worth on the issue date, and why the deal was terminated. That last point matters because directors resigned the same day, claim #15476, and investors need to know whether the two events are connected.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion, claim #15459, alongside an admission that its 12-month cash runway was uncertain, claim #15462. What has not been said is the number of machines, the hashrate in stated units, the site location and power contract (MW, price per kWh, utility counterparty), and how the expansion would be funded given the going-concern language.

The company disclosed on 2018-07-12 Nasdaq deadlines to add two independent directors by 2019-01-07 and to fix its compensation committee by 2019-06-28, cited as claims #15465 and #15468. It then disclosed on 2018-08-21 that compliance had been regained after Michael Berg and Fred Thiel joined, claims #15435, #15437 and #15438. What has not been said is which specific listing-rule deficiency was cured on that date, whether the separate compensation-committee deadline was also satisfied, and whether the 50,000-option grants to Thiel and Michael Rudolph, claims #15396 and #15394, were approved by a fully independent committee.

The company disclosed on 2018-04-01 an interim CEO and a 10-K deadline of 2018-04-15, cited as claims #16644 and #16645. What has not been said is the interim CEO's name and pay, whether the 10-K was filed on time or under an NT 10-K extension, and whether any auditor change or material weakness explained the timing.

In the patent-licensing era, the company disclosed on 2014-03-30 that patent revenue had begun and that IP Navigation Group was a key supplier, cited as claims #21014 and #21024. It also disclosed licensing agreements with Siemens and Zhone, claims #21018 and #21022. What has not been said is the revenue attributable to each licensee, the fee or revenue-share terms paid to IP Navigation Group, and the purchase price for each acquired portfolio.

Two caveats on the record itself. Claims #20895 through #20903 (Ashish and Jagdish Thakkar, 19 African countries, the Mara Foundation, the UN Women and EY partnerships) appear to describe Mara Group, an unrelated African conglomerate, so they were not used. The claim list was also truncated after claim #15396, so later disclosures were not evaluated.

The company announced the termination of its Global Bit Ventures merger on 2018-06-28, cited as claims #15477 and #15461, along with a break-up fee paid in 3,000,000 shares. It disclosed director resignations on the same date, cited as claim #15476. What has not been said is why the deal was terminated, what the fee shares were worth on the issue date and how that charge was booked, and which directors resigned and whether their departures were tied to the failed merger.

On 2018-10-15 the company disclosed a grant of 5,000,000 stock options to Merrick Okamoto and a $350,000 base salary, cited as claims #15389 and #15388. Two months earlier it had reserved 10,000,000 shares under the 2018 Equity Incentive Plan, cited as claim #15460. What has not been said is the strike price, vesting schedule and any performance conditions on a single grant equal to half the plan's reserve. The company also has not said who served as the interim CEO it disclosed on 2018-04-01, cited as claim #16644, or how that person was paid.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that flagged uncertainty about a 12-month cash runway, cited as claim #15462. What has not been said is the scale of the expansion in machines, hash rate or megawatts. Nor has it said where the miners are physically hosted, what it pays per kWh, or how the build-out would be funded given the going-concern language.

On 2014-03-30 the company said patent revenue generation had commenced and that it had licensing agreements with Siemens Technology and Zhone, cited as claims #21014, #21018 and #21022. In the same filing it named IP Navigation Group as a key supplier, cited as claim #21024. What has not been said is how much revenue each license produced, and whether those were one-time or running royalties. It also has not said what share of recoveries went to IP Navigation Group or back to the original patent sellers under any contingent-payment terms.

The 2014-03-30 filing referenced the company's employee count and patent portfolio composition, cited as claims #21026 and #21013, and a string of portfolio acquisitions from Delphi, Intergraph, Mosaid, CyberFone, TeleCommunication Systems and others, cited as claims #21016 through #21023. What has not been said in the record is the actual headcount, the number of patents held, or the purchase price paid for each portfolio. Without those, a reader cannot judge whether these assets were worth what was paid for them.

The company received Nasdaq deficiency deadlines on 2018-07-12, cited as claims #15465 through #15468, including a requirement to add two independent directors and a compensation committee deadline of 2019-06-28. It then announced compliance regained on 2018-08-21, cited as claim #15438, after appointing Michael Berg and naming Fred Thiel compensation committee chair, cited as claims #15435 and #15437. What has not been said is whether the second independent director seat and the full compensation committee requirement were actually satisfied by their deadlines. Weeks after his appointment, Thiel himself received 50,000 options, cited as claim #15396, and the company has not said how that bears on his independence when setting executive pay.

The company disclosed on 2014-03-30 a string of patent portfolio acquisitions from Mosaid, Siemens Technology, Intergraph, TeleCommunication Systems, Delphi, Zhone, CyberFone and TT IP, cited as claims #21016, #21031, #21020, #21019, #21023, #21035, #21017 and #21021; what has not been said is the purchase price, upfront-versus-contingent split, and any revenue-share owed back to each seller. The same filing names IP Navigation Group as a key supplier (claim #21024) without stating what share of licensing or litigation proceeds that firm takes.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is who received those shares, the dollar value on the issuance date, why the deal failed, and whether the director resignations disclosed on 2018-06-28 (claim #15476) were connected to the termination.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459; what has not been said is where the machines sit, how many miners or how much hashrate was deployed, the megawatt capacity of the host site, and who supplies the power and on what contract term.

The company disclosed on 2018-10-15 option grants to CEO Merrick Okamoto described as 5,000,000 shares (claim #15389), 2,500,000 issued (claim #15385) and 2,500,000 unvested with change-of-control acceleration (claim #15386); what has not been said is the strike price, vesting schedule, what counts as a change of control, and how a single executive's grant reconciles with the 10,000,000-share reserve of the 2018 Equity Incentive Plan (claim #15460), which it would consume roughly half of.

The company disclosed on 2018-08-13 substantial doubt about a 12-month cash runway, cited as claim #15462; what has not been said is the monthly burn rate, the dollar amount of financing needed, and the intended source. Combined with an interim CEO (claim #16644) whose name and appointment terms are not stated and a 10-K deadline of 2018-04-15 (claim #16645) with no confirmation it was met, a reader cannot tell whether the company was current in its filings during that stretch.

The company disclosed on 2018-08-21 that it regained NASDAQ compliance after appointing Michael Berg and Fred Thiel (claims #15438, #15435, #15437); what has not been said is whether the separate compensation committee requirement with a 2019-06-28 deadline (claim #15468) was resolved, and whether either new independent director had any prior business, investment or consulting relationship with management. Separately, claims #20895 through #20903 (Mara Foundation, Ashish Thakkar, operations in 26 countries) appear to describe the unrelated African Mara Group and should not be read as company disclosures.

The company disclosed on 2018-10-15 that Merrick Okamoto received 2.5 million stock options (claim #15385), that 2.5 million unvested options carry change-of-control acceleration (claim #15386), and that he holds 5 million options in total (claim #15389). What has not been said is how these figures reconcile into one grant schedule. What are the exercise prices, the vesting triggers for each tranche, and the change-of-control definition that would accelerate them?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger ended with a break-up fee payable in 3,000,000 shares (claim #15461), following the termination first reported on 2018-06-28 (claim #15477), when directors also resigned (claim #15476). What has not been said is the dollar value of those shares at issuance, who received them, and whether any resigning director had a relationship with Global Bit Ventures. Why did the merger fail, and why was the fee paid in stock rather than cash?

The company disclosed on 2018-08-13 a cryptocurrency mining expansion (claim #15459) in the same filing that flagged doubt about its 12-month cash runway (claim #15462). What has not been said is any operational unit behind the expansion: machine count, hashrate, site location, power contract terms or cost per kWh, and hosting counterparty. How would a company uncertain of a year's liquidity fund the expansion, and with what dilution?

The company disclosed on 2018-07-12 a series of NASDAQ deadlines: a compliance plan due 2018-08-24 (claim #15466), two new independent directors by 2019-01-07 (claim #15465), listing-rule compliance by 2019-01-07 (claim #15467), and a compensation committee requirement by 2019-06-28 (claim #15468). It then stated on 2018-08-21 that compliance was regained (claim #15438), with appointments of Michael Berg (claim #15435) and Fred Thiel (claim #15437). What has not been said is whether that regained compliance covered the separate compensation-committee deadline, and whether NASDAQ issued formal written confirmation. Relatedly, the company named an interim CEO on 2018-04-01 without identifying them in the record (claim #16644), and never confirmed whether the 10-K due 2018-04-15 (claim #16645) was filed on time.

The company disclosed on 2014-03-30 licensing agreements with Zhone (claim #21022) and Siemens Technology (claim #21018), a key supplier relationship with IP Navigation Group (claim #21024), the start of patent revenue (claim #21014), and an employee count (claim #21026). What has not been said is the dollar value of any license, the revenue-share or contingency-fee split owed to IP Navigation Group, how concentrated revenue was across those licensees, and the actual headcount. Separately, the record never resolves the real-estate disposition announced on 2013-02-04 (claim #21044): was the business sold, and on what terms?

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461). What has not been said is who received those shares, how they were valued on the issue date, and what, if anything, Marathon got in return, such as mining equipment, hosting rights or IP. Nor has it said why a merger that apparently justified a break-up fee in stock was abandoned.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459). In the same filing it flagged uncertainty over whether it had 12 months of cash runway (claim #15462). What has not been said is the scale or location of the mining operation: machine count, hashrate, megawatts under contract, site address, power counterparty or price per kWh. It also has not said how an expansion is to be funded when the company is warning on going concern.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO (claims #15384, #15387) with a $350,000 base salary (claim #15388). The same filing gives option figures of 5,000,000 shares (claim #15389) and 2,500,000 issued plus 2,500,000 unvested with change-of-control acceleration (claims #15385, #15386). What has not been said is whether these figures overlap or add up, the strike prices, the vesting schedules, and the triggers that define a change of control. Investors would also want to know whether Okamoto or entities affiliated with him had any prior financing, consulting or related-party relationship with the company before he took both top roles.

The company disclosed director resignations on 2018-06-28 (claim #15476), then a set of NASDAQ deadlines for new independent directors, compensation committee composition and a compliance plan (claims #15465, #15466, #15467, #15468). It reported regaining compliance on 2018-08-21 (claim #15438) after appointing Michael Berg and Fred Thiel (claims #15435, #15437). What has not been said is why the prior directors resigned, whether any disagreement with management was involved, and how the 50,000-share option grant to Thiel (claim #15396) bears on the independence assessment. It also has not confirmed whether the 2019-06-28 compensation committee deadline was separately satisfied.

The company disclosed on 2014-03-30 that patent revenue generation had commenced (claim #21014), alongside licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is the dollar value or structure of those licenses (lump sum or running royalty), how much of reported revenue each counterparty accounted for, and the contingency-fee or revenue split owed to IP Navigation Group. The cost of acquiring the CyberFone, Delphi, TeleCommunication Systems, Intergraph and Mosaid portfolios (claims #21017, #21023, #21019, #21020, #21016) is also undisclosed, as is any backend revenue-share owed to the sellers.

The company disclosed an interim Chief Executive Officer on 2018-04-01 (claim #16644) alongside a 10-K filing deadline of 2018-04-15 (claim #16645). What has not been said in this record is who the interim CEO was, why the prior CEO departed, and whether the annual report was filed on time or under an extension. Separately, claims #20895 through #20903 (Mara Foundation, Ashish and Jagdish Thakkar, operations in 26 countries) appear to describe the unrelated Mara Group rather than Marathon. They were not treated as company disclosures here.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is why Marathon, rather than the counterparty, bore the fee, what the shares were worth on the issuance date, and which specific breach or condition failure triggered the termination.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO at a $350,000 base salary with 5,000,000 stock options, 2,500,000 issued outright and 2,500,000 unvested with change-of-control acceleration, cited as claims #15384 through #15389; what has not been said is the strike price, the vesting schedule, how change of control is defined, and whether Okamoto had any prior investor or financing relationship with the company. The grant is also equal to half of the 10,000,000-share 2018 Equity Incentive Plan reserve, cited as claim #15460, and the company has not said whether the board considered that concentration or what remains for other employees.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, alongside going-concern-style uncertainty about a 12-month cash runway, cited as claim #15462; what has not been said is the number and model of machines, the installed or targeted hash rate, the physical site and power contract terms including the price per kWh, and how the expansion would be funded given the runway warning.

The company disclosed that directors resigned on 2018-06-28, cited as claim #15476, that it faced NASDAQ deadlines to add two independent directors and fix compensation committee composition, cited as claims #15465 through #15468, and that it regained compliance on 2018-08-21 with Michael Berg and Fred Thiel appointed to committee chairs, cited as claims #15435, #15437 and #15438; what has not been said is which directors resigned, whether any resignation involved a disagreement with management, and whether the separate compensation committee deadline of 2019-06-28 was independently satisfied.

The company disclosed licensing agreements with Siemens Technology and Zhone, a key supplier relationship with IP Navigation Group, and the commencement of patent revenue on 2014-03-30, cited as claims #21014, #21018, #21022 and #21024; what has not been said is the dollar consideration of each license, the share of recoveries paid to IP Navigation Group or original patent sellers, and what percentage of revenue came from a single licensee.

The company disclosed an Interim Chief Executive Officer and a 10-K filing deadline of 2018-04-15 in a 2018-04-01 record, cited as claims #16644 and #16645; what has not been said in the record is the interim CEO's name and compensation, or whether the 10-K was filed on time. Claims #20895 through #20903 describe Mara Group, an African conglomerate founded by Ashish J. Thakkar, and appear to be attributed to Marathon by mistake, so they were not used here; the claim list was also truncated after claim #15384.

The company disclosed on 2018-08-13 that its Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claim #15461, following the 2018-06-28 termination and director resignations, cited as claims #15477 and #15476. What has not been said is who received those shares, what they were worth on issuance, whether the resigning directors had any connection to Global Bit Ventures, and why the merger was abandoned. Why did a terminated deal cost shareholders 3,000,000 shares, and who was paid?

The company disclosed on 2018-10-15 that Merrick Okamoto, named CEO and Chairman, received 5,000,000 stock options, cited as claims #15384 and #15389, two months after a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claim #15460. What has not been said is the strike price, the vesting schedule, any performance conditions, or whether Okamoto or his affiliates had prior financing or other related-party dealings with the company. Why did half the plan reserve go to a single executive, on what terms, and what other arrangements exist between the company and Okamoto?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, and on the same date flagged uncertainty about its 12-month cash runway, cited as claim #15462. What has not been said is the number of miners, hashrate, site location, power cost or contract counterparty, or how the expansion would be funded given the going-concern doubt. Where are the machines, who supplies the power and at what rate, and what financing pays for the build-out?

The company disclosed on 2014-03-30 a series of patent portfolio acquisitions from Siemens, Delphi, Intergraph, TeleCommunication Systems, Zhone, Mosaid and others, cited as claims #21016 through #21035, alongside a key supplier relationship with IP Navigation Group, cited as claim #21024, and the start of patent revenue, cited as claim #21014. What has not been said is the consideration paid for each portfolio, whether sellers retained back-end revenue shares, what IP Navigation Group is paid, or how much revenue each portfolio has produced. What share of each licensing or litigation recovery goes to the original owners and to IP Navigation Group, and which portfolios account for the revenue?

The company disclosed on 2018-07-12 deadlines to appoint two independent directors and to bring its compensation committee into Nasdaq compliance by 2019-06-28, cited as claims #15465 and #15468, and on 2018-08-21 that it had regained compliance with Michael Berg and Fred Thiel chairing the audit and compensation committees, cited as claims #15438, #15435 and #15437. What has not been said is whether the compensation committee met independence and size requirements when it approved the October 2018 option grants, cited as claims #15389 and #15396, including Thiel's own 50,000 options. Was the committee that approved the CEO's 5,000,000-option grant fully constituted and independent at the time, and did any member grant awards to themselves?

The company disclosed on 2018-04-01 an interim CEO and a 10-K filing deadline of 2018-04-15, cited as claims #16644 and #16645, but the record does not name the interim officer or confirm the filing was made on time, and a 2014 employee count, cited as claim #21026, appears with no figure. Who served as interim CEO, was the 2017 10-K filed timely, and how many employees did the company have through its pivot from patents to mining? Claims #20895 through #20903, dated 2013-2014, describe Mara Group, the African conglomerate founded by Ashish Thakkar, and appear to be misattributed to this company; the claim list also ends at claim #15388, so later disclosures were not reviewed.

The company disclosed on 2018-08-13 that it was expanding into cryptocurrency mining, cited as claim #15459, while warning in the same filing of uncertainty over its 12-month cash runway, cited as claim #15462. What has not been said is where the machines are or will be sited, how many units and what hash rate or megawatts are involved, whether a third-party host is involved and under what contract, and how the expansion will be funded given the stated going-concern risk.

The company disclosed the termination of the Global Bit Ventures merger on 2018-06-28, cited as claim #15477, and a share-based break-up fee of 3,000,000 shares, cited as claim #15461. What has not been said is which side paid the fee and who received the shares, what value was assigned to them at issuance, whether any registration rights or lock-up attached, and why the merger was terminated. The coincident director resignations, cited as claim #15476, are also not explained.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO at a $350,000 base salary with 5,000,000 options, cited as claims #15388 and #15389. Half of those options, 2,500,000, carry change-of-control acceleration, cited as claims #15385 and #15386. What has not been said is the strike price, the vesting schedule and any performance conditions on the unvested tranche, or how a single grant equal to half of the 10,000,000-share 2018 Equity Incentive Plan reserve, cited as claim #15460, was sized. It is also unexplained why the change-of-control trigger was attached to a company with a disclosed cash-runway problem.

The company disclosed that Fred Thiel was named Compensation Committee chairman on 2018-08-21 and that Thiel and Michael Rudolph each received 50,000 options on 2018-10-15, cited as claims #15437, #15396 and #15394. Michael Berg was named Audit Committee chair, cited as claim #15435. What has not been said is which committee members approved the Okamoto package, whether any director had a prior relationship with the new CEO, and whether the Nasdaq compensation-committee requirement carrying a 2019-06-28 deadline was ever separately confirmed as satisfied. That deadline is cited as claim #15468, distinct from the general compliance regained in claim #15438.

The company disclosed an Interim Chief Executive Officer and a 10-K deadline of 2018-04-15, cited as claims #16644 and #16645. What has not been said is the interim officer's name and compensation, whether the 10-K was filed on time or under an extension, and what led to the leadership vacancy that preceded the October appointment.

The company disclosed in its 2014-03-30 filing a string of patent portfolio acquisitions and licenses involving Siemens, Zhone, Delphi, Intergraph, Mosaid, TeleCommunication Systems and CyberFone, cited as claims #21016 through #21023, #21031 and #21035. It also named IP Navigation Group as a key supplier, cited as claim #21024, and announced that patent revenue had commenced, cited as claim #21014. What has not been said is the purchase price or revenue-share terms for each portfolio, what share of recoveries flows to IP Navigation Group or to the original patent owners, and how much revenue each portfolio generated. It is also unexplained why the Sampo IP acquisition announced on 2013-02-04, cited as claim #21042, reappears as a separate acquisition and rights grant in 2014, cited as claims #21015 and #21028.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares (claim #15461), following the termination announced 2018-06-28 (claim #15477). What has not been said is which party bears the fee, whether the shares were ever issued, at what per-share value, and how the obligation was recorded.

The company disclosed on 2018-10-15 that Merrick Okamoto was named Executive Chairman and CEO at a $350,000 base salary with 5,000,000 stock options (claims #15387, #15388, #15389). The same filing cites 2,500,000 options issued and 2,500,000 unvested options with change-of-control acceleration (claims #15385, #15386), against a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460). What has not been said is the strike price, the vesting schedule or performance conditions on the unvested tranche, and the rationale for granting half the plan reserve to a single executive while 12-month cash runway was in doubt (claim #15462).

