jain.com — data centers, connectivity and security, analyzed daily

Category: AI Infrastructure

  • Sharon AI’s 9.95% GPU Loan Shows Lenders Underwrite AI Chips Through Their Contracts

    Sharon AI’s 9.95% GPU Loan Shows Lenders Underwrite AI Chips Through Their Contracts

    TL;DR · 30-second read

    The Short Version

    An Australian company called Sharon AI has borrowed 356 million US dollars and pledged the computer chips it is buying as security, much like a mortgage uses the house.

    The chips are powerful Nvidia processors that run artificial intelligence. The loan costs about 10 percent a year, and Goldman Sachs is among the lenders.

    The twist: lenders get repaid from what customers pay to rent those chips. So the loan really rests on Sharon AI’s customer deals, and it is only the first of several loans behind a plan to install 68,000 chips by mid-2027.

    SharonAI Holdings Inc. (NASDAQ: SHAZ), an Australian provider of AI computing infrastructure, announced on October 1, 2026 that it has entered its first GPU-backed debt facility: a US$356 million committed, senior secured loan held in a special purpose vehicle and priced at a fixed 9.95%, excluding fees. The facility is secured against the GPUs and their associated cash flows, and proceeds will fund compute capacity dedicated to customer contracts.

    Lenders include Goldman Sachs and select large private credit funds alongside Australian, Asian and global investors, with Jarden Australia acting as sole financial advisor and arranger. Sharon AI called it the first in an expected series of GPU financings supporting more than 68,000 NVIDIA GPUs deployed by mid-2027, and said it has now secured more than US$2.6 billion of debt and equity capital over ten months against a customer contract book worth more than US$8.8 billion.

    Executive Summary

    Sharon AI has opened a new funding channel for its GPU fleet: debt secured directly by the chips and the customer revenue they generate. The US$356 million facility is modest against the company’s 68,000-GPU build-out plan, but Sharon AI presents it as the template for a series of similar financings, shifting part of its hardware funding from equity to borrowed money at a fixed 9.95% rate before fees.

    The significance lies in the structure. Because the security package combines hardware and contracted cash flows, the lenders are effectively underwriting Sharon AI’s customer contracts, with GPU resale value as the fallback. That makes the quality, length and enforceability of those contracts, rather than raw chip count, the variable that determines how much Sharon AI can borrow and at what price.

    The deal also lands days after Sharon AI disclosed a change of auditor to Ernst & Young and announced a confidential-computing collaboration with VAST Data aimed at regulated customers, two developments that bear on how lenders and investors will judge the company’s reporting and its contract pipeline.

    The Chip Is the Collateral, the Contract Is the Repayment

    Sharon AI describes the facility as senior secured and GPU-backed, held in a special purpose vehicle (SPV), a separate legal entity set up to own specific assets and borrow against them. The security is “the GPUs and associated cash flows,” and proceeds will fund compute “dedicated to customer contracts.” Senior secured means these lenders are paid first and hold a direct claim on the pledged assets.

    Put those pieces together and the lenders’ position becomes clear. In normal operation, the debt is serviced from what customers pay to use the specific GPUs inside