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459), alongside a going-concern-style warning that cash may not cover the next 12 months (claim #15462). What has not been said is the scale of the expansion: machine count, hashrate, site location, power contract terms and price, any hosting counterparty, and how the expansion would be financed given the stated runway risk.

The company disclosed on 2018-06-28 that directors had resigned (claim #15476). It then faced a NASDAQ deadline to add two independent directors by 2019-01-07 and a separate compensation-committee compliance deadline of 2019-06-28 (claims #15465, #15467, #15468). It reported compliance regained on 2018-08-21 after appointing Michael Berg and Fred Thiel (claims #15435, #15437, #15438). What has not been said is who resigned and why, whether any disagreement with management was cited, and why the compensation-committee deadline remained open after compliance was declared regained.

The company disclosed an Interim Chief Executive Officer and a 10-K filing deadline of 2018-04-15 (claims #16644, #16645). What has not been said is the interim CEO's identity and compensation, and whether the 10-K was filed on time or under an NT 10-K extension.

The company disclosed in its 2014-03-30 filing that patent revenue generation had commenced (claim #21014). That filing also cited licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is the revenue attributable to each licensee, the upfront versus running-royalty split, and the revenue-share or fee economics paid to IP Navigation Group out of those recoveries.

The company disclosed a cluster of patent portfolio acquisitions on 2014-03-30, from Mosaid, CyberFone, TeleCommunication Systems, Intergraph, Delphi, Siemens and Zhone, cited as claims #21016, #21017, #21019, #21020, #21023, #21031 and #21035, alongside licensing agreements with Siemens and Zhone (claims #21018, #21022). What has not been said is what was paid for each portfolio, whether the sellers retain a back-end share of enforcement proceeds, and whether the Siemens and Zhone licenses are license-backs to the same parties that sold the patents. If they are, how much of the reported patent revenue (claim #21014) is actually recycled seller consideration?

The company named IP Navigation Group as a key supplier on 2014-03-30, cited as claim #21024, while disclosing active enforcement litigation (claim #21025). What has not been said is the fee structure: is it a contingency percentage of recoveries, a fixed retainer, or both? Does any officer or director have an economic interest in IP Navigation Group?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a 3,000,000-share break-up fee, cited as claims #15461 and #15477. Directors resigned on the same date as the termination (claim #15476). What has not been said is whether those shares were actually issued and at what per-share value the fee was booked. Which directors left, did any resignation cite a disagreement, and did any departing director have a relationship with Global Bit Ventures?

The company disclosed Merrick Okamoto's appointment as Executive Chairman and CEO on 2018-10-15 at a $350,000 base salary with 5,000,000 options, split into two 2,500,000 tranches, one carrying change-of-control acceleration (claims #15384 to #15389). What has not been said is the exercise price, the vesting schedule, and any performance conditions on either tranche. It is also unstated whether this grant draws on the 10,000,000-share 2018 Equity Incentive Plan reserve disclosed in claim #15460, which would put half the pool with one person. Finally, did Okamoto or his affiliates hold pre-existing equity or financing positions in the company?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that flagged uncertainty about a 12-month cash runway (claim #15462). What has not been said is the number and model of machines, the hash rate, the power capacity in megawatts, the hosting site location and electricity rate, and how the expansion was to be funded given the going-concern language.

The company disclosed an interim CEO on 2018-04-01 alongside a 10-K deadline of 2018-04-15, cited as claims #16644 and #16645. Separately, it said on 2018-08-21 that it had regained NASDAQ compliance (claim #15438), even though the 2018-07-12 notices (claims #15465 to #15468) set board and compensation-committee deadlines running into 2019. What has not been said is who the interim CEO was and whether the 10-K was filed on time. Did the 2018-08-21 compliance notice resolve every deficiency, or only the audit-committee one?

Separately, claims #20895 to #20903 (Mara Foundation, the Thakkar family, UN Women and EY partnerships, operations in 26 countries) appear to describe the African conglomerate Mara Group rather than Marathon Patent Group or Marathon Digital. They should be checked for entity misattribution before any gap is drawn from them.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 it described a break-up fee of 3,000,000 shares (claim #15461). What has not been said is why the deal was terminated, whether those shares have actually been issued, what they were worth on the issue date, and how the charge was booked. The same-day director resignations (claim #15476) raise a further question: which directors left, and did they cite a disagreement with the company?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options on his appointment as Executive Chairman and CEO (claims #15387, #15389). The 2018 Equity Incentive Plan reserves only 10,000,000 shares (claim #15460). What has not been said is the strike price, the vesting schedule, any performance conditions, or his cash salary. The CFO's $180,000 base salary was disclosed (claim #15391), so the omission of the CEO's salary stands out. So does the fact that half of the plan reserve went to one person.

The company disclosed on 2018-08-21 that Fred Thiel was appointed Compensation Committee Chairman (claim #15437), and on 2018-10-15 that he received 50,000 options (claim #15396). What has not been said is who approved the option grants to the compensation committee's own members. It is also unclear whether a second independent director was ever seated to meet the compensation committee requirement (claims #15465, #15468), since only Michael Berg's appointment is on record (claim #15435).

The company disclosed on 2018-08-13 a cryptocurrency mining expansion (claim #15459) alongside going-concern-style uncertainty about its 12-month cash runway (claim #15462). What has not been said is the number of miners, the installed hashrate, where the machines are hosted, the power price and capacity in megawatts, and who the hosting counterparty is. It is also unclear how the expansion would be funded given the stated runway risk.

The company disclosed on 2014-03-30 that patent revenue had begun (claim #21014). It named licensing agreements with Siemens Technology and Zhone (claims #21018, #21022) and a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is how much revenue each licensee contributed, what revenue share or contingency fee IP Navigation Group receives, and how many employees the company had, since the headcount claim (claim #21026) carries no figure.

The company disclosed on 2018-04-01 that an Interim Chief Executive Officer was in place (claim #16644) and that the 10-K was due by 2018-04-15 (claim #16645). What has not been said is who the interim CEO was, why the prior CEO left, and whether the 10-K was filed on time or required an extension notice.

The company disclosed the termination of the Global Bit Ventures merger on 2018-06-28, cited as claims #15477 and #15461, along with a break-up fee paid in 3,000,000 shares, and on the same date that directors resigned (claim #15476). What has not been said is why the deal collapsed, what the shares were worth on the termination date, who received them, which directors left, and whether those resignations were tied to the failed merger.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459. What has not been said is where the machines are located, how many miners were deployed or ordered, the expected hashrate or megawatts, the power price, and whether the site is owned or run by an unnamed third-party hosting provider.

The company disclosed 2.5 million stock options to the CEO on 2018-10-15 (claim #15385), drawn from a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460), plus grants to David Lieberman and James Crawford (claims #15391, #15392, #15393). What has not been said is the exercise price, vesting schedule, and any performance conditions for these grants, what roles Lieberman and Crawford hold, and what interim CEO Merrick Okamoto (claim #16648) was paid when appointed in April 2018.

The company disclosed substantial doubt about a 12-month cash runway on 2018-08-13, cited as claim #15462, while expanding mining capacity at the same time. What has not been said is the cash balance and monthly burn, how the expansion is being funded, and what dilution shareholders should expect from the financing needed to close the gap.

The company disclosed a delayed 10-K filing deadline of 2018-04-15 (claim #16645) and a series of Nasdaq compliance deadlines (claims #15466, #15467, #15468, #15465), later reporting regained compliance on 2018-08-21 (claim #15438). What has not been said is the cause of the late 10-K, whether the auditor or internal-control findings were involved, and whether the 2019 compensation-committee deadline was ever formally met.

The company disclosed licensing agreements with Zhone and Siemens and the start of patent revenue on 2014-03-30 (claims #21022, #21018, #21014), and named IP Navigation Group as a key supplier (claim #21024). What has not been said is how much each license paid, how proceeds are split with IP Navigation Group and the original patent sellers under contingency arrangements, and how many employees (claim #21026 gives no figure) actually run a portfolio of more than seven acquired patent families.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares (claim #15461), following the termination announced 2018-06-28 (claim #15477). What has not been said is who received those shares, at what per-share value the fee was booked, whether the shares were actually issued or settled another way, and why the merger was abandoned when the cryptocurrency mining expansion (claim #15459) depended on it.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that flagged doubt about its 12-month cash runway (claim #15462). What has not been said is how many machines or how much hash rate the expansion involves, where the miners are hosted (site address, host counterparty, power contract and price per kWh), and what financing will cover both the expansion and the runway gap through 2019-08-13.

The company disclosed 2.5 million stock options to Merrick Okamoto on his appointment as Executive Chairman and CEO on 2018-10-15 (claims #15385 and #15387), drawn from a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460), alongside 50,000-share director grants (claims #15395 and #15396). What has not been said is the exercise price, vesting schedule and any performance conditions on the CEO grant, what he was paid while interim CEO from 2018-04-01 (claim #16648), and how much of the plan reserve remains after these grants.

The company disclosed on 2018-08-21 that it had regained compliance with NASDAQ listing rules and seated three new directors (claims #15438, #15435, #15436, #15437). Yet the 2018-07-12 deadlines for the compliance plan, the two independent directors, and compensation-committee composition remain recorded as delayed (claims #15466, #15465, #15468), and the 2018-04-15 10-K deadline is also recorded as delayed (claim #16645). What has not been said is which specific rule was breached, whether NASDAQ formally closed each deficiency, and why the annual report was late.

The company disclosed on 2014-03-30 that it had begun generating patent revenue (claim #21014) through licensing agreements with Siemens and Zhone (claims #21018 and #21022), supported by a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is the revenue attributable to each licensee, the upfront and contingent purchase prices paid for each acquired portfolio (claims #21016, #21017, #21019, #21020, #21021, #21023), and the fee or revenue share IP Navigation Group retains. Separately, the real estate disposition flagged on 2013-02-04 (claim #21044) was never followed by a disclosed sale price or completion date.

The record attributes to this company claims dated 2013-04-04 and 2014 that describe the Mara Foundation, Ashish and Jagdish Thakkar, operations in 26 countries, and partnerships with UN Women and EY (claims #20895 through #20903). These appear to belong to the unrelated African conglomerate Mara Group rather than Marathon Patent Group. What has not been resolved is whether any actual corporate relationship exists, and absent one, these claims should not be read as Marathon disclosures.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares (claim #15461, following the termination noted on 2018-06-28 in claim #15477). What has not been said is who received those shares, how they were valued at issuance, whether they carried registration rights or lock-ups, and how the fee was recorded. The filing also does not say why a fee was owed by the company rather than by the counterparty.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459), in the same filing that flagged uncertainty over a 12-month cash runway (claim #15462). What has not been said is the scale of the expansion, the site and its power supply, and how it will be paid for. Missing specifics include the number and model of miners, target hashrate, the physical facility, the power contract terms and cost per kWh, and whether hosting partners are involved. Without that, the plan cannot be reconciled with the going-concern language.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO (claim #15387) and that 2.5 million unvested options carry change-of-control acceleration (claim #15386). It also disclosed option grants of 200,000 shares to David Lieberman (claim #15392) and 100,000 shares to James Crawford (claim #15393). What has not been said is the strike price, vesting schedule and holder of the accelerating options, or the cash compensation attached to the combined Chairman/CEO role. Also undisclosed is whether any related-party relationships exist between these executives and the company's financing counterparties. This matters more because the board was still working toward compensation committee compliance (claim #15468), drawn from the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460).

The company disclosed on 2018-06-28 that directors resigned (claim #15476), which led to NASDAQ compliance deadlines (claims #15465, #15466, #15467) and later a statement that compliance was regained (claim #15438). What has not been said is which directors left, whether any resignation involved a disagreement with management, and whether the departures were tied to the failed Global Bit Ventures merger announced the same day.

The company disclosed on 2018-04-01 an interim CEO change (claims #16644, #16648) and a 10-K filing deadline of 2018-04-15 that was delayed (claim #16645). What has not been said is why the annual report was late, whether the delay involved auditor issues or an accounting review, and whether any restatement or internal-control weakness followed.

In its 2014-03-30 annual filing, the company described patent revenue generation commencing (claim #21014) and licensing agreements with Zhone and Siemens Technology (claims #21022, #21018). It also named IP Navigation Group as a key supplier (claim #21024). What has not been said is the dollar value of license revenue by counterparty, the revenue-share or contingency-fee split paid to IP Navigation Group, and how much of the acquired portfolios' purchase price was contingent on future recoveries. The employee count (claim #21026) is recorded without a figure, so it is also unclear how a firm with no R&D spend (claim #21027) staffs its enforcement litigation (claim #21025).

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), and on 2018-08-13 that a share-based break-up fee of 3,000,000 shares was owed (claim #15461); what has not been said is why the deal was terminated, what dollar value was assigned to those shares at issuance, how the fee was booked, and whether the shares carried registration or resale rights that added dilution.

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459), alongside going-concern-style uncertainty about a 12-month cash runway (claim #15462); what has not been said is the number and model of miners deployed, the hosting site's location and contracted power capacity in megawatts, the all-in power cost per kWh, and how the expansion would be funded given the stated runway risk.

Merrick Okamoto was named Interim CEO on 2018-04-01 (claims #16644, #16648) and Executive Chairman and CEO on 2018-10-15, with 2.5 million unvested options carrying change-of-control acceleration (claims #15387, #15386); what has not been said is the strike price, vesting schedule, and definition of a change-of-control trigger for those options, or what cash salary and bonus terms accompanied the permanent appointment. The same gap applies to the 100,000 and 200,000 option grants to James Crawford and David Lieberman (claims #15393, #15392), which were disclosed without strike, vesting, or role context.

The company disclosed a 2018 Equity Incentive Plan reserve of 10,000,000 shares on 2018-08-13 (claim #15460) shortly after resolving a compensation committee compliance deficiency (claims #15468, #15437); what has not been said is what that reserve represents as a percentage of shares outstanding, how many shares were already granted from it by year-end, and whether the newly constituted compensation committee approved the October 2018 grants.

The 10-K due 2018-04-15 was delayed (claim #16645), followed by NASDAQ compliance-plan and board-independence deadlines (claims #15466, #15467, #15465) that were later reported as regained on 2018-08-21 (claim #15438); what has not been said is the specific cause of the late 10-K, whether it involved auditor disagreement or restatement, and the actual filing date.

The 2014-03-30 filing lists more than a dozen patent portfolio acquisitions, including CyberFone, Mosaid, Siemens, Intergraph, TeleCommunication Systems, Delphi, Zhone, and TT IP (claims #21016 through #21036), licensing agreements with Zhone and Siemens (claims #21022, #21018), patent revenue commencement (claim #21014), and IP Navigation Group as a key supplier (claim #21024); what has not been said is the purchase consideration and any revenue-share owed back to the sellers for each portfolio, the economics of the IP Navigation relationship, and how much of reported licensing revenue came from each named licensee or litigation settlement (claim #21025).

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477. What has not been said is why the merger was terminated, whether those shares were actually issued and when, what they were worth on the issuance date, and how the fee was accounted for in the period's results.

The company announced a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459. What has not been said is how many machines were involved and of what model, what hashrate they were expected to deliver, where they would be sited, who the hosting or power counterparty was, and at what contracted cost per kilowatt-hour.

The company disclosed uncertainty about its 12-month cash runway on 2018-08-13, cited as claim #15462, alongside a 10,000,000-share 2018 Equity Incentive Plan reserve, cited as claim #15460. What has not been said is the cash balance and monthly burn rate behind that going-concern language, and the size and structure of the financing management expected to rely on, including how much of the plan reserve could dilute holders before that runway question was resolved.

The company disclosed on 2018-10-15 option grants of 100,000 shares to James Crawford and 50,000 shares each to directors Fred Thiel, Michael Berg and Michael Rudolph, cited as claims #15393 through #15396. It had also disclosed Merrick D. Okamoto as Interim Chief Executive Officer on 2018-04-01, cited as claims #16644 and #16648. What has not been said is the strike price, vesting schedule and grant-date fair value of those options, what role Crawford held that justified twice the director grant, and what compensation or related-party arrangements applied to Okamoto during his interim tenure.

The company listed IP Navigation Group as a key supplier in its 2014-03-30 filing, cited as claim #21024, alongside licensing agreements with Zhone and Siemens Technology, cited as claims #21022 and #21018, and a string of portfolio acquisitions including CyberFone, Intergraph, Delphi and TeleCommunication Systems, cited as claims #21017, #21020, #21023 and #21019. What has not been said is the consideration paid for each portfolio, the license fees received, and the revenue-share or contingency terms owed to IP Navigation Group out of litigation and licensing proceeds.

The company indicated on 2018-04-01 that its 10-K annual report filing deadline of 2018-04-15 would slip, cited as claim #16645, and on 2018-07-12 it carried NASDAQ compliance deadlines for its compensation committee and board independence, cited as claims #15466 through #15468 and #15465. What has not been said is the cause of the 10-K delay, including whether an auditor change or a restatement was involved, and the date on which the report was ultimately filed.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares (claim #15461), following the 2018-06-28 termination and same-day director resignations (claims #15477, #15476). What has not been said is why the deal collapsed, which directors resigned and whether that was connected to the termination, and what fair value and P&L charge were recorded for the 3,000,000 shares issued.

On 2018-08-13 the company announced a cryptocurrency mining expansion (claim #15459) while also flagging substantial doubt about a 12-month cash runway (claim #15462). What has not been said is the scale of the expansion: machine count, hashrate, megawatts, where the miners are hosted, the power price per kWh, and how a company uncertain of its own runway intended to fund the capex.

On 2018-10-15 the company disclosed 2.5 million stock options for Merrick Okamoto on his appointment as Executive Chairman and CEO (claim #15385), after he had served as interim CEO since 2018-04-01 (claim #16648). It also disclosed 50,000-share grants to directors including Fred Thiel, the Compensation Committee chair (claims #15396, #15395, #15437). What has not been said is the strike price, vesting schedule, and performance conditions, what share of the 10,000,000-share 2018 Equity Incentive Plan reserve these grants consumed (claim #15460), and whether the compensation chair abstained from approving awards that included his own.

The company disclosed on 2018-04-01 a 10-K filing deadline of 2018-04-15 that was missed (claim #16645), during the same period as the interim CEO arrangement (claim #16644). What has not been said is the cause of the delay: whether it reflected auditor turnover, an unresolved accounting position, or an internal-control weakness, and whether any restatement followed.

The 2014-03-30 annual disclosures list more than ten patent portfolio acquisitions, including Siemens, Delphi, Intergraph, TeleCommunication Systems, and Mosaid (claims #21031, #21023, #21020, #21019, #21016). They also announce that patent revenue had commenced (claim #21014) and that licensing agreements were signed with Siemens and Zhone (claims #21018, #21022). What has not been said is the purchase price or revenue-share terms of each portfolio, the dollar value of each license, and the share of recoveries paid to IP Navigation Group, which is named as a key supplier (claim #21024).

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 that the break-up fee was paid in stock, 3,000,000 shares (claim #15461). What has not been said is what those shares were worth when issued, who received them, whether they carried registration rights or lock-ups, and how much dilution they caused for existing holders.

The company disclosed on 2014-03-30 that it had started generating patent revenue (claim #21014), that it held licensing agreements with Zhone and Siemens Technology (claims #21022 and #21018), and that IP Navigation Group was a key supplier (claim #21024). What has not been said is how much revenue each license produced, whether those licenses were one-time payments or recurring royalties, and what share of recoveries went to IP Navigation Group under its fee arrangement.

The company disclosed on 2014-03-30 a string of patent portfolio acquisitions from Delphi, Intergraph, TeleCommunication Systems, CyberFone, Mosaid, Siemens, Zhone and TT IP (claims #21016 through #21036). What has not been said is the purchase price of each portfolio, how each deal split cash and stock, and whether any sellers kept a back-end revenue share that would reduce the company's net from enforcement litigation (claim #21025).

The company disclosed on 2018-08-13 a plan to expand cryptocurrency mining (claim #15459), alongside uncertainty about having 12 months of cash runway (claim #15462). What has not been said is the number of machines or the hash rate targeted, where the mining site is, what power capacity and electricity rate were contracted, and how the expansion would be paid for given the going-concern language.

The company disclosed on 2018-10-15 compensation terms of a $180,000 salary and 200,000 options for David Lieberman, plus option grants to James Crawford and Michael Rudolph of 100,000 and 50,000 shares (claims #15391 through #15394), all drawn from a 10,000,000-share 2018 plan reserve (claim #15460). What has not been said is the strike prices, the vesting schedules, and Lieberman's title. It has also not disclosed any compensation for interim CEO Merrick Okamoto (claim #16648), who was named in 2018.

The company disclosed that directors resigned on 2018-06-28 (claim #15476) and that it later fell short of NASDAQ board and compensation committee requirements, with remediation deadlines and then a 2018-08-21 announcement that compliance was regained (claims #15465 through #15468 and #15438). It also disclosed that its 10-K was late against a 2018-04-15 deadline (claim #16645). What has not been said is why the directors left, whether any departure involved a disagreement with management, and what caused the 10-K delay.

The company disclosed on 2018-08-13 that its terminated merger with Global Bit Ventures carried a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is whether those shares were actually issued, at what per-share value the fee was booked, and whether any further liability to Global Bit Ventures remains. Why did a terminated merger require the company to pay, rather than receive, a break-up fee?

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, alongside a warning that its cash might not last the next 12 months, cited as claim #15462; what has not been said is how many miners it owns, what hash rate it targets, where the machines are physically located, who supplies the power and at what rate, and how much cash it had against that expansion. How large was the actual funding gap, and what financing was planned to close it?

The company disclosed on 2014-03-30 a series of patent portfolio acquisitions from Sampo, Mosaid, CyberFone, Siemens, Delphi, Zhone, TeleCommunication Systems, Intergraph and TT IP, cited as claims #21015 through #21036, together with licensing agreements with Siemens and Zhone (#21018, #21022) and a key supplier relationship with IP Navigation Group (#21024); what has not been said is the purchase price of each portfolio, the revenue-share or contingent payments owed back to the sellers, and the fee terms paid to IP Navigation Group. How much of any licensing or litigation recovery does the company actually keep?

The company disclosed on 2018-04-01 that Merrick D. Okamoto was serving as Interim Chief Executive Officer, cited as claims #16644 and #16648, and on 2018-10-15 that David Lieberman would receive a $180,000 salary and 200,000 options while directors received option grants (#15391 through #15394); what has not been said is Okamoto's pay or equity, the strike price and vesting schedule of these options, and how the 10,000,000-share 2018 Equity Incentive Plan reserve (#15460) compares with total shares outstanding. How much dilution could the plan cause if it were fully granted?

The company disclosed a missed 10-K filing deadline of 2018-04-15, cited as claim #16645, and a set of NASDAQ compliance deadlines covering the plan submission, independent directors and the compensation committee (#15465 through #15468), before stating on 2018-08-21 that compliance had been regained (#15438); what has not been said is why the annual report was late, whether the auditor raised any issue, and whether the separate compensation committee deadline of 2019-06-28 was met on its own terms. What caused the filing delay, and what governance changes followed?

The company disclosed an employee count and a patent portfolio composition on 2014-03-30, cited as claims #21026 and #21013, but the record carries no actual headcount or patent count. How many employees and how many patents did the company hold when it began generating patent revenue (#21014)?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477; what has not been said is who received those shares, at what per-share value the fee was booked, how large the resulting charge was against a balance sheet already flagged for going-concern risk, and why the deal was abandoned.

The company disclosed on 2018-10-15 that interim CEO Merrick Okamoto received 5,000,000 stock options, cited as claim #15389, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claim #15460; what has not been said is the strike price, vesting schedule and performance conditions on a grant equal to half the entire plan, or whether the newly seated compensation committee chaired by Fred Thiel, cited as claim #15437, approved it independently.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that flagged doubt about a 12-month cash runway, cited as claim #15462; what has not been said is how many machines were deployed or ordered, the expected hashrate, the physical site and its power contract and cost per kWh, and how the expansion would be funded given the runway warning.

The company disclosed licensing agreements with Siemens and Zhone and the start of patent revenue on 2014-03-30, cited as claims #21018, #21022 and #21014, alongside a key supplier relationship with IP Navigation Group, cited as claim #21024; what has not been said is how much revenue each licensee contributed, what share of recoveries went back to the original patent sellers, and what contingent-fee percentage or minimum payment IP Navigation Group was owed.

The company disclosed an employee count and the absence of any research and development spending on 2014-03-30, cited as claims #21026 and #21027, but the record carries no actual headcount; what has not been said is how many people ran a business that held more than ten acquired portfolios, cited as claims #21028 through #21036, and pursued enforcement litigation, cited as claim #21025, and how much of that work was outsourced to contingency counsel.

The company disclosed Nasdaq deficiency deadlines on 2018-07-12, including a compensation committee compliance deadline of 2019-06-28 that is still marked delayed, cited as claims #15466, #15467 and #15468, then announced on 2018-08-21 that compliance was regained, cited as claim #15438; what has not been said is which specific listing rules were breached, whether the compensation committee requirement was separately cured, and whether Nasdaq formally closed the matter in writing.

The company disclosed a cluster of patent portfolio acquisitions on 2014-03-30, cited as claims #21016, #21017, #21019, #21020, #21023, #21031 and #21034, covering portfolios from Mosaid, CyberFone, TeleCommunication Systems, Intergraph, Delphi, Siemens and TT IP. What has not been said is the consideration paid for each portfolio, or whether the sellers retained back-end shares of future licensing and litigation proceeds that would reduce the company's net economics.

The company disclosed licensing agreements with Zhone and Siemens Technology on 2014-03-30, cited as claims #21022 and #21018, alongside acquisitions of those same parties' patent portfolios, cited as claims #21035 and #21031. What has not been said is whether these licenses were license-backs granted to the sellers as part of the purchase, and if so what royalty, if any, flows to the company and how the license-back affects the value of the acquired patents.

The company disclosed IP Navigation Group as a key supplier on 2014-03-30, cited as claim #21024, during the same period it reported active patent enforcement litigation and the start of patent revenue, cited as claims #21025 and #21014. What has not been said is the fee structure, whether contingency or fixed, what share of gross recoveries IP Navigation receives, and how concentrated the company's enforcement program is on this single vendor.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477, and on 2018-06-28 that directors had resigned, cited as claim #15476. What has not been said is which party paid the fee to whom, the fair value of those shares at the time, why the merger failed, and whether the director resignations were connected to the termination.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, in the same filing that raised doubt about its 12-month cash runway, cited as claim #15462. What has not been said is the installed and targeted hashrate or machine count, the site location and power arrangement, the capital required, and how the expansion would be financed given the going-concern language.

The company disclosed a 10,000,000-share 2018 Equity Incentive Plan and grants of options to David Lieberman (200,000 shares, alongside a $180,000 base salary), James Crawford (100,000) and Michael Rudolph (50,000) on 2018-10-15, cited as claims #15460 and #15391 through #15394, two months after Rudolph joined the board as an independent director, cited as claim #15436. What has not been said is the strike prices and vesting schedules of these grants, or what compensation Merrick Okamoto received as interim CEO, cited as claim #16648, in a period when the 10-K filing was delayed, cited as claim #16645.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), with a break-up fee of 3,000,000 shares (claim #15461), and that directors resigned the same day (claim #15476). What has not been said is which directors left, whether their departures were tied to the failed deal, what the break-up shares were worth when issued, and which party triggered the termination.

The company announced a cryptocurrency mining expansion on 2018-08-13 (claim #15459). What has not been said is the number of machines or hashrate being added, where the miners would be housed, and who would supply power or hosting, and on what contract terms. Without a site or counterparty, the expansion cannot be measured.

On 2018-08-13 the company flagged uncertainty about its 12-month cash runway (claim #15462) and, in the same filing, reserved 10,000,000 shares under the 2018 Equity Incentive Plan (claim #15460). What has not been said is the cash balance and monthly burn behind the going-concern language, how the company planned to fund the mining expansion, and what percentage of shares outstanding the 10M-share reserve represents.

The company disclosed on 2018-10-15 a $180,000 base salary and 200,000 options for David Lieberman (claims #15391, #15392), along with option grants to James Crawford and Michael Rudolph (claims #15393, #15394). What has not been said is the strike price, vesting schedule, or Lieberman's title. It is also unclear why Rudolph, appointed an independent director and Nominating Committee chair on 2018-08-21 (claim #15436), received options within two months of joining. Interim CEO Merrick Okamoto's pay (claim #16648) is not disclosed at all.

The company missed its 2018-04-15 10-K deadline (claim #16645) and then faced a series of NASDAQ compliance deadlines (claims #15465, #15466, #15467, #15468) before saying on 2018-08-21 that compliance was regained (claim #15438). What has not been said is why the 10-K was late, including whether auditor or accounting-restatement issues were involved, and which specific listing rule deficiencies were cured.

In the patent era, the company said on 2014-03-30 that revenue generation had commenced (claim #21014), citing licensing agreements with Zhone and Siemens (claims #21022, #21018) and a key supplier relationship with IP Navigation Group (claim #21024). What has not been said is the dollar value of each license, the revenue share or contingency fee owed to IP Navigation Group, and the purchase prices and patent counts for the CyberFone, Delphi, Intergraph, TeleCommunication Systems and TT IP portfolios (claims #21017, #21023, #21020, #21019, #21034). The 2013-02-04 plan to explore disposing of the real estate business (claim #21044) also has no recorded outcome.

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461), while directors resigned on the same date as the termination (claim #15476). What has not been said is who triggered the termination, what dollar value and accounting charge attached to those 3,000,000 shares, and whether the departing directors were connected to the Global Bit Ventures side of the deal.

The company disclosed on 2018-10-15 that Merrick Okamoto received a $350,000 base salary (claim #15388) and options on 5,000,000 shares (claim #15389), two months after reserving 10,000,000 shares under the 2018 Equity Incentive Plan (claim #15460). What has not been said is the strike price, vesting schedule and any performance conditions on a grant equal to half the plan's reserve, or when and on what terms Okamoto's role moved from interim CEO (claims #16648, #16644) to permanent.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion (claim #15459) in the same filing that flagged uncertainty about a 12-month cash runway (claim #15462). What has not been said is the number of machines, hash rate, hosting site location, power contract price and term, or how the expansion was to be financed without deepening the going-concern risk.

The company disclosed on 2014-03-30 that patent revenue generation had commenced (claim #21014), that it had licensing agreements with Siemens Technology and Zhone (claims #21018, #21022), and that IP Navigation Group was a key supplier (claim #21024). What has not been said is the amount of licensing revenue by counterparty, what share of recoveries is owed back to the original patent sellers or to IP Navigation Group, and how concentrated revenue was in any single licensee.

The company disclosed on 2013-02-04 that it was exploring disposition of its real estate business (claim #21044) and authorizing a 2-for-3 reverse stock split (claim #21046). What has not been said is whether the real estate business was ever sold, to whom and for what consideration, or whether the reverse split was actually effected and on what date.

The company disclosed a 10-K filing deadline of 2018-04-15 that slipped (claim #16645), and on 2018-07-12 a series of NASDAQ compliance deadlines (claims #15466, #15467, #15468), with listing compliance later reported regained on 2018-08-21 (claim #15438). What has not been said is why the 10-K was late, and whether the separate compensation committee compliance deadline of 2019-06-28 (claim #15468) was satisfied or required a further extension.

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee of 3,000,000 shares (claim #15461), and on 2018-06-28 that the merger ended and directors resigned (claims #15477, #15476). What has not been said is which directors resigned and whether their departures were tied to the failed deal. The company also has not said what the shares were worth on the issue date, what charge was booked, or which party walked away and why.

The company disclosed on 2018-10-15 that CEO Merrick Okamoto received 2.5 million stock options, all of which accelerate on a change of control (claims #15385, #15386, #15387). What has not been said is the strike price, vesting schedule, and grant-date fair value. Nor has it explained how a single grant equal to a quarter of the 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460) was approved. That grant was made while the company was still working toward its compensation committee compliance deadline (claims #15468, #15437).

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459), alongside going-concern language about whether it had 12 months of cash (claim #15462). What has not been said is any operating detail: machine count, hashrate, power capacity in megawatts, site location, or the hosting or power counterparty. It also has not explained how the expansion would be funded given the stated cash runway risk.

The company disclosed on 2014-03-30 that patent revenue had begun and that it had signed licensing agreements with Siemens Technology and Zhone (claims #21014, #21018, #21022). It disclosed more than a dozen portfolio acquisitions the same day, including CyberFone, Mosaid, Intergraph, TeleCommunication Systems, Delphi and TT IP (claims #21015 through #21036). What has not been said is the dollar value of any license or the price paid for any portfolio. The company also has not disclosed revenue-share or contingent payments owed to the sellers, or the fee split with IP Navigation Group, which it named as a key supplier (claim #21024).

On 2018-07-12 the company listed Nasdaq deadlines for a compliance plan, board independence and compensation committee makeup (claims #15465, #15466, #15467, #15468). On 2018-08-21 it said compliance had been regained after appointing Berg, Rudolph and Thiel (claims #15438, #15435, #15436, #15437). What has not been said is which specific listing rule was cured on that date. The company also has not said whether the separate compensation committee deficiency, with a 2019-06-28 deadline, was resolved at the same time.

The company disclosed a 10-K filing deadline of 2018-04-15, which is recorded as delayed (claim #16645), at the same time Okamoto was named interim CEO (claims #16644, #16648). What has not been said is the actual filing date, the reason for the delay given in any NT 10-K, and whether the auditor or the leadership change contributed to it.

The company disclosed on 2018-08-13 that the terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claim #15461, and on 2018-06-28 that the merger was terminated and directors resigned, cited as claims #15477 and #15476. What has not been said is what value was assigned to those 3,000,000 shares, whether they were actually issued and to whom, how they were accounted for, and which directors resigned and whether their departures were connected to the failed deal.

The company disclosed a cryptocurrency mining expansion on 2018-08-13, cited as claim #15459, alongside a going-concern style warning about its 12-month cash runway, cited as claim #15462. What has not been said is the scale of that expansion in any usable unit (machine count, hashrate or megawatts), where the miners are physically hosted, under what power or hosting contract, and how the expansion was to be funded given the runway warning.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO at a $350,000 base salary, holding 2.5 million unvested options with change-of-control acceleration, cited as claims #15387, #15388 and #15386, after he served as interim CEO from 2018-04-01, cited as claim #16648. What has not been said is the strike price and vesting schedule of those options, how a change of control is defined for acceleration purposes, and whether Okamoto or his affiliates had any prior financing or related-party relationship with the company.

The company disclosed on 2014-03-30 that patent revenue generation had commenced and that it held licensing agreements with Siemens Technology and Zhone, cited as claims #21014, #21018 and #21022, while naming IP Navigation Group as a key supplier, cited as claim #21024. What has not been said is how much revenue each license produced, whether those licenses were one-time settlements or recurring royalties, and what share of recoveries IP Navigation Group retained under its contingency arrangement.

The company disclosed on 2018-07-12 a series of NASDAQ deadlines, including two new independent directors by 2019-01-07 and compensation committee compliance by 2019-06-28, cited as claims #15465, #15467 and #15468, then said on 2018-08-21 that compliance had been regained with three board appointments, cited as claims #15435 through #15438. What has not been said is whether the compensation committee requirement was satisfied by that date, or what equity the new directors received beyond Fred Thiel's 50,000 options, cited as claim #15396, given that he chairs the committee that sets the CEO's pay.

The company disclosed a series of patent portfolio acquisitions on 2014-03-30, including the Sampo, Mosaid, CyberFone, Siemens, TeleCommunication Systems, Intergraph, TT IP and Delphi portfolios (claims #21015 through #21036, several of them recorded twice). What has not been said is the consideration paid for each portfolio, whether any sellers kept revenue-share or contingent-payment rights, and how much of the acquired portfolio was already in active litigation. Nor has the company explained the economics of its 'key supplier' relationship with IP Navigation Group (claim #21024), including whether that firm is paid a contingency share of enforcement recoveries.

The company announced licensing agreements with Siemens Technology and Zhone on 2014-03-30 (claims #21018 and #21022). In the same filing it said patent revenue generation had commenced (claim #21014) and described ongoing enforcement litigation (claim #21025). What has not been said is the dollar value, term, or structure of either license (lump-sum or running royalty), or what share of reported revenue came from settlements versus negotiated licenses.

On 2018-06-28 and 2018-08-13 the company disclosed that its Global Bit Ventures merger was terminated with a break-up fee payable in 3,000,000 shares (claims #15477 and #15461). On the same day as the termination, it disclosed director resignations (claim #15476). What has not been said is which side triggered the termination, how the share-based fee was valued and who ultimately received the shares, and whether the departing directors had any relationship to Global Bit Ventures or disagreed with management.

In its 2018-08-13 filing the company flagged uncertainty about whether it had 12 months of cash runway (claim #15462) while announcing an expansion of cryptocurrency mining (claim #15459). What has not been said is how the expansion would be funded, how many miners it involved and what hash rate they targeted, and where the machines were to be sited. The filing also gave no power cost or hosting counterparty for the expansion.

On 2018-10-15 the company disclosed option grants of 50,000 shares each to directors Thiel, Berg and Rudolph and 100,000 shares to James Crawford (claims #15393 through #15396). These followed a 10,000,000-share 2018 Equity Incentive Plan reserve (claim #15460) and the appointment of Merrick D. Okamoto as interim CEO (claims #16644 and #16648). What has not been said is the strike price, vesting schedule and grant-date value of these options, or what role Crawford held to justify double the director grant. The company has also not described Okamoto's interim compensation or any related-party dealings involving him.

The company disclosed a delayed 10-K deadline of 2018-04-15 (claim #16645) and a sequence of NASDAQ compliance deadlines, including the compliance plan, listing rule and compensation committee deadlines (claims #15466 through #15468). It reported regaining compliance on 2018-08-21 (claim #15438). What has not been said is what caused the late annual report, whether the auditor changed or raised any going-concern or internal-control findings, and whether the employee count (claim #21026) ever stated an actual number.

The company disclosed on 2018-08-13 that its terminated Global Bit Ventures merger carried a share-based break-up fee of 3,000,000 shares, cited as claims #15461 and #15477. What has not been said is which party paid or received those shares, what they were worth at the termination date, whether they were actually issued, and why the deal collapsed about a month after directors resigned (#15476).

The company disclosed on 2018-10-15 that Merrick Okamoto was named Executive Chairman and CEO and received 2.5 million stock options with change-of-control acceleration, cited as claims #15384, #15385 and #15386. What has not been said is the strike price, the vesting schedule, or what triggers acceleration. That grant is a quarter of the 10,000,000-share 2018 Equity Incentive Plan reserve (#15460), and the company has also not said what Okamoto was paid while serving as interim CEO (#16648).

The company disclosed that directors resigned on 2018-06-28 (#15476) and that it faced Nasdaq board-independence and compensation-committee deadlines (#15465, #15466, #15467, #15468). It then disclosed on 2018-08-21 that compliance was regained with new appointments (#15435, #15436, #15437, #15438). What has not been said is who resigned, whether any resignation involved a disagreement with management, and whether the separate compensation-committee requirement due 2019-06-28 (#15468) was ever confirmed as satisfied.

The company disclosed on 2018-08-13 a cryptocurrency mining expansion alongside substantial doubt about a 12-month cash runway, cited as claims #15459 and #15462. What has not been said is how many machines it operates, the hash rate, where the mining sites are, what it pays for power and on what contract terms, and how the expansion will be financed given the going-concern language.

The company set a 10-K filing deadline of 2018-04-15, which is recorded as delayed (claim #16645), during an interim-CEO transition (#16644). What has not been said is when the report was actually filed, what reason the late-filing notice gave, and whether the delay was tied to audit, controls or restatement issues.

The company disclosed on 2014-03-30 that patent revenue had begun, that it held licensing agreements with Siemens and Zhone, and that it had a key supplier relationship with IP Navigation Group, cited as claims #21014, #21018, #21022 and #21024. What has not been said is how much revenue each license produced, what share of recoveries went to IP Navigation Group or to the original patent sellers, and what headcount sat behind the undisclosed employee count (#21026).

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 described a break-up fee of 3,000,000 shares (claim #15461); what has not been said is why the deal failed, what value was assigned to those shares on issuance, and whether any GBV mining equipment, hosting arrangements or personnel stayed with Marathon after termination. Who received the 3 million shares, and at what accounting value?

The company disclosed a cryptocurrency mining expansion on 2018-08-13 (claim #15459) in the same filing that flagged uncertainty over a 12-month cash runway (claim #15462); what has not been said is the number of machines, installed hashrate, site location, or the megawatts and price of the power contract behind the expansion. How was the buildout to be funded given the going-concern language, and where is the capacity physically located?

The company disclosed on 2018-10-15 that Merrick Okamoto, interim CEO since 2018-04-01 (claims #16644, #16648), was made Executive Chairman and CEO with 2.5 million options carrying change-of-control acceleration (claims #15384, #15385, #15386, #15387); what has not been said is the strike price, vesting schedule, the precise change-of-control trigger, what he was paid as interim CEO, or any related-party dealings. How do those 2.5 million options compare with the 10 million-share reserve under the 2018 Equity Incentive Plan (claim #15460) and with shares outstanding at the time?

The company disclosed director resignations on 2018-06-28 (claim #15476), a Nasdaq notice with a compliance-plan deadline of 2018-08-24 and a compensation-committee deadline of 2019-06-28 (claims #15466, #15467, #15468), and then regained compliance on 2018-08-21 by appointing Michael Berg, S. Michael Rudolph and Fred Thiel (claims #15435, #15436, #15437, #15438); what has not been said is how many directors left, why they left, and whether any of them raised disagreements with management. Did any departing director submit a resignation letter describing a dispute?

The company disclosed a 10-K filing deadline of 2018-04-15 that was missed (claim #16645) during the same period an interim CEO was installed (claim #16648); what has not been said is the cause of the delay, whether it involved an auditor change or a restatement, or the actual filing date. Did the late filing reflect an accounting or internal-control problem?

The company disclosed on 2014-03-30 a string of patent portfolio acquisitions from Siemens, Delphi, Intergraph, TeleCommunication Systems, Zhone, Mosaid, CyberFone and TT IP (claims #21015 through #21036), licensing agreements with Siemens and Zhone (claims #21018, #21022), IP Navigation Group as a key supplier (claim #21024), and the start of patent revenue (claim #21014); what has not been said is the purchase price and seller revenue-share terms for each portfolio, IP Navigation's contingency percentage, or which licensees produced the first revenue. How much of each litigation recovery (claim #21025) flowed back to the original patent sellers and to IP Navigation?

The company disclosed on 2014-03-30 that IP Navigation Group was a key supplier, cited as claim #21024, alongside active patent enforcement litigation (claim #21025); what has not been said is what economics govern that relationship. What share of any recovery goes to IP Navigation, how is it paid, and could the company run its enforcement campaigns without it?

The company disclosed on 2014-03-30 a string of patent portfolio acquisitions from Siemens, Delphi, Intergraph, TeleCommunication Systems, Zhone, Mosaid, CyberFone and TT IP, cited as claims #21031, #21036, #21033, #21032, #21035, #21029, #21030 and #21034, together with licensing agreements with Siemens and Zhone (claims #21018 and #21022); what has not been said is the purchase price of each portfolio or whether the sellers keep a back-end revenue share. Are the Siemens and Zhone licenses a license-back of the same patents, and on what royalty terms? The same filing announced that patent revenue had begun (claim #21014) and gave an employee count (claim #21026), but neither figure is quantified in the record.

The company disclosed on 2018-06-28 that its Global Bit Ventures merger was terminated, cited as claim #15477, with a 3,000,000-share break-up fee (claim #15461) and director resignations on the same date (claim #15476); what has not been said is which party triggered the termination and why. Who received the break-up shares, at what value were they recorded, and did any departing director cite a disagreement with management?

The company disclosed on 2018-08-13 a plan to expand cryptocurrency mining, cited as claim #15459, while also flagging uncertainty about its 12-month cash runway (claim #15462); what has not been said is how the expansion will be sized, sited or funded. How many machines and what hashrate are targeted, at which facility, with what power contract and cost per kWh, and against what cash balance and monthly burn?

The company disclosed on 2018-10-15 that 2.5 million unvested options carry change-of-control acceleration, cited as claim #15386, alongside director grants to Crawford, Rudolph and Berg (claims #15393, #15394 and #15395) under a 10,000,000-share 2018 Equity Incentive Plan (claim #15460); what has not been said is who holds the 2.5 million accelerating options. Interim CEO Merrick Okamoto was named on 2018-04-01 (claim #16648), but his compensation and his tenure as 'interim' remain undisclosed. Did the compensation committee under Fred Thiel (claim #15437) approve the acceleration terms, and what are the strike prices and vesting schedules?

The company disclosed on 2018-04-01 a 10-K filing deadline of 2018-04-15, cited as claim #16645, which is recorded as delayed; what has not been said is the cause of the delay. Was it tied to the auditor, to the CEO transition recorded the same day (claim #16644), or to an accounting issue, and was any restatement or material weakness later reported?

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options (claim #15389), and separately that 2,500,000 options were issued with another 2,500,000 unvested and subject to change-of-control acceleration (claims #15385, #15386). What has not been said is the strike price, the vesting schedule for the unvested tranche, or how 'change of control' is defined. It is also unclear why a single executive holds half of the 10,000,000-share 2018 Equity Incentive Plan reserve (claims #15460, #15373), and whether that grant was the reason Proposal No. 5 was withdrawn under Nasdaq Rule 5635(b) (claim #15372).

The company disclosed a 26,700-square-foot Quebec mining facility on 2019-06-02 and a lease through November 2022 carrying a $120,150 figure on 2019-07-01 (claims #15230, #15219, #15229). What has not been said is the facility's address, its contracted power capacity in megawatts, the electricity supplier and tariff, and whether $120,150 is a monthly, annual, or total-term amount. It is also unclear whether the 1,400 deployed mining units (claim #15220) all sit at that site, and what their model and aggregate hashrate are.

The company disclosed on 2018-06-28 and 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee of 3,000,000 shares (claims #15477, #15461), on the same date that directors resigned (claim #15476). What has not been said is who received those shares, what value the company assigned to them, and whether any of the resigning directors were affiliated with Global Bit Ventures.

The company disclosed on 2019-07-01 a $250,000 cryptocurrency patent acquisition from XpresSpa Group alongside a 30% waterfall interest in a patent special purpose entity (claims #15218, #15223). What has not been said is who holds the remaining interest in that entity, how the waterfall is structured, and whether any director or officer has a relationship with XpresSpa Group that would make this a related-party transaction.

The company disclosed a 12-month cash runway uncertainty on 2018-08-13 (claim #15462), then asserted operating cash sufficiency for twelve months on 2019-07-31 (claim #15197). In between, it set up a $7,472,417 at-the-market program with H.C. Wainwright at a 3% commission (claims #15202, #15203, #15204, #15228). What has not been said is how much of the ATM has actually been drawn, at what average price, and what the monthly cash burn is. With only three full-time employees (claim #15222), the company has also not named the outside providers handling financial operations and 24-hour security (claims #15200, #15199), or what it pays them.

The company disclosed a string of Nasdaq compliance deadlines in 2018 and 2019, including the $1.00 minimum bid, minimum equity, and board-independence requirements (claims #15466, #15467, #15383, #15244, #15196), and obtained shareholder authorization for a reverse stock split by 2018-12-31 (claim #15332). What has not been said is the approved split ratio, whether the split was executed and when, and whether Nasdaq has formally confirmed that each deficiency was cured.

The company disclosed on 2018-10-15 that Merrick Okamoto received stock options on 5,000,000 shares (claim #15389), and separately that 2.5 million options were issued to the CEO (claim #15385) and 2.5 million unvested options carry change-of-control acceleration (claim #15386). What has not been said is the strike price, the vesting schedule for the unaccelerated tranche, and whether any compensation was paid for the roughly six months he served as interim CEO from 2018-04-01 (claims #16644, #16648). What change-of-control trigger releases the second 2.5 million, and how does the full grant compare with the 10,000,000-share 2018 Equity Incentive Plan reserve (claims #15460, #15373)?

The company disclosed on 2019-07-01 a Quebec data center lease through November 2022 at 120,150 USD (claim #15219), a 26,700 square foot facility (claim #15230), and 1,400 deployed Bitcoin mining units (claim #15220). What has not been said is whether 120,150 USD is a monthly, annual, or total-term figure, who the landlord or hosting provider is, and the facility's power allocation in megawatts. Nor has the company given the machine model, aggregate hashrate, or electricity rate. What is the contracted power capacity at the site, and at what cost per kilowatt-hour?

The company disclosed on 2019-07-01 that it bought a cryptocurrency patent from XpresSpa Group for 250,000 USD (claim #15218) together with a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the remaining 70%, what the SPE is called, where the company sits in the waterfall's priority, and whether any enforcement or licensing actions have been filed. Has the SPE returned any recoveries against the 250,000 USD outlay?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee paid in 3,000,000 shares (claims #15461, #15477). On the termination date of 2018-06-28, directors also resigned (claim #15476). What has not been said is which party terminated and why, whether the 3,000,000 shares were actually issued and at what value, and which directors resigned and whether their departures were connected to the deal. Were any of the resigning directors affiliated with Global Bit Ventures?

The company disclosed on 2019-07-18 an at-the-market program with H.C. Wainwright capped at 7,472,417 USD, with a 3% commission (claims #15202, #15203, #15204). This came alongside repeated doubt about a twelve-month cash runway (claims #15462, #15197). What has not been said is how much of the ATM had been sold as of each subsequent quarter, at what average price, and what the monthly cash burn was against the cash on hand. How many months of operations did the ATM proceeds actually fund?

The company disclosed on 2019-07-31 two service arrangements, 24-hour security services (claim #15199) and outsourced financial operations management (claim #15200), while reporting only three full-time employees (claim #15222). What has not been said is who the counterparties are, what the contracts cost, and whether any of them are related parties to management or directors. Who performs the company's finance function, and how is that provider compensated?

The company disclosed Merrick Okamoto's appointment as Executive Chairman and CEO on 2018-10-15 with a $350,000 base salary, cited as claims #15387 and #15388. That same filing puts the CEO's option grant at 2.5 million shares in claim #15385 and 5 million shares in claim #15389, and separately cites 2.5 million unvested options carrying change-of-control acceleration in claim #15386. What has not been said is how these figures reconcile: is the total 5 million with half unvested, or two separate 2.5 million tranches? At what strike price, and on what vesting schedule? And how does the grant fit within the 10,000,000-share 2018 Equity Incentive Plan reserve cited in claims #15460 and #15373?

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated and that directors resigned, cited as claims #15477 and #15476. It later cited a 3,000,000-share break-up fee in claim #15461. What has not been said is who received those shares and at what value they were recorded. Nor has it said why the deal was terminated, or whether the resigning directors had any affiliation with Global Bit Ventures.

The company disclosed a 26,700-square-foot Quebec mining facility on 2019-06-02 and a lease running through November 2022 at $120,150 on 2019-07-01, cited as claims #15230, #15229 and #15219. What has not been said is whether $120,150 is a monthly, annual or total-term figure. The company has also not given the site's contracted power capacity in megawatts, its electricity rate, or its utility interconnection terms, even though mining economics depend on them. Beyond the Granby reference in claim #15198, the facility's physical address is also unstated.

The company disclosed on 2019-07-31 that bitcoin mining at the Granby facility, 24-hour security and financial operations management were handled through partnerships, cited as claims #15198, #15199 and #15200. The company also reported only three full-time employees on 2019-07-01, cited as claim #15222. What has not been said is who these counterparties are, what they are paid, and whether any of them are related parties. With so small a staff, those operating agreements effectively run the business.

The company disclosed a deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220. What has not been said is the miner model, the aggregate hashrate in TH/s or PH/s, how many units are actually hashing versus merely installed, and what the fleet cost. Without those figures the 1,400 count cannot be turned into expected revenue.

The company disclosed a $250,000 purchase of cryptocurrency patents from XpresSpa Group and a 30% waterfall interest in a patent special purpose entity on 2019-07-01, cited as claims #15218 and #15223. It also disclosed a $7,472,417 at-the-market program with H.C. Wainwright on 2019-07-18, cited as claims #15202 and #15203, while flagging doubts about having twelve months of cash, cited as claims #15462 and #15197. What has not been said is who holds the other 70% of the patent entity and whether that holder is related to management. The company also has not said how much of the ATM capacity has been drawn to date, or at what average price.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated, cited as claim #15477, and on 2018-08-13 described a share-based break-up fee of 3,000,000 shares, cited as claim #15461. What has not been said is who received those shares, what they were worth at issuance, whether they were ever actually issued given the fee's delayed status, and why the deal fell apart. Did the termination coincide with the director resignations disclosed the same day, cited as claim #15476, whose names and reasons were also not given?

The company disclosed Merrick Okamoto's CEO compensation on 2018-10-15, citing 2,500,000 options in claim #15385, 5,000,000 options in claim #15389, and 2,500,000 unvested options with change-of-control acceleration in claim #15386. What has not been said is how these figures reconcile. Is the grant 5,000,000 split evenly between vested and unvested tranches, or 2,500,000 total? What are the strike price, the vesting schedule, and the triggers that would accelerate the unvested half? The same filing combines the CEO and Chairman roles, cited as claims #15384 and #15387, so which independent body approved an award of this size relative to a $350,000 base salary, cited as claim #15388?

The company disclosed a 26,700-square-foot Quebec mining facility on 2019-06-02, cited as claim #15230, and a lease running through November 2022 at a figure of $120,150, cited as claim #15219. What has not been said is whether that $120,150 is monthly, annual, or the total lease cost. The filings also give no street address, no landlord name, no contracted power capacity in megawatts, and no Hydro-Québec rate or interconnection terms. How much of the 1,400 deployed mining units, cited as claim #15220, sit at this site, and what model and aggregate hashrate do they represent?

The company disclosed service partnerships for 24-hour security and for financial operations management on 2019-07-31, cited as claims #15199 and #15200. What has not been said is who the counterparties are, what they are paid, and whether any of them is related to management or directors. This matters for a company reporting only 3 full-time employees, cited as claim #15222: how much of its operations are outsourced to these unnamed providers?

The company disclosed a $250,000 patent acquisition from XpresSpa Group and a 30% waterfall interest in a patent special purpose entity on 2019-07-01, cited as claims #15218 and #15223. What has not been said is the name of the SPE, who holds the remaining 70%, and how the waterfall ranks Marathon against other holders. Has the portfolio produced any licensing or litigation proceeds so far?

The company disclosed a string of Nasdaq compliance deadlines between 2018 and 2019, cited as claims #15467, #15383, #15244, and #15196. It also disclosed a reverse split authorization, cited as claim #15332, and a $7,472,417 at-the-market program with H.C. Wainwright, cited as claims #15202 and #15203. What has not been said is whether the reverse split was carried out and at what ratio, and how much of the ATM capacity has been sold and at what average price. Given the recurring going-concern language, cited as claims #15462 and #15197, how many months of cash runway does management actually project?

The company disclosed on 2018-06-28 that its merger with Global Bit Ventures was terminated (claim #15477), and on 2018-08-13 described a share-based break-up fee of 3,000,000 shares (claim #15461); what has not been said is who received those shares, how they were valued at issuance, and whether the fee was ever fully settled, given the claim still carries a delayed status.

The company disclosed a 26,700-square-foot Quebec mining facility on 2019-06-02 (claim #15230), a data center lease there through November 2022 at 120,150 USD on 2019-07-01 (claim #15219), 1,400 deployed mining units (claim #15220), and Bitcoin mining at a Granby facility described as a partnership on 2019-07-31 (claim #15198). What has not been said is whether the 120,150 USD is a monthly, annual, or total-term figure, who the landlord and hosting partner are, what contracted power capacity in megawatts and utility rate back the site, and what model and aggregate hashrate the 1,400 machines represent.

The company disclosed on 2018-10-15 that Merrick Okamoto became Executive Chairman and CEO at a 350,000 USD base salary (claims #15387 and #15388). The same filing references 2.5 million options issued (claim #15385), 2.5 million unvested options with change-of-control acceleration (claim #15386), and a 5,000,000-share option figure (claim #15389). What has not been said is how these figures reconcile, what the strike prices and vesting schedules are, and what share price or transaction value would trigger the acceleration.

The company disclosed on 2019-07-01 a 250,000 USD cryptocurrency patent acquisition from XpresSpa Group (claim #15218) and a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the remaining 70% of that entity, how the waterfall is ordered, and whether any officer or director has a relationship with the seller or the SPE that would make this a related-party transaction.

The company disclosed on 2019-07-18 an at-the-market program with H.C. Wainwright sized at 7,472,417 USD at a 3% commission (claims #15202, #15203 and #15204). It also flagged twelve-month cash runway uncertainty in 2018 and 2019 (claims #15462 and #15197) and reported only three full-time employees (claim #15222). What has not been said is how much of the ATM was actually drawn, at what average price, and how many months of operating and power costs the proceeds covered.

The company disclosed a string of Nasdaq compliance deadlines between 2018 and 2019, all now marked delayed (claims #15466, #15467, #15383, #15244 and #15196), alongside a reverse stock split authorization due by 2018-12-31 (claim #15332). What has not been said is whether the reverse split was executed and at what ratio, and which deficiencies, whether minimum bid, stockholders' equity, or committee composition, were formally cured versus extended. The company also disclosed a 'financial operations management' partnership on 2019-07-31 (claim #15200) without naming the counterparty or the fee terms.

The company disclosed a Quebec mining facility of 26,700 square feet on 2019-06-02 (claim #15230), a data center lease through November 2022 carrying a figure of $120,150 on 2019-07-01 (claim #15219), a deployment of 1,400 mining units (claim #15220), and Bitcoin mining at a Granby facility on 2019-07-31 (claim #15198). What has not been said is whether the $120,150 is a monthly, annual or total lease commitment, who the landlord or hosting counterparty is, and what contracted power capacity in megawatts and utility rate back the site. Nor has the company given the model, hashrate or ownership status of the 1,400 machines, so how much capacity is actually being leased?

The company disclosed the termination of the Global Bit Ventures merger with a 3,000,000-share break-up fee on 2018-08-13 (claims #15461 and #15477), and on the same 2018-06-28 date it disclosed that directors had resigned (claim #15476). What has not been said is whether those shares were ever issued, which named holders received them, what fair value was booked, and whether the director resignations were connected to the failed deal. Who were the departing directors, and did they have any relationship with Global Bit Ventures?

The company disclosed on 2018-10-15 that Merrick Okamoto received 2.5 million options with change-of-control acceleration (claims #15385 and #15386), and in the same filing cited 5,000,000 options for him (claim #15389), alongside a $350,000 base salary (claim #15388). The 2018 Equity Incentive Plan reserves 10,000,000 shares (claims #15460 and #15373). What has not been said is whether 5 million is the total grant and 2.5 million the unvested tranche, or whether these are separate awards, and what the strike prices and vesting triggers are. It is also unclear how much of the 10-million-share reserve the October 2018 grants to Okamoto, Lieberman, Crawford and the three new directors consumed (claims #15392 through #15396).

The company disclosed on 2018-11-13 that it withdrew Proposal No. 5 because of NASDAQ Rule 5635(b) (claim #15372), a rule that governs issuances resulting in a change of control. On 2018-12-11 it also sought approval for non-public issuances above 20 percent (claim #15333). What has not been said is which investor or related party would have crossed the change-of-control threshold under the withdrawn proposal, and on what terms the replacement 20-percent issuance was placed and to whom.

The company disclosed a $250,000 purchase of cryptocurrency patents from XpresSpa Group on 2019-07-01 (claim #15218), together with a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the remaining 70% of that entity, what licensing or litigation revenue it has produced, and whether any Marathon officer or director has a relationship with XpresSpa or the entity's other participants.

The company disclosed a $7,472,417 at-the-market program with H.C. Wainwright on 2019-07-18 (claims #15202 and #15203), paying a 3% commission (claim #15204). It also disclosed going-concern-style cash-runway uncertainty (claims #15462 and #15197), a three-person workforce (claim #15222), and a series of delayed Nasdaq bid-price and equity deadlines alongside a reverse-split authorization (claims #15383, #15196 and #15332). What has not been said is how much of the at-the-market capacity has been drawn and at what average price. The company also has not said whether the reverse split was effected and at what ratio, or whether Nasdaq formally confirmed that the minimum-equity deficiency was cured.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), and on 2018-08-13 that the termination carried a share-based break-up fee of 3,000,000 shares (claim #15461). What has not been said is whether those shares were actually issued, to which GBV holders, at what per-share value they were recorded, and how the charge flowed through the income statement.

The company disclosed on 2018-10-15 that Merrick Okamoto was named Executive Chairman and CEO at a $350,000 base salary (claims #15387, #15388). The same filing puts his option grant at 5,000,000 shares (claim #15389) in one place and 2,500,000 shares in another (claim #15385), with 2,500,000 unvested options carrying change-of-control acceleration (claim #15386). What has not been said is how those figures reconcile, along with the strike price, vesting schedule and grant-date fair value. A related open question is what Proposal No. 5, withdrawn on 2018-11-13 over Nasdaq Rule 5635(b) change-of-control concerns (claim #15372), would have done and which investor it would have benefited.

The company disclosed on 2019-07-01 a Quebec data center lease through November 2022 tied to a figure of $120,150 (claim #15219), a 26,700 square foot facility (claim #15230), and 1,400 bitcoin miners deployed (claim #15220). What has not been said is whether $120,150 is a monthly, annual or total-term figure. The filings also omit the miner model, the aggregate hashrate, the contracted power in megawatts and the electricity rate or utility counterparty.

The company disclosed on 2019-07-31 that bitcoin mining at the Granby facility, 24-hour security, and financial operations management are all handled through partnerships (claims #15198, #15199, #15200), while reporting only three full-time employees (claim #15222). What has not been said is who these counterparties are, what they are paid, and whether any of them is related to management or directors.

The company disclosed on 2019-07-01 that it bought cryptocurrency patents from XpresSpa Group for $250,000 (claim #15218) and holds a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is the name of that entity and who holds the other 70%. The company also has not described the waterfall terms or said whether any licensing or enforcement revenue has been realized.

The company disclosed twelve-month cash runway uncertainty on 2018-08-13 and again on 2019-07-31 (claims #15462, #15197). It also disclosed a $7,472,417 at-the-market program with H.C. Wainwright at a 3% commission (claims #15202, #15203, #15204), a reverse split authorization (claim #15332) and approval for a non-public issuance above the 20% threshold (claim #15333). What has not been said is how much of the ATM has been drawn and at what average price. The filings also leave out the reverse split ratio actually applied and the identity and pricing of the non-public issuance counterparty. Note that the record provided ends mid-entry at claim #15196, so later disclosures may close some of these gaps.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated, cited as claim #15477, and on 2018-08-13 described a share-based break-up fee of 3,000,000 shares, cited as claim #15461. What has not been said is which party paid the fee, whether the shares were actually issued, and what value per share was used to book the fee.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options, cited as claim #15389. It separately disclosed 2,500,000 options issued and 2,500,000 unvested options with change-of-control acceleration, cited as claims #15385 and #15386. What has not been said is the strike price, the vesting schedule, and the grant-date fair value. A reader would also want to know why half of the 10,000,000-share 2018 Equity Incentive Plan reserve, cited as claims #15460 and #15373, went to a single executive.

The company disclosed on 2018-11-13 that Proposal No. 5 was withdrawn to comply with NASDAQ Rule 5635(b), cited as claim #15372. It disclosed on 2018-12-11 that shareholders approved a non-public issuance above the 20 percent threshold, cited as claim #15333. What has not been said is who the intended investor or acquirer was, and how much ownership or control that party would have gained.

The company disclosed on 2019-07-01 a Quebec data center lease through November 2022 at 120,150 USD, cited as claim #15219. On 2019-06-02 it gave a facility size of 26,700 square feet, cited as claim #15230, and it disclosed 1,400 mining units deployed, cited as claim #15220. What has not been said is whether the lease figure is monthly, annual, or the total for the term. The filings also omit the site's contracted power capacity in megawatts, the utility rate or interconnection arrangement, and the machine models and aggregate hashrate behind the 1,400 units.

The company disclosed on 2019-07-31 partnerships covering Bitcoin mining operations at the Granby facility, 24-hour security, and financial operations management, cited as claims #15198, #15199 and #15200. What has not been said is who those counterparties are, what they are paid, and whether any of them are related parties. This matters because the company reported only three full-time employees on 2019-07-01, cited as claim #15222.

The company disclosed on 2019-07-01 a 250,000 USD patent acquisition from XpresSpa Group and a 30 percent waterfall interest in a patent special purpose entity, cited as claims #15218 and #15223. It also flagged 12-month cash-runway uncertainty on 2018-08-13, cited as claim #15462, and set up a 7,472,417 USD at-the-market program with H.C. Wainwright on 2019-07-18, cited as claims #15202 and #15203. What has not been said is who holds the other 70 percent of the patent entity and how the waterfall is ordered. The filings also do not say how much of the ATM had been drawn, at what average price, or how many months of runway that capital actually buys.

The company disclosed on 2019-07-31 that Bitcoin mining at its Granby facility is run through a partnership that also covers 24-hour security and financial operations management, cited as claims #15198, #15199 and #15200. What has not been said is who that operator is, how it is paid (a fixed fee, a share of mined coin, or a markup on power), and whether it has any relationship with management. This matters because the company reported only three full-time employees on 2019-07-01 (claim #15222).

The company disclosed on 2019-06-02 and 2019-07-01 a 26,700-square-foot Quebec data center leased through November 2022 at a cost of $120,150, cited as claims #15230, #15229 and #15219. What has not been said is whether $120,150 is a monthly, annual, or total-term figure. The company also has not named the landlord, given the facility's power capacity in megawatts, or said anything about the Hydro-Québec tariff or interconnection that governs its electricity cost.

The company disclosed on 2019-07-01 the deployment of 1,400 Bitcoin mining units, cited as claim #15220. What has not been said is the machine model, the total hashrate in TH/s, the power draw, or the cost and financing of the fleet. A first-year analyst would need those numbers to test the Ether mining plans in claim #15231 and the cash-runway uncertainty flagged on 2018-08-13 in claim #15462.

The company disclosed on 2018-10-15 a 5,000,000-share option grant to Merrick Okamoto, split into 2.5 million issued and 2.5 million unvested with change-of-control acceleration, cited as claims #15389, #15385 and #15386. That grant alone equals half of the 10,000,000-share 2018 Equity Incentive Plan reserve in claims #15460 and #15373. What has not been said is the strike price, the vesting schedule, and the change-of-control definition, or why Proposal No. 5 was withdrawn on 2018-11-13 over NASDAQ Rule 5635(b) (claim #15372).

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee of 3,000,000 shares, and on 2018-06-28 that directors resigned, cited as claims #15461, #15477 and #15476. What has not been said is who received the shares, the value placed on them, which directors left, and whether the resignations were tied to the failed deal.

The company disclosed on 2018-12-11 shareholder approval for a reverse stock split and a non-public issuance of up to 20 percent, cited as claims #15332 and #15333. On 2019-07-01 it disclosed a $250,000 patent purchase from XpresSpa Group carrying a 30 percent waterfall interest in a patent special purpose entity, cited as claims #15218 and #15223. What has not been said is the split ratio, the identity and pricing of the private-placement investors, the name of the SPE, and who holds the other 70 percent of its waterfall.

Marathon disclosed a Quebec mining facility measured at 26,700 square feet on 2019-06-02, cited as claim #15230, and a data center lease running through November 2022 carrying a figure of $120,150 on 2019-07-01, cited as claim #15219; what has not been said is any power number at all — no contracted megawatts, no Hydro-Quebec interconnection or energization status, no rate per kilowatt-hour, and no street address for the Granby site — nor whether the $120,150 is monthly rent, annual rent, or the aggregate obligation over the full lease term. Square footage is not capacity, and a reader cannot convert one into the other.

The company disclosed deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220, alongside mining operations at the Granby facility on 2019-07-31, cited as claim #15198; what has not been said is the machine model, the per-unit and aggregate hashrate, the power draw implied by those 1,400 units, how many were actually energized and hashing versus merely delivered, and how that fleet reconciles against the 26,700 square feet in claim #15230. A first-year analyst's first question on the next call is simply: what is total installed and operating terahash, and on what date did it reach that level?

The company disclosed a partnership for 24-hour security services and one for financial operations management, both on 2019-07-31, cited as claims #15199 and #15200, and mining operations at Granby under a partnership structure, cited as claim #15198; what has not been said is who any of these counterparties are, what they are paid, whether the arrangements are fixed-fee or revenue-share, their termination provisions, or whether any counterparty is an affiliate of a director or officer. Announcing an operating dependency while withholding the name of the operator leaves the reader unable to assess concentration or related-party risk.

The company disclosed acquiring a cryptocurrency patent from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, and a 30% waterfall interest in a patent special purpose entity on the same date, cited as claim #15223; what has not been said is who holds the other 70% of the SPE, where 30% sits in the waterfall and behind what preferred return or litigation-funding recovery, what enforcement or contingency-fee obligations Marathon bears, and what board, officer, or shareholder overlap existed between Marathon and XpresSpa Group at the time the $250,000 price was set. A purchase price from a counterparty with potential affiliation, paired with an unexplained residual interest, is precisely the fact pattern that requires related-party framing.

The company disclosed, all on 2018-10-15, a $350,000 base salary and 5,000,000 options for Merrick Okamoto (claims #15388 and #15389), a separate 2.5 million option grant to the CEO (claim #15385), 2.5 million unvested options carrying change-of-control acceleration (claim #15386), a $180,000 salary and 200,000 options for CFO David Lieberman (claims #15391 and #15392), and director grants of 50,000 to 100,000 shares each (claims #15393 through #15396) — against a 2018 Equity Incentive Plan reserve of 10,000,000 shares (claims #15460 and #15373) and a company that reported three full-time employees on 2019-07-01 (claim #15222). What has not been said is the exercise price, vesting schedule, and grant-date fair value of any of these awards, whether claims #15385, #15386 and #15389 describe overlapping tranches of a single package or genuinely additive grants, what percentage of fully diluted shares the total represents, and how a reserve of that size is justified against a three-person payroll.

The company disclosed a shelf registration of $7,472,417 on 2019-06-02 and again on 2019-07-01, cited as claims #15228 and #15217, an At The Market program of the identical $7,472,417 with H.C. Wainwright at a 3% commission on 2019-07-18, cited as claims #15203, #15202 and #15204, and separately flagged 12-month cash runway uncertainty on 2018-08-13, cited as claim #15462; what has not been said is how much of that capacity has actually been drawn, at what average price, how many shares were issued, and how the proceeds map to the equipment and lease commitments in claims #15220 and #15219. Related and still open: the reverse stock split authorized on 2018-12-11, cited as claim #15332, was disclosed without a ratio or an effective date even though the $1.00 Nasdaq minimum bid price deadline of 2019-04-15, cited as claim #15383, has since passed.

The company disclosed a 26,700 square foot mining facility in Quebec on 2019-06-02, cited as claim #15230, with a lease running through November 2022 at a figure of $120,150 disclosed on 2019-07-01, cited as claim #15219, and operations described as being at a "Granby facility" in claim #15198. What has not been said is whether $120,150 is a monthly, annual, or aggregate lease obligation, what the contracted electrical capacity of the site is in megawatts, what the all-in power price per kilowatt-hour is under the Hydro-Québec arrangement, and whether the interconnection is firm or interruptible — square footage is not capacity, and for a miner the megawatt and power-price figures are the only ones that determine unit economics.

The company disclosed deployment of 1,400 units of bitcoin mining equipment on 2019-07-01, cited as claim #15220, alongside a three-person full-time headcount in claim #15222. What has not been said is the make and model of the machines, aggregate hashrate in petahashes per second, efficiency in joules per terahash, the capitalized cost of the fleet, and critically whether all 1,400 units were energized and hashing at the Granby site as of that date or merely purchased and in transit — and how a three-employee organization is staffing a facility of that size, i.e. whether operations are self-run or outsourced to a third-party operator whose name and fee structure have not been identified.

The company disclosed a cryptocurrency patent acquisition from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30% waterfall interest in a patent special purpose entity, cited as claim #15223. What has not been said is whether any Marathon officer or director held a concurrent board seat, executive role, or equity position at XpresSpa Group at the time of the transaction, whether the $250,000 price was supported by an independent valuation or approved by disinterested directors, who holds the remaining 70% of the SPE, and where in the waterfall Marathon's 30% sits relative to litigation-funding and counsel recoveries.

The company disclosed on 2018-10-15 that Merrick Okamoto received a $350,000 base salary, cited as claim #15388, plus 5,000,000 stock options, cited as claim #15389, and a separate 2,500,000-option grant, cited as claim #15385, of which 2,500,000 unvested options accelerate on a change of control, cited as claim #15386 — against a CFO package of $180,000 and 200,000 options, cited as claims #15391 and #15392, and director grants of 50,000 to 100,000 options in claims #15393 through #15396. What has not been said is the exercise price, vesting schedule, and grant-date fair value of any of these awards, whether the 5,000,000 and 2,500,000 figures are cumulative or overlapping descriptions of the same grant, and what percentage of fully diluted shares outstanding the CEO's awards represent — a question sharpened by the fact that the 2018 Equity Incentive Plan reserved only 10,000,000 shares in total, cited as claims #15460 and #15373.

The company disclosed termination of the Global Bit Ventures merger on 2018-06-28, cited as claim #15477, with a share-based break-up fee of 3,000,000 shares, cited as claim #15461, and simultaneous director resignations, cited as claim #15476. What has not been said is who received the 3,000,000 shares and at what deemed value, whether the resigning directors were GBV affiliates or designees, whether any of them or their affiliates were counterparties to the break-up fee, and what diligence finding or financing failure actually caused a signed merger to be unwound.

The company disclosed an At The Market program and shelf registration sized at exactly $7,472,417 on three separate occasions — claims #15228, #15217, and #15203 — with a 3% agent commission to H.C. Wainwright, cited as claim #15204, while simultaneously flagging twelve-month cash runway uncertainty in claims #15462 and #15197 and seeking authority to issue over 20% of shares non-publicly in claim #15333 and a reverse stock split in claim #15332. What has not been said is the public float calculation underlying that unusually precise cap and whether it reflects the one-third baby-shelf limitation, how much of the $7.47 million had actually been drawn as of each disclosure, the average price and share count issued to date, and what month-end cash balance and monthly operating burn the going-concern language in claim #15197 is measured against.

The company disclosed regaining Nasdaq compliance on 2018-08-21, cited as claim #15438, after a sequence of compliance-plan and director-appointment deadlines in claims #15465 through #15468, then disclosed a fresh listing rule violation on 2019-05-23, cited as claim #15243, a new plan deadline in claim #15244, a minimum stockholders' equity filing obligation in claim #15196, and a separate $1.00 minimum bid price deadline in claim #15383. What has not been said is the actual stockholders' equity figure versus the applicable threshold, the specific rule cited in the 2019 notice as distinct from the 2018 notices, and what closing bid price the stock held when the reverse split authorization in claim #15332 lapsed past its 2018-12-31 target without a disclosed ratio or effective date.

The company disclosed a 26,700 square foot Quebec mining facility leased through November 2022 on 2019-06-02 and 2019-07-01, cited as claims #15230, #15229 and #15219, with a figure of $120,150 attached to the lease and 1,400 mining units deployed under claim #15220; what has not been said is the street address of the Granby site, whether $120,150 is monthly, annual or aggregate rent, what interconnected power capacity in megawatts the site holds, what the all-in power price per kilowatt-hour is, and what aggregate hashrate those 1,400 machines represent — so on what basis should a reader model the facility's cost per terahash?

The company disclosed mining operations and 24-hour security services at the Granby facility on 2019-07-31, cited as claims #15198 and #15199, and separately disclosed a total of three full-time employees on 2019-07-01, cited as claim #15222; what has not been said is who the counterparties to those operating and security arrangements are, whether they are hosting providers, affiliates of the landlord, or related parties, and what the contracted fees and termination rights are — who actually runs a 1,400-machine site that the issuer staffs with three people?

The company disclosed the acquisition of cryptocurrency patents from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30 percent waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is whether overlapping officers or directors sat on both sides of that transaction, whether the price was supported by an independent valuation or a disinterested board vote, who holds the other 70 percent of the SPE, and what the waterfall's priority tiers and expense-recovery thresholds are before Marathon sees a dollar — was this an arm's-length purchase or a related-party transfer?

The company disclosed compensation for Merrick Okamoto on 2018-10-15 comprising a $350,000 base salary and option grants cited as claims #15388, #15389, #15385 and #15386, totaling as recorded 5,000,000 shares plus a separately described 2,500,000-share CEO grant and 2,500,000 unvested options carrying change-of-control acceleration, alongside a 2018 Equity Incentive Plan reserve of 10,000,000 shares cited as claims #15460 and #15373; what has not been said is whether the 5,000,000 figure subsumes or sits on top of the 2,500,000 grants, what the exercise prices, vesting schedules and grant-date fair values are, and what percentage of the plan pool and of shares outstanding went to one executive in a single day — is the CEO's grant one award or three?

The company disclosed termination of the Global Bit Ventures merger on 2018-06-28, cited as claim #15477, alongside director resignations the same day, cited as claim #15476, and then disclosed on 2018-08-13 a 3,000,000-share break-up fee tied to that termination, cited as claim #15461; what has not been said is the dollar value ascribed to those shares, whether they were actually issued and when, who the recipients were, and whether any resigning director or their affiliates held an interest in GBV or received any portion of the fee — why did the fee surface six weeks after the termination was announced?

The company disclosed a shareholder-authorized reverse stock split on 2018-12-11 with a stated deadline of 2018-12-31, cited as claim #15332, against a Nasdaq minimum bid price deficiency due 2019-04-15, cited as claim #15383, a regained-compliance announcement on 2018-08-21, cited as claim #15438, a fresh listing violation on 2019-05-23, cited as claims #15243 and #15244, and a minimum stockholders' equity filing deadline of 2019-11-13, cited as claim #15196; what has not been said is the split ratio, whether it was ever effected and on what date, and which specific listing standard the 2019 deficiency concerns — how many separate Nasdaq deficiencies were open simultaneously, and what is the current stockholders' equity shortfall in dollars?

The company disclosed an At The Market program of exactly $7,472,417 with H.C. Wainwright on 2019-07-18 at a 3 percent commission plus $25,000 of counsel expenses, cited as claims #15203, #15202, #15204, #15205 and #15217, while separately flagging twelve-month cash sufficiency uncertainty on 2018-08-13 and 2019-07-31, cited as claims #15462 and #15197; what has not been said is the cash balance and monthly operating burn underlying the going-concern language, how much of the ATM had been sold and at what average price as of each filing, whether the oddly precise ceiling reflects a baby-shelf General Instruction I.B.6 limit that shrinks with the stock, and at what bitcoin price the 1,400 deployed machines cover cash costs — how many months of runway does the ATM actually buy?

The company disclosed on 2019-07-01 that it acquired cryptocurrency patents from XpresSpa Group for $250,000, cited as claim #15218, and simultaneously disclosed a 30% waterfall interest in a patent special purpose entity, cited as claim #15223. What has not been said is who holds the other 70% of that SPE, whether any Marathon director or officer had a concurrent economic or board relationship with XpresSpa Group at the time of the purchase, and what the waterfall tiers actually are — at what recovery level does Marathon's 30% begin to pay, and were any related-party approvals or fairness procedures used to set the $250,000 price?

The company disclosed a Quebec mining facility of 26,700 square feet on 2019-06-02, cited as claim #15230, with a lease running through November 2022 tied to a figure of $120,150, cited as claims #15219 and #15229, and separately disclosed the deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220. What has not been said is whether $120,150 is monthly, annual, or the aggregate lease obligation; what the contracted power capacity in megawatts is; the name of the utility or interconnection counterparty and the all-in energy rate per kilowatt-hour; and the model, aggregate hashrate, and energy draw of those 1,400 units — so how many megawatts and what petahash does 26,700 square feet in Granby actually support?

The company disclosed bitcoin mining operations at the Granby facility together with 24-hour security services and financial operations management on 2019-07-31, cited as claims #15198, #15199 and #15200, while disclosing a total of three full-time employees on 2019-07-01, cited as claim #15222. What has not been said is the identity of the counterparty performing hosting, security, and operational management, the fee structure of that arrangement, and its term and termination rights — who actually runs the site, and what does Marathon pay them per month or per megawatt?

The company disclosed an At The Market program with H.C. Wainwright on 2019-07-18 with a capacity of $7,472,417, cited as claims #15202, #15203 and #15217, a 3% sales agent commission, cited as claim #15204, and a $25,000 legal reimbursement, cited as claim #15205, the same capacity figure having first appeared in the 2019-06-02 shelf disclosure, cited as claim #15228. What has not been said is whether that oddly precise ceiling reflects the one-third public-float cap for smaller issuers, how much had actually been sold as of each reporting date, at what weighted-average price, and how many shares that represents against shares outstanding — what is the realized dilution to date and how much headroom remains if the float shrinks?

The company disclosed on 2018-10-15 a $350,000 base salary for Merrick Okamoto, cited as claim #15388, together with option grants of 5,000,000 shares, cited as claim #15389, 2,500,000 shares, cited as claim #15385, and 2,500,000 unvested options carrying change-of-control acceleration, cited as claim #15386, alongside grants to David Lieberman, James Crawford, and the three new directors, cited as claims #15392, #15393, #15394, #15395 and #15396 — all against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claims #15460 and #15373. What has not been said is whether these option figures are distinct grants or overlapping descriptions of the same award, their exercise prices, vesting schedules, and grant-date fair values, and how much of the 10,000,000 share pool remained available after them — what percentage of fully diluted shares did insiders hold in options at year-end 2018?

The company disclosed the termination of the Global Bit Ventures merger on 2018-06-28, cited as claim #15477, the resignation of directors on the same date, cited as claim #15476, and later characterized the termination as carrying a 3,000,000 share break-up fee on 2018-08-13, cited as claim #15461. What has not been said is who received those 3,000,000 shares, the value recorded for them, why the deal was abandoned after diligence, and whether the director resignations on that same day were connected to the termination — did any departing director or affiliate receive any portion of the break-up consideration?

The company disclosed a going-concern style 12-month cash runway uncertainty on 2018-08-13, cited as claim #15462, and a cascade of Nasdaq deficiencies including the $1.00 minimum bid price deadline of 2019-04-15, cited as claim #15383, a reverse stock split authorization due by 2018-12-31 that is recorded as delayed, cited as claim #15332, and a stockholders' equity deficiency with plan deadlines in 2019, cited as claims #15243, #15244 and #15196. What has not been said is the cash balance and monthly burn rate underlying the runway warning, the dollar amount by which stockholders' equity fell short of the Nasdaq minimum, and whether the authorized reverse split was ever effected and at what ratio — what is the quarterly cash burn, and what equity shortfall must the ATM close to cure the listing deficiency?

The company disclosed a 26,700 square foot mining facility in Quebec on 2019-06-02, cited as claim #15230, alongside a lease running through November 2022 carrying a figure of $120,150 on 2019-07-01, cited as claim #15219, and bitcoin mining at a "Granby facility" on 2019-07-31, cited as claim #15198; what has not been said is the megawatt capacity of that site, the contracted power rate per kWh, the street address and utility interconnection point, and whether $120,150 is monthly, annual, or the aggregate rent over the full 41-month term. Square footage is not a capacity unit for a mining operation — what is the energized load, and what is the escalator on the power contract?

The company disclosed deployment of 1,400 units of bitcoin mining equipment on 2019-07-01, cited as claim #15220, while simultaneously disclosing a full-time headcount of three employees on the same date, cited as claim #15222; what has not been said is the make and model of those machines, the aggregate hashrate in PH/s or EH/s, how many of the 1,400 were actually energized and hashing as of the filing date versus purchased or in transit, and the identity of the third party that must be operating and maintaining the site, because three full-time employees cannot. Is there a hosting or O&M counterparty at Granby, what does it charge, and why has its name been withheld?

The company disclosed a shelf registration figure of $7,472,417 on 2019-06-02, cited as claim #15228, repeated the identical figure on 2019-07-01, cited as claim #15217, and again as At-The-Market program capacity on 2019-07-18 with H.C. Wainwright as agent at a 3.0 percent commission, cited as claims #15203, #15202 and #15204; what has not been said is whether that number is original shelf capacity or remaining availability net of prior sales, how many dollars and shares were actually sold under the ATM in each subsequent quarter, and the weighted-average issuance price. The same figure appearing unchanged across three dates raises the question of whether it was ever drawn at all, or whether the disclosure simply was not updated.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO at a base salary of $350,000 with 5,000,000 stock options, cited as claims #15387, #15388 and #15389, and separately disclosed 2,500,000 options issued to the CEO and 2,500,000 unvested options carrying change-of-control acceleration, cited as claims #15385 and #15386; what has not been said is whether the 2.5 million grants are a subset of the 5 million or incremental to it, the exercise price and grant-date fair value of each tranche, the vesting schedule, and the strike price relative to the then-prevailing market price. Given that the 2018 Equity Incentive Plan reserve of 10,000,000 shares is disclosed twice, on 2018-08-13 and 2018-11-12, cited as claims #15460 and #15373, is that one pool or two, and what percentage of shares outstanding did the CEO's grants alone represent on a fully diluted basis?

The company disclosed acquisition of a cryptocurrency patent from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, and a 30 percent waterfall interest in a patent special purpose entity on the same date, cited as claim #15223; what has not been said is whether any officer, director, or significant shareholder of the company held a concurrent interest in XpresSpa Group or in the SPE, who holds the other 70 percent of the waterfall, what the SPE's litigation or licensing strategy is, and how the $250,000 price was determined or whether an independent valuation or audit committee approval was obtained. A transaction between two small-cap companies with a shared promoter orbit is precisely where related-party context is expected and here absent.

The company disclosed termination of the Global Bit Ventures merger on 2018-06-28, cited as claim #15477, with a share-based break-up fee of 3,000,000 shares disclosed on 2018-08-13, cited as claim #15461, and disclosed on the same June date that directors had resigned, cited as claim #15476; what has not been said is who received those 3,000,000 shares, whether any recipient was a departing director or an affiliate of one, the dollar value recorded for the fee, and whether the resignations and the termination were causally linked. Which directors resigned, and was the break-up consideration ever actually issued?

The company disclosed a Nasdaq minimum bid price compliance deadline of $1.00 by 2019-04-15 on 2018-10-21, cited as claim #15383, and a reverse stock split authorization due by 2018-12-31 on 2018-12-11, cited as claim #15332, both of which remain in delayed status; what has not been said is whether the reverse split was ever executed, at what ratio, on what effective date, and how compliance with the bid price rule was ultimately resolved. The record shows four separate Nasdaq deficiency and compliance-plan deadlines across 2018 and 2019, cited as claims #15466, #15467, #15196 and #15244, all still carrying delayed status — what is the current listing status, and what remains outstanding?

The company disclosed a Quebec mining facility of 26,700 square feet on 2019-06-02, cited as claim #15230, alongside a lease running to November 2022 at $120,150 on 2019-07-01, cited as claim #15219, and a deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220; what has not been said is any electrical figure at all — no contracted megawatts, no energy rate in cents per kilowatt-hour, no utility or interconnection agreement, no street address in Granby, and no miner model or aggregate hashrate, leaving square footage and unit count standing in for capacity. Is the $120,150 an annual rent, a monthly rent, or the total obligation across the 41-month term, and what is the all-in power cost per bitcoin mined at the deployed fleet's efficiency?

The company disclosed three separate Granby-related arrangements on 2019-07-31 — mining operations at the facility, cited as claim #15198, 24-hour security services, cited as claim #15199, and financial operations management, cited as claim #15200 — without naming a single counterparty to any of them. What has not been said is who the hosting, security, and back-office providers are, what each is paid, what the termination provisions are, and whether any of the three is affiliated with a director, an officer, or a holder of record, which matters acutely at a company that reported three full-time employees on 2019-07-01, cited as claim #15222.

The company disclosed a cryptocurrency patent acquisition from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30% waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is who holds the remaining 70% of that waterfall, which patents were conveyed by number, what board or officer overlap existed between the two public companies at the time of the transaction, and whether the related-party review was conducted by disinterested directors. Has the SPE generated any licensing revenue, and what is the waterfall's priority stack ahead of the 30% interest?

The company disclosed on 2018-10-15 a $350,000 base salary for Merrick Okamoto, cited as claim #15388, and option grants of 5,000,000 shares, cited as claim #15389, plus 2,500,000 options, cited as claim #15385, of which 2,500,000 unvested options accelerate on a change of control, cited as claim #15386; what has not been said is the exercise price, vesting schedule, grant-date fair value, or the resulting percentage of fully diluted shares. Compounding this, Fred Thiel was named Compensation Committee Chairman on 2018-08-21, cited as claim #15437, and received a 50,000-share option grant on 2018-10-15, cited as claim #15396 — who set the chair's own grant, and was the CEO award benchmarked against any peer group?

The company disclosed a 2018 Equity Incentive Plan reserve of 10,000,000 shares on 2018-08-13 and again on 2018-11-13, cited as claims #15460 and #15373, authorization to issue non-public securities equal to 20% on 2018-12-11, cited as claim #15333, a reverse stock split authorization due by 2018-12-31 that remains delayed, cited as claim #15332, and a $1.00 minimum bid price cure deadline of 2019-04-15, also delayed, cited as claim #15383; what has not been said is the shares-outstanding denominator against which any of these dilution figures should be read, the split ratio actually selected, and the effective date. Why did the authorized split lapse past its own deadline while the bid-price deficiency was still outstanding?

The company disclosed an At The Market program of $7,472,417 on 2019-07-18, cited as claim #15203, matching to the dollar the shelf figure cited on 2019-06-02 and 2019-07-01, claims #15228 and #15217, with a 3% sales agent commission, cited as claim #15204, against a stated twelve-month cash runway uncertainty from 2018-08-13, cited as claim #15462; what has not been said is how much of that capacity has actually been sold, at what weighted average price, what cash balance and monthly burn rate the going-concern language rests on, and how the three-share-reserve, ATM, and 20% authorizations interact. Separately, the 3,000,000-share break-up fee tied to the terminated Global Bit Ventures merger on 2018-08-13, cited as claim #15461, and the director resignations of 2018-06-28, cited as claim #15476, are disclosed without naming who received those shares or whether any recipient overlapped with the departing board.

The company disclosed on 2019-07-01 that it had deployed 1,400 bitcoin mining units, cited as claim #15220, and separately described a 26,700 square foot Quebec facility on 2019-06-02, cited as claim #15230, with a lease running through November 2022 at a stated figure of $120,150, cited as claims #15229 and #15219; what has not been said is the facility's megawatt interconnection capacity, its all-in power price per kWh, its street address, or whether the $120,150 is annual rent, monthly rent, a deposit, or the total contracted lease obligation. A reader cannot convert 1,400 machines into hashrate or into a cost-per-bitcoin figure without the miner model, and none is given. Which utility or substation serves the Granby site, and under what rate schedule?

The 2019-07-31 disclosures name Granby as the site of mining operations and add a 24-hour security arrangement and an outsourced financial operations management function, cited as claims #15198, #15199 and #15200; what has not been said is who the counterparties are. With a full-time headcount of three employees as of 2019-07-01, cited as claim #15222, essentially the entire operating stack appears to be vendor-supplied, so who hosts the machines, who operates them, who provides security, who keeps the books, what are the contract terms and termination rights, and are any of these parties affiliated with officers or directors?

The company disclosed a cryptocurrency patent acquisition from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, alongside a 30% waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is whether this was a related-party transaction, given the overlapping investor and board networks between the two issuers, who holds the other 70% of the waterfall, what the SPE is named, who controls enforcement decisions, and how the $250,000 price was determined or by whom it was reviewed.

The compensation disclosures of 2018-10-15 record a $350,000 base salary and 5,000,000 stock options for Merrick Okamoto, cited as claims #15388 and #15389, plus separate line items of 2,500,000 options to the CEO and 2,500,000 unvested options carrying change-of-control acceleration, cited as claims #15385 and #15386, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claims #15460 and #15373. What has not been said is whether the 5,000,000 figure is inclusive of or incremental to the two 2,500,000 tranches, what the exercise prices and vesting schedules are, and what percentage of fully diluted shares the CEO's grant represents when the top executive's award alone may consume half the entire plan reserve.

The company disclosed an At The Market Offering program with H.C. Wainwright on 2019-07-18 with capacity of $7,472,417 and a 3% commission, cited as claims #15202, #15203 and #15204, matching the shelf figures cited as claims #15228 and #15217; what has not been said is why the capacity is that precise and unusual number, which strongly suggests a baby-shelf limitation tied to one-third of non-affiliate float, nor how much has actually been sold to date, at what average price, and how many shares were issued. On the next call, what is the remaining unused ATM capacity, and does the float calculation reset in a way that shrinks it further if the stock declines?

The company flagged 12-month cash runway uncertainty on 2018-08-13, cited as claim #15462, and then disclosed a Nasdaq minimum stockholders' equity compliance filing due 2019-11-13, cited as claim #15196, following a violation noted 2019-05-23, cited as claims #15243 and #15244; what has not been said is the actual cash balance, the monthly operating burn, the size of the stockholders' equity shortfall in dollars, and the specific plan to close it. How much bitcoin was mined and how much was sold rather than held in each quarter, and at what realized price?

The company authorized a reverse stock split by 2018-12-31, cited as claim #15332, and faced a $1.00 minimum bid price deadline of 2019-04-15, cited as claim #15383, both carried on the record as delayed; what has not been said is whether the split was ever effected, at what ratio, and on what date, or how compliance with the bid price rule was otherwise resolved before the separate 2019 deficiencies arose.

The Global Bit Ventures merger was terminated as disclosed on 2018-06-28, cited as claim #15477, with a share-based break-up fee of 3,000,000 shares, cited as claim #15461, on the same date that directors resigned, cited as claim #15476; what has not been said is which directors left, whether any resignation involved a disagreement with the company requiring disclosure, the dollar value ascribed to the 3,000,000 shares, and whether any GBV principal retained an ongoing equity or contractual relationship with the company afterward.

The company disclosed a Quebec data center of 26,700 square feet on 2019-06-02, cited as claim #15230, with a lease running through November 2022 at a stated $120,150 on 2019-07-01, cited as claim #15219; what has not been said is the site's street address, its contracted megawatts and interconnection arrangement with the local utility, the all-in power price per kWh, and whether $120,150 is an annual rent, a monthly figure, or the aggregate over the lease term — square footage is a real estate unit, not a capacity unit, and nothing in the record converts one to the other.

The company disclosed deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220, alongside a stated plan to mine ether, cited as claim #15231; what has not been said is the miner model and per-unit hashrate, the aggregate hash power in PH/s, the energized-versus-received split as of the filing date, the capital committed to the ether plan, or the machines' expected useful life. A first-year analyst would ask on the next call how many of the 1,400 units were actually hashing, what the fleet efficiency in J/TH is, and how many bitcoin were produced in the quarter.

The company disclosed a three-million-share break-up fee tied to the terminated Global Bit Ventures merger on 2018-08-13, cited as claim #15461, following the termination announced on 2018-06-28, cited as claim #15477, on the same day directors resigned, cited as claim #15476; what has not been said is the dollar value ascribed to those shares, whether the departing directors had any affiliation with or economic interest in Global Bit Ventures, and why a terminated transaction carried a share-denominated fee payable by the company rather than a cash fee or none at all.

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options, cited as claim #15389, while separately disclosing 2,500,000 options issued to the CEO, cited as claim #15385, and 2,500,000 unvested options carrying change-of-control acceleration, cited as claim #15386, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claims #15460 and #15373; what has not been said is whether these are three grants, one grant described three ways, or overlapping tranches, nor the exercise price, vesting schedule, grant-date fair value, or the resulting share of the entire plan reserve consumed by a single executive alongside his $350,000 salary, cited as claim #15388.

The company disclosed the purchase of cryptocurrency patents from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30% waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is whether any officer or director of Marathon held a board seat, equity position, or other interest in XpresSpa at the time, the identity and manager of the special purpose entity, the tiers and priority of the waterfall above and below the 30% strip, and whether the transaction was reviewed as a related-party matter by an independent committee.

The company disclosed an At The Market program with H.C. Wainwright of $7,472,417 on 2019-07-18, cited as claims #15202 and #15203, at a 3% commission, cited as claim #15204, plus $25,000 of counsel expense reimbursement, cited as claim #15205, and the identical $7,472,417 figure appears in shelf disclosures on 2019-06-02 and 2019-07-01, cited as claims #15228 and #15217; what has not been said is the public float that produced that unusually precise cap, how much had actually been sold and at what average price, and how the remaining capacity reconciles with the going-concern language on twelve-month cash sufficiency, cited as claims #15462 and #15197, at a company reporting three full-time employees, cited as claim #15222.

The company disclosed shareholder authorization of a reverse stock split on 2018-12-11, cited as claim #15332, against a Nasdaq $1.00 minimum bid price cure deadline of 2019-04-15, cited as claim #15383; what has not been said is the ratio selected, the effective date, or whether the split was ever implemented — and the record likewise names no counterparty, fee, or term for the 24-hour security services and financial operations management arrangements disclosed on 2019-07-31, cited as claims #15199 and #15200, leaving open whether either vendor is a related party.

The company announced bitcoin mining operations at the Granby facility on 2019-07-31, cited as claims #15198, #15199 and #15200, describing mining operations, 24-hour security services and financial operations management as if performed by someone else; what has not been said is who that counterparty is, whether the hosting, security and financial-management roles sit with one vendor or three, what the contract term and fee structure are, and whether any of those parties is a related party of an officer or director. The gap is sharpened by claim #15222, which puts total full-time headcount at three employees as of 2019-07-01 — who physically operates a 26,700 square foot site, and under what agreement?

Marathon disclosed a Quebec mining facility of 26,700 square feet on 2019-06-02, cited as claim #15230, and a data center lease running through November 2022 with a figure of $120,150 on 2019-07-01, cited as claim #15219; what has not been said is the period that $120,150 covers — monthly, annual, or aggregate over the full lease — nor the site's street address, its electrical service in megawatts, the contracted power rate per kilowatt-hour, the Hydro-Québec interconnection or energization status, and whether the lease price includes power at all. Square footage is not capacity: what is the energized megawatt figure behind the 26,700 square feet?

The company disclosed deployment of 1,400 bitcoin mining units on 2019-07-01, cited as claim #15220, alongside plans to mine ether on 2019-06-02, cited as claim #15231; what has not been said is the miner make and model, the aggregate hashrate in petahash, how many of the 1,400 units were actually energized and hashing versus purchased or in transit, the capitalized cost and depreciation life, and — given the separate ether plan — whether any GPU capacity was ever procured or the plan quietly abandoned. A first-year analyst would ask for monthly bitcoin production and all-in cost per coin against these 1,400 units.

Marathon disclosed acquiring cryptocurrency patents from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30% waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is the name of the special purpose entity, who holds the other 70%, the tiers and thresholds of the waterfall, who controls enforcement and litigation funding decisions, and — most importantly — what overlapping officer, director or shareholder relationships existed between Marathon and XpresSpa at the time that would make this a related-party transaction requiring proxy disclosure.

On 2018-10-15 the company disclosed option grants of 5,000,000 shares to Merrick Okamoto, cited as claim #15389, plus a 2,500,000 share CEO grant and 2,500,000 unvested options carrying change-of-control acceleration, cited as claims #15385 and #15386, against a 2018 Equity Incentive Plan reserve of 10,000,000 shares, cited as claims #15460 and #15373; what has not been said is whether these are one grant described three ways or successive grants that consume the entire plan reserve, and in either case the exercise price, vesting schedule, and grant-date fair value are absent. Given the $1.00 minimum bid price deficiency disclosed six days later on 2018-10-21, cited as claim #15383, at what strike were these options struck relative to a sub-$1 stock, and what change-of-control event would accelerate the 2,500,000 unvested shares?

The company authorized a reverse stock split on 2018-12-11 with a 2018-12-31 date attached, cited as claim #15332, and separately obtained approval to issue non-public securities above the 20% threshold, cited as claim #15333, after regaining Nasdaq compliance on 2018-08-21, cited as claim #15438 — yet a new listing rule violation followed on 2019-05-23, cited as claim #15243; what has not been said is whether the reverse split was ever effected and at what ratio, why authority granted in December did not cure the bid price deficiency whose deadline was 2019-04-15, and which specific rule the May 2019 violation concerns. Was the 20% issuance authority used, and if so, to whom and at what discount?

The At The Market program with H.C. Wainwright was disclosed on 2019-07-18 at a capacity of $7,472,417 with a 3% sales agent commission, cited as claims #15202, #15203 and #15204, against going-concern language about twelve-month cash sufficiency on 2018-08-13 and 2019-07-31, cited as claims #15462 and #15197; what has not been said is the cash balance and monthly operating burn those going-concern statements are measured against, how much of the ATM had actually been sold and at what average price, and why the capacity is an unrounded $7,472,417 — is it the one-third public float cap under General Instruction I.B.6, and what public float does that imply? How many months of runway do current cash plus undrawn ATM capacity fund at the present burn rate?

The Global Bit Ventures merger was disclosed as terminated on 2018-06-28, cited as claim #15477, with directors resigning the same day, cited as claim #15476, but a 3,000,000 share break-up fee surfaced only in the 2018-08-13 disclosure, cited as claim #15461; what has not been said is who received those shares, the dollar value recorded and the expense period in which it was booked, whether the resigning directors were GBV designees or affiliates receiving any portion of the fee, and why a termination announced in June had its share-based cost disclosed six weeks later.

The company disclosed a Quebec data center lease of 26,700 square feet running through November 2022 at a stated $120,150 on 2019-06-02 and 2019-07-01, cited as claims #15230, #15229 and #15219, and separately described bitcoin mining operations at a 'Granby facility' on 2019-07-31, cited as claim #15198; what has not been said is the street address and interconnection detail of that site, the contracted megawatt capacity behind the square footage, whether the $120,150 figure is annual rent, total lease consideration or a deposit, and whether 'Granby' and the 26,700-square-foot lease are the same building or two sites. Square footage is not a power number, and a miner cannot be sized from it.

The company disclosed deployment of 1,400 units of bitcoin mining equipment on 2019-07-01, cited as claim #15220, alongside a full-time employee headcount of three on the same date, cited as claim #15222; what has not been said is the miner model and aggregate hashrate, the energized-versus-received split of those 1,400 units, the total load they draw against the leased facility, and who physically operates and maintains the fleet given a three-person payroll — the 24-hour security services and financial operations management arrangements announced on 2019-07-31, cited as claims #15199 and #15200, name no counterparty and disclose no fee.

The company disclosed hosting at a bitcoin mining facility with electricity costs of $0.055 per KwH on 2019-08-28, cited as claim #15153; what has not been said is the identity of the host, the contract term and termination rights, whether $0.055 is an all-in hosting rate or a pass-through power rate before a hosting margin, whether there is a minimum take-or-pay commitment, and how many megawatts are contracted at that price. A rate without a committed quantity or a term is not a cost structure.

The company disclosed a cryptocurrency patent acquisition from XpresSpa Group for $250,000 on 2019-07-01, cited as claim #15218, together with a 30% waterfall interest in a patent special purpose entity, cited as claim #15223; what has not been said is whether XpresSpa was a related party at the time through overlapping officers, directors or beneficial owners, whether the $250,000 price was set by arm's-length negotiation or board fairness determination, the name and capitalization of the special purpose entity, and where the 30% interest sits in the waterfall — that is, what recovery thresholds and priority returns must be satisfied before the company receives a dollar.

The company disclosed a 2018 Equity Incentive Plan share reserve of 10,000,000 shares on 2018-08-13 and again on 2018-11-12, cited as claims #15460 and #15373, and then on 2018-10-15 disclosed grants of 5,000,000 options and 2,500,000 options to Merrick Okamoto, 200,000 to David Lieberman, 100,000 to James Crawford and 50,000 each to three directors, cited as claims #15389, #15385, #15392, #15393, #15394, #15395 and #15396, plus 2,500,000 unvested options carrying change-of-control acceleration, cited as claim #15386; what has not been said is whether the 5,000,000 and 2,500,000 figures are two separate grants or one grant described twice, the exercise prices and vesting schedules, how much of the 10,000,000-share reserve remained after a single day's grants, and whether the accelerating 2,500,000 tranche is a subset of those awards or additional to them. Against base salaries of $350,000 and $180,000, cited as claims #15388 and #15391, the option grants are the compensation, and their terms are the undisclosed part.

The company disclosed a shelf registration and At The Market program capacity of $7,472,417 on 2019-06-02, 2019-07-01 and 2019-07-18, cited as claims #15228, #15217 and #15203, with a 3% sales agent commission to H.C. Wainwright, cited as claims #15204 and #15202, and disclosed 8,720,405 shares of common stock outstanding post-transaction on 2019-08-28, cited as claim #15152, following a reverse stock split authorization that slipped past its 2018-12-31 date, cited as claim #15332; what has not been said is the reverse split ratio and effective date, how many dollars of the ATM had actually been sold and at what average price as of the 8,720,405 share count, and whether the oddly precise $7,472,417 is a baby-shelf General Instruction I.B.6 one-third-of-float cap that shrinks as the stock falls. The first-year analyst question for the next call is simple: how many ATM dollars remain available today, and at what share price does that capacity go to zero.

The company disclosed twelve-month cash sufficiency uncertainty on 2018-08-13 and an operating cash sufficiency statement on 2019-07-31, cited as claims #15462 and #15197, both now carried as delayed, while disclosing a Nasdaq minimum stockholders' equity deficiency on 2019-05-23 and a compliance filing obligation, cited as claims #15243, #15244 and #15196 — this after having regained compliance on 2018-08-21, cited as claim #15438; what has not been said is the actual cash balance and monthly operating burn underlying the going-concern language, the dollar size of the stockholders' equity shortfall against the Nasdaq threshold, and how a $7.5 million ATM is expected to close that gap. Separately, the Global Bit Ventures merger termination carrying a 3,000,000-share break-up fee, cited as claims #15461 and #15477, has never been accompanied by the reason for termination, the value ascribed to those 3,000,000 shares, or whether they were in fact issued.

The company disclosed on 2019-07-31 that bitcoin mining operations at its Granby facility, 24-hour security, and financial operations management were all handled through partnerships (claims #15198, #15199, #15200), while reporting only three full-time employees on 2019-07-01 (claim #15222). What has not been said is who these counterparties are, what they are paid, and whether any of them are related to officers or directors. For a company this small, those vendors effectively run the business.

The company disclosed on 2019-07-01 a Quebec data center lease running through November 2022 at $120,150 (claim #15219), a 26,700 square foot facility (claim #15230), and deployment of 1,400 mining units (claim #15220). What has not been said is whether the $120,150 is a monthly, annual, or total-term figure. The company also has not given the facility's street address, contracted power capacity in megawatts, utility interconnection, miner model, or aggregate hashrate. Without those, the capacity claim cannot be tied to revenue potential.

The company disclosed on 2019-08-28 a hosting arrangement at $0.055/kWh (claim #15153). What has not been said is the name of the hosting provider, the contracted power capacity, the contract term, or whether the rate is fixed or pass-through. It is also unclear whether this arrangement is the Granby facility or a separate site.

The company disclosed on 2018-10-15 that Merrick Okamoto received 2.5 million stock options (claim #15385) and, in a separate item, 5,000,000 options (claim #15389). It also disclosed change-of-control acceleration on 2.5 million unvested options (claim #15386), against a 10,000,000-share 2018 Equity Incentive Plan reserve (claims #15460, #15373). What has not been said is which option figure is correct and what the strike price and vesting terms are. The company also has not said why roughly half the plan reserve went to one executive. Nor has it explained whether Proposal No. 5, withdrawn on 2018-11-13 over NASDAQ Rule 5635(b) change-of-control concerns (claim #15372), was connected to Okamoto's holdings.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated and that directors resigned the same day (claims #15477, #15476). On 2018-08-13 it described a 3,000,000-share break-up fee (claim #15461). What has not been said is which party paid the fee, and whether it was issued or cancelled. The company also has not named the resigning directors or said whether their departures related to the failed merger.

The company disclosed on 2019-07-01 a $250,000 patent acquisition from XpresSpa Group paired with a 30% waterfall interest in a patent special purpose entity (claims #15218, #15223). It separately disclosed a $7,472,417 H.C. Wainwright at-the-market program at a 3% commission (claims #15203, #15204). It also flagged twelve-month cash runway doubts twice (claims #15462, #15197). What has not been said is who holds the other 70% of the patent entity, what the waterfall priority is, and how much of the at-the-market capacity has actually been drawn.

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated (claim #15477), that directors resigned the same day (claim #15476), and later a break-up fee payable in 3,000,000 shares (claim #15461). What has not been said is which directors left, why the deal collapsed, whether the fee shares were actually issued, and at what per-share value they were booked. Was the resignation connected to the termination?

The company disclosed on 2018-10-15 option grants to Merrick Okamoto of 5,000,000 shares (claim #15389), described elsewhere as 2,500,000 issued plus 2,500,000 unvested with change-of-control acceleration (claims #15385, #15386), against a 10,000,000-share 2018 Equity Incentive Plan reserve (claims #15460, #15373). What has not been said is the strike price, the vesting triggers for the unvested tranche, and why half the plan went to one executive. It is also unexplained what change of control Proposal No. 5 would have caused under Nasdaq Rule 5635(b) before it was withdrawn on 2018-11-13 (claim #15372), and who would have gained control.

The company disclosed on 2019-06-02 and 2019-07-01 a 26,700-square-foot Quebec facility (claim #15230) leased through November 2022 at a cited 120,150 USD (claim #15219), later tied to Granby (claim #15198). What has not been said is whether 120,150 USD is a monthly, annual, or total-term figure, the site's street address, its contracted power capacity in megawatts, and the name of the landlord or operator. Who provides the 24-hour security and financial operations management described as partnerships (claims #15199, #15200)?

The company disclosed 1,400 deployed mining units on 2019-07-01 (claim #15220), then on 2019-08-28 an acquisition of 6,000 Bitmain S-9 Antminers (claim #15150) hosted at 0.055 USD/kWh (claim #15153). What has not been said is the seller, the purchase price and how it was financed, the hosting counterparty and site, the resulting hashrate, and the fate of the 1,400 existing units. Was the 0.055 USD/kWh rate all-in or electricity only, and for what term?

The company disclosed on 2019-07-01 a 250,000 USD patent purchase from XpresSpa Group (claim #15218) and a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the remaining interest in the entity, whether any party is related to management or directors, and what the waterfall priority actually pays out before the company's 30% applies.

The company disclosed a string of Nasdaq compliance deadlines that slipped (claims #15466, #15467, #15468, #15383, #15244, #15196), a reverse split authorization (claim #15332), going-concern-style runway language (claims #15462, #15197), and a 7,472,417 USD H.C. Wainwright ATM at a 3% commission (claims #15202, #15203, #15204). What has not been said is the reverse split ratio and whether it was executed, how much of the ATM was drawn and at what average price, and whether the ATM proceeds alone were expected to cover the twelve-month runway with only three full-time employees on staff (claim #15222).

The company disclosed on 2019-06-02 a 26,700 square foot Quebec mining facility (claim #15230) and on 2019-07-01 a lease there through November 2022 at 120,150 USD (claim #15219). Operations at a Granby facility followed on 2019-07-31 (claim #15198). What has not been said is whether the 120,150 USD is annual, monthly or total, and whether it is in USD or CAD. The company also has not said whether Granby is the same site, what megawatt load is contracted, or which utility and interconnection serve it. So what power capacity actually backs the Quebec footprint?

The company disclosed on 2019-08-28 a hosting arrangement at 0.055 USD/kWh (claim #15153) and on 2019-07-01 a deployment of 1,400 mining units (claim #15220). What has not been said is who the hosting counterparty is, how long the rate is fixed and how much power it covers. The company also has not given the miner model or aggregate hashrate behind the 1,400 units. The Granby security and financial-operations partners (claims #15199, #15200) are likewise unnamed. Who are these counterparties, and on what terms?

The company disclosed on 2019-07-01 a 250,000 USD patent acquisition from XpresSpa Group (claim #15218) and a 30% waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the other 70%, what cash flows the waterfall applies to, and whether any director or officer has a relationship with XpresSpa or the SPE. Should this have been presented as a related-party transaction?

The company disclosed on 2018-10-15 that CEO Merrick Okamoto received 5,000,000 stock options (claim #15389). It described these as 2,500,000 issued plus 2,500,000 unvested options with change-of-control acceleration (claims #15385, #15386), alongside a 350,000 USD base salary (claim #15388). What has not been said is the strike price, the vesting schedule or performance hurdles, and what counts as a change of control. The company also has not disclosed any bonus terms. What is the full economic value of the package, measured against the 10,000,000 shares reserved under the 2018 plan (claims #15460, #15373)?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a 3,000,000-share break-up fee (claim #15461). On 2018-06-28 it disclosed that directors resigned in connection with the termination (claims #15477, #15476). What has not been said is which party paid the fee to whom, whether the shares were ever issued and at what valuation, and which directors left and why. How did the break-up fee affect the share count?

The company disclosed a chain of Nasdaq compliance deadlines, all now marked delayed (claims #15467, #15468, #15383, #15244, #15196). It also disclosed a reverse split authorization due by 2018-12-31 (claim #15332), approval for non-public issuances above 20% (claim #15333) and a 2,335,000-share private placement on 2019-08-28 (claim #15151). What has not been said is the split ratio and whether a split was executed. The company also has not named the placement investors or disclosed the per-share price and discount. What dilution did these financings cause relative to the H.C. Wainwright ATM capacity of 7,472,417 USD (claim #15203)?

The company disclosed on 2018-10-15 that Merrick Okamoto received stock options on 5,000,000 shares (claim #15389), split into 2,500,000 options (claim #15385) and 2,500,000 unvested options with change-of-control acceleration (claim #15386). What has not been said is the exercise price, the vesting schedule or performance conditions for the unvested tranche, or what counts as a change of control. Given that Proposal No. 5 was withdrawn on 2018-11-13 over NASDAQ Rule 5635(b) change-of-control concerns (claim #15372), what did that proposal cover, and who would have gained control or economics under it?

The company disclosed on 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee of 3,000,000 shares (claim #15461), and on 2018-06-28 that directors resigned in connection with the termination (claims #15476, #15477). What has not been said is who received the break-up shares, how they were valued, whether they carried registration rights, and which directors resigned and why. Were any of the resigning directors affiliated with Global Bit Ventures?

The company disclosed a Quebec mining facility of 26,700 square feet on 2019-06-02 (claim #15230), a lease running through November 2022 tied to a figure of $120,150 on 2019-07-01 (claim #15219), and mining operations at a Granby facility on 2019-07-31 (claim #15198). What has not been said is whether $120,150 is monthly, annual, or total lease cost, whether Granby is the same site as the Quebec facility, and what the site's contracted power capacity in megawatts is, along with the utility counterparty and interconnection terms. How many megawatts does the facility actually draw?

The company disclosed 1,400 bitcoin mining units deployed on 2019-07-01 (claim #15220), alongside a full-time headcount of 3 employees (claim #15222), and on 2019-08-28 a hosting arrangement at $0.055/kWh (claim #15153). What has not been said is the miner model, aggregate hashrate, the share of units actually operating, the name of the hosting provider, and the contract term and minimum-power commitments. The same gap applies to the unnamed providers of 24-hour security and financial operations management (claims #15199, #15200). Who are these counterparties, and are any of them related parties?

The company disclosed on 2019-07-01 a $250,000 cryptocurrency patent acquisition from XpresSpa Group (claim #15218) and a 30% waterfall interest in a patent special-purpose entity (claim #15223). What has not been said is which patents were acquired, the name and other owners of the SPE, and where Marathon sits in the waterfall. What monetization, if any, has the patent portfolio produced relative to its cost?

The company disclosed an all-stock private placement of 2,335,000 shares on 2019-08-28 (claim #15151), a $7,472,417 at-the-market program with H.C. Wainwright at a 3% commission (claims #15202, #15203, #15204), and repeated doubts about its twelve-month cash runway (claims #15462, #15197). What has not been said is who the placement purchasers were, the per-share price, how much of the ATM was actually drawn and at what average price, and whether the repeated Nasdaq equity and bid-price deficiencies (claims #15243, #15196, #15383) were cured by these raises or by the reverse split authorized on 2018-12-11 (claim #15332). When, if ever, was that reverse split effected?

The company disclosed on 2019-07-31 that Bitcoin mining at its Granby facility runs under a partnership that also covers 24-hour security and financial operations management (claims #15198, #15199, #15200), and on 2019-08-28 that hosting is priced at 0.055 USD/kWh (claim #15153). What has not been said is who the counterparty is. Is it the same party for hosting, security and financial management, and is that party related to any director, officer or placement investor?

The company disclosed a 26,700 square foot Quebec facility on 2019-06-02 (claim #15230) and a lease through November 2022 figured at 120,150 USD (claims #15219, #15229). What has not been said is whether that figure is monthly, annual or total. The company has also not given the facility's street address, its contracted megawatt capacity, or its utility and interconnection arrangement, and it has not said whether the 0.055 USD/kWh rate is fixed or indexed.

The company disclosed on 2019-07-01 that 1,400 mining units were deployed (claim #15220) with only three full-time employees (claim #15222). What has not been said is the miner model, the aggregate hashrate, the power draw, or the monthly bitcoin produced. Without those figures, the hosting cost and the cash-runway warnings (claims #15462, #15197) cannot be checked against unit economics.

The company disclosed on 2018-10-15 that CEO Merrick Okamoto received 2.5 million stock options (claim #15385) but also 5 million options (claim #15389), with 2.5 million unvested options carrying change-of-control acceleration (claim #15386). What has not been said is how these figures reconcile, or what the exercise price, vesting schedule and grant-date fair value are. The company has also not explained how these grants relate to the 10 million share 2018 Equity Incentive Plan reserve (claims #15460, #15373).

The company disclosed on 2018-06-28 and 2018-08-13 that the Global Bit Ventures merger was terminated with a break-up fee paid as 3,000,000 shares (claims #15477, #15461), and that directors resigned on the same date (claim #15476). What has not been said is who received the shares, what value was assigned to them, and whether they were actually issued. The company has also not said which directors resigned or whether the resignations were tied to the termination.

The company disclosed on 2019-08-28 an all-stock private placement of 2,335,000 common shares (claim #15151), and on 2019-07-18 a 7,472,417 USD at-the-market program with H.C. Wainwright (claims #15202, #15203). What has not been said is who bought the placement shares, at what price, and whether any purchaser is linked to the Granby hosting counterparty. The company has also not said how much of the ATM capacity has been drawn, or whether the reverse split authorized on 2018-12-11 (claim #15332) was carried out, and at what ratio, to address the recurring Nasdaq deficiencies (claims #15383, #15243, #15196).

The company disclosed on 2018-10-15 that Merrick Okamoto received 5,000,000 stock options (claim #15389), including 2,500,000 options (claim #15385) and 2,500,000 unvested options with change-of-control acceleration (claim #15386). The 2018 Equity Incentive Plan reserved only 10,000,000 shares (claims #15460, #15373). What has not been said is the strike price, the vesting schedule for the unvested tranche, what counts as a change of control, and why half the plan reserve went to a single executive who also chairs the board. How was that grant priced relative to the grants of 50,000 to 200,000 options made to directors and the CFO on the same date (claims #15392 to #15396)?

The company disclosed on 2018-11-13 that Proposal No. 5 was withdrawn to comply with NASDAQ Rule 5635(b) (claim #15372). On 2018-12-11 it disclosed approval for a non-public issuance above the 20 percent threshold (claim #15333). What has not been said is who the intended investor or acquirer was, and whose holdings would have triggered a change of control under 5635(b). Was the later approved issuance directed to the same unnamed counterparty, and on what terms?

The company disclosed on 2018-08-13 a 3,000,000-share break-up fee tied to terminating the Global Bit Ventures merger (claims #15461, #15477). What has not been said is whether those shares were actually issued, and at what per-share value the fee was recorded. It is also unclear how the fee was adjusted, if at all, for the reverse split authorized on 2018-12-11 (claim #15332), given that only 8,720,405 common shares were outstanding by 2019-08-28 (claim #15152).

The company disclosed a Quebec data center lease through November 2022 valued at 120,150 USD (claim #15219) for a 26,700 square foot facility (claims #15229, #15230), plus operations at a Granby facility (claim #15198). What has not been said is whether 120,150 USD is a monthly, annual, or total-term figure, and whether the Quebec and Granby references describe the same site. The filings also give no street address, no contracted megawatt capacity, and no utility or interconnection arrangement behind a site deploying 1,400 mining units (claim #15220).

The company disclosed on 2019-08-28 a hosting arrangement priced at 0.055 USD/kWh (claim #15153), and on 2019-07-31 it disclosed third-party 24-hour security and financial operations management services (claims #15199, #15200). What has not been said is the name of the hosting provider or the service counterparties, the contract term, and whether any of these parties are related to management. With only three full-time employees (claim #15222), who actually operates the mining fleet, and what are the fixed fees beyond the per-kWh rate?

The company disclosed on 2019-07-01 the purchase of cryptocurrency patents from XpresSpa Group for 250,000 USD (claim #15218), along with a 30 percent waterfall interest in a patent special purpose entity (claim #15223). What has not been said is who holds the other 70 percent and what the waterfall priority is. The company also has not said how much of the 7,472,417 USD At The Market capacity with H.C. Wainwright (claims #15203, #15202) had been drawn. This matters given the repeated twelve-month cash runway uncertainty (claims #15462, #15197). How much cash did the ATM actually raise, and does any patent monetization proceeds flow to parties other than shareholders?

The company disclosed on 2018-06-28 that the Global Bit Ventures merger was terminated and that directors resigned, cited as claims #15477 and #15476, and on 2018-08-13 described a share-based break-up fee of 3,000,000 shares, cited as claim #15461 and marked delayed. What has not been said is which directors resigned and why, whether the 3,000,000 shares were actually issued, at what per-share value the fee was recorded, and whether the resignations were tied to the failed merger.

The company disclosed on 2018-10-15 that Merrick Okamoto received stock options described as 5,000,000 shares, cited as claim #15389, and separately as 2,500,000 options issued plus 2,500,000 unvested options carrying change-of-control acceleration, cited as claims #15385 and #15386. What has not been said is the exercise price, the vesting triggers for the unvested tranche, and whether this grant is connected to Proposal No. 5, which was withdrawn on 2018-11-13 over Nasdaq Rule 5635(b) change-of-control concerns, cited as claim #15372. The company also has not identified who would receive the non-public issuance of up to 20 percent approved on 2018-12-11, cited as claim #15333.

The company disclosed a 26,700 square foot Quebec mining facility, cited as claim #15230, leased through November 2022 at a figure of $120,150, cited as claim #15219, and bitcoin mining at a Granby facility, cited as claim #15198. What has not been said is whether $120,150 is a monthly, annual, or total-term amount, who the landlord or operating partner is, and the site's contracted power capacity in megawatts.

The company disclosed on 2019-08-28 the acquisition of 6,000 Bitmain S-9 miners, cited as claim #15150, hosted at an electricity cost of $0.055/kWh, cited as claim #15153, after reporting 1,400 deployed units on 2019-07-01, cited as claim #15220. What has not been said is the seller, the purchase price or consideration, the identity and location of the hosting provider, and when the 6,000 units would be energized, including the implied hashrate.

The company disclosed on 2019-07-01 a $250,000 cryptocurrency patent acquisition from XpresSpa Group, cited as claim #15218, alongside a 30 percent waterfall interest in a patent special purpose entity, cited as claim #15223. What has not been said is who holds the remaining 70 percent, whether any officer or director has a relationship with XpresSpa or the SPE, and whether the waterfall has produced any cash.

The company disclosed going-concern-style runway uncertainty on 2018-08-13 and 2019-07-31, cited as claims #15462 and #15197, repeated Nasdaq deficiencies, cited as claims #15383, #15243 and #15196, a reverse split authorization by 2018-12-31, cited as claim #15332, and a $7,472,417 ATM with H.C. Wainwright, cited as claims #15203 and #15202, while reporting three full-time employees, cited as claim #15222. What has not been said is the reverse split ratio and whether it was effected, how much of the ATM had been drawn and at what average price, and who provides the outsourced financial operations management and security services, cited as claims #15200 and #15199.

The company disclosed on 2018-10-15 that Merrick Okamoto was appointed Executive Chairman and CEO with a $350,000 base salary and 5,000,000 stock options (claims #15387, #15388, #15389). The same filing records 2,500,000 options issued to the CEO and 2,500,000 unvested options carrying change-of-control acceleration (claims #15385, #15386). What has not been said is how the 5,000,000 and 2,500,000 figures reconcile, what the strike price and vesting schedule are, and which grants would accelerate. Related to this, the company withdrew Proposal No. 5 on 2018-11-13 over NASDAQ Rule 5635(b), the change-of-control rule (claim #15372). Which holder or transaction would have crossed the control threshold, and how did that interact with the CEO's accelerated options?

The company disclosed a 26,700 square foot Quebec mining facility leased through November 2022 at a cited $120,150 (claims #15230, #15219, #15229). What has not been said is whether $120,150 is a monthly, annual or total-term figure. The company also has not given the facility's contracted power in megawatts, its Hydro-Quebec rate or interconnection terms, or the street address of the Granby site referenced separately (claim #15198).

On 2019-08-28 the company disclosed a hosting arrangement at $0.055 per kWh and the acquisition of 6,000 Bitmain S-9 Antminers (claims #15153, #15150). This followed a 2019-07-01 statement of 1,400 deployed units (claim #15220). What has not been said is who the hosting provider and the miner seller are, what was paid for the 6,000 machines, where they will be physically hosted, and what hosting term and minimum power commitment the $0.055 rate is tied to. The company also has not given a resulting hashrate figure that would let the unit counts be checked.

The company described third-party arrangements for its Granby mining operations, 24-hour security, and financial operations management (claims #15198, #15199, #15200) while reporting only 3 full-time employees (claim #15222). What has not been said is who these counterparties are, what they are paid, and whether any are affiliated with officers or directors. That matters given how much of the operation appears to be outsourced.

The company disclosed a $250,000 cryptocurrency patent acquisition from XpresSpa Group and a 30% waterfall interest in a patent special purpose entity (claims #15218, #15223). What has not been said is who holds the remaining interests in that entity, how the waterfall is ordered, and whether any director or officer has ties to XpresSpa. Separately, the terminated Global Bit Ventures merger carried a 3,000,000-share break-up fee (claims #15477, #15461). Were those shares actually issued, and at what recorded value?

The company put in place a $7,472,417 at-the-market program with H.C. Wainwright at a 3% commission (claims #15202, #15203, #15204). It did so while twice flagging doubt about twelve months of cash runway (claims #15462, #15197) and a Nasdaq minimum equity deficiency with a 2019-11-13 filing deadline (claim #15196). What has not been said is how much of the ATM has been drawn, at what average price and resulting share count, and whether the proceeds were enough to cure the equity deficiency. The company also has not said how the reverse split authorized for 2018-12-31 was used (claim #15332).

The company disclosed on 2018-10-15 that Merrick Okamoto received stock options covering 5,000,000 shares (claim #15389). It also disclosed 2.5 million options issued to the CEO (claim #15385) and 2.5 million unvested options with change-of-control acceleration (claim #15386). What has not been said is whether these are two halves of one 5,000,000-share grant or separate awards, and what the strike price, vesting schedule and expiry are. Separately, Proposal No. 5 was withdrawn on 2018-11-13 over NASDAQ Rule 5635(b) change-of-control concerns (claim #15372), so what did that proposal actually authorize, and did it involve the CEO's equity or a specific investor?

The company disclosed an all-stock purchase of 6,000 Bitmain S-9 miners for 2,335,000 common shares on 2019-08-28 (claims #15150, #15151, #15136). Post-transaction shares outstanding were 8,720,405 (claim #15152), and the seller was later named as SelectGreen at a $4,086,250 value (claims #15062, #15071). What has not been said is who beneficially owns SelectGreen, and whether any insider is connected to it. The block works out to roughly 27% of outstanding shares, so how was the deal structured to stay clear of the 20% issuance and change-of-control limits flagged in claims #15333 and #15372? It is also unexplained why $1,852,477 of the share issuance was still pending as of 2019-11-07 (claim #15066).

The company disclosed on 2019-08-28 that the new miners would be hosted at an electricity cost of $0.055/kWh (claim #15153). What has not been said is the hosting provider's name, the site location, the contracted megawatts, the contract term, and whether hosting fees sit on top of the power rate. It is also unclear whether those 6,000 units are deployed alongside the 1,400 units already disclosed (claim #15220).

The company disclosed on 2019-06-02 and 2019-07-01 a 26,700 sq ft Quebec mining facility leased through November 2022 at $120,150 (claims #15229, #15230, #15219). What has not been said is whether $120,150 is an annual, monthly or total-term figure, and what the street address and power capacity in MW are. Does the site have a direct Hydro-Québec interconnection or tariff allocation? The 2019-06-02 mention of Ether mining plans (claim #15231) has also never been followed by any deployment or abandonment update.

The company disclosed on 2019-07-31 and again on 2019-11-07 unnamed partners for Granby facility operations, financial operations management and 24-hour security services (claims #15198, #15200, #15199, #15068, #15070, #15069). With only three full-time employees (claim #15222), what has not been said is who these counterparties are, what they are paid, and whether any are related parties. Given how much of the business has been outsourced, the omission is significant.

The company disclosed a $7,472,417 at-the-market program with H.C. Wainwright on 2019-07-18 (claims #15202, #15203), with only $79,817 raised by 2019-11-11 (claim #15061). It also disclosed guidance that operating cash would suffice for twelve months (claim #15197). What has not been said is how many shares were sold and at what average price, and what cash balance and monthly burn support that guidance. Relatedly, what is the counterparty and economic value of the 30% waterfall interest in the patent special-purpose entity tied to the $250,000 XpresSpa patent purchase (claims #15223, #15218)?

The company disclosed on 2019-11-07 a Granby facility operations partnership, a 24-hour security services partnership and a financial operations services partnership (claims #15068, #15069, #15070), none of which names a counterparty; what has not been said is who operates Granby, where exactly the site is, how much power it has and on what terms, and whether the miners bought in 2020 (claims #14868, #14869, #14870) are hosted there or elsewhere. Who runs the site Marathon's hashrate depends on, and how many megawatts does it have?

The company disclosed on 2019-11-07 and again on 2020-08-13 the purchase of 6,000 S-9 miners from SelectGreen Blockchain for $4,086,250 (claims #15071, #15062, #14851, #14883), tied to a pending $1,852,477 share issuance (claim #15066) and stockholder approval for issuing more than 19.99% (claim #15048); what has not been said is how the price was split between cash and stock, how many shares SelectGreen received, whether SelectGreen became a significant or related holder, and whether the S-9s were ever deployed. Why do the August 2020 hashrate figures of 73 PH/s and 56 PH/s (claims #14857, #14856) appear to leave the 6,000 S-9s out?

The company disclosed on 2020-06-28, 2020-07-22 and 2020-08-05 a target of about 240 PH/s once the announced miners are deployed (claims #14906, #14887, #14871), but its itemized figures come to only 129 PH/s (claims #14857, #14856). The remainder depends on 1,000 S-19 Pro units expected only in 'Q4 2020' (claim #14855). What has not been said is the firm delivery date and payment status of those 1,000 units, whether the 660 mid-August units (claims #14873, #14854) actually arrived, and which MicroBT model was bought, given that filings variously say M30S+, M31S+ and M305+ (claims #14884, #14872, #14852). What is the energized, hashing capacity today versus 'when fully deployed'?

The company disclosed on 2020-06-28 and 2020-07-22 executive stock and cash awards totaling 1,164,000 restricted stock units (claims #14908, #14891), made while it reported 3 full-time employees (claim #14890) and a Nasdaq bid-price deficiency (claims #14943, #14893). What has not been said is who received the units, the vesting and performance conditions, the dollar value of the cash component, and whether the grants were benchmarked or approved by an independent committee. How were awards of this size justified for a three-person company facing a delisting notice?

The company disclosed on 2020-05-19 and 2020-07-22 an amendment reducing the conversion price on a $999,105.60 Bi-Coastal Consulting convertible note (claims #14913, #14892); what has not been said is the old and new conversion prices, how many shares the note now converts into, and who owns or controls Bi-Coastal Consulting. Is Bi-Coastal a related party, and what did the company receive in exchange for the lower conversion price?

The company disclosed at-the-market figures of $79,817 (claim #15061), $7,472,417 capacity (claim #14915), a $7.5M H.C. Wainwright agreement with a 2019-07-19 date (claim #14848), $7.1M sold in the first half of 2020 (claim #14849), a $6.9M underwritten offering of 5,293,246 shares (claims #14850, #14895) and a $100M S-3 (claim #14874). What has not been said is whether these amounts are gross or net, what the $79,817 figure represents, the average sale price and total share count, fees paid to Wainwright, and how much capacity remains. With the company claiming regained Nasdaq compliance as of August 6, 2020 (claim #14864), has it received written confirmation from Nasdaq, and how much further dilution is planned to pay for the Q4 miner deliveries?

The company disclosed an exclusive power purchase agreement with Beowulf on 2020-10-12, cited as claims #14791 and #14801, alongside a $14,000,000 Phase I construction cost (#14793) and a Phase II cost "per megawatt" (#14794). What has not been said is the PPA term, the contracted megawatts, the site's physical location and interconnection point, or the actual Phase II per-megawatt figure. Is the $0.006/kWh rate (#14800) the all-in power price, or only an operating and maintenance charge on top of an undisclosed energy tariff?

The company disclosed on 2020-08-13 a target of 294 PH/s once all S-19 Pro miners are installed, cited as claim #14858. That conflicts with the approximately 240 PH/s figure repeated on 2020-06-28, 2020-07-22 and 2020-08-05 (#14906, #14887, #14871), and with the 129 PH/s actually running (#14857 plus #14856). What has not been said is how the target moved, and whether 1,660 S-19 Pros (#14853) is right, given that the itemized batches of 660 and 500 (#14885, #14886) total only 1,160. The same filings call the MicroBT unit an M30S+, M31S+ and M305+ (#14884, #14872, #14856), so which model was actually bought?

The company disclosed on 2020-08-17 a $23,159,173.80 purchase of 10,500 Antminer S-19 Pros with a minimum aggregate hashrate of 1,155,000 TH/s, delivered through June 2021, cited as claims #14842, #14843, #14844 and #14845. What has not been said is the tranche-by-tranche delivery schedule, the deposit already paid versus the balance still owed, and remedies if hashrate falls short. It also has not said where these units will be hosted and at what power cost, given that Compute North (#14861) and the Beowulf site (#14791) both appear without allocation.

The company disclosed on 2020-07-22 and 2020-05-19 an amendment reducing the conversion price on a $999,105.60 convertible note held by Bi-Coastal Consulting, cited as claims #14892 and #14913. What has not been said is the old and new conversion price, the resulting share count and dilution, why a lower price was granted while the company was issuing equity (#14895, #14850), and whether Bi-Coastal has any relationship with insiders.

The company disclosed on 2020-06-28 and 2020-07-22 executive stock and cash awards of 1,164,000 units, cited as claims #14908 and #14891. At the same time it reported three full-time employees (#14890) and flagged going-concern doubt (#14865). What has not been said is which executives received the grants, how they split, the vesting or performance conditions, and the cash component's dollar value. The 3,000,000-share issuance and 150,000-share Phase I completion bonus tied to the O&M agreement (#14796, #14797, #14792) leave a similar question: who is the counterparty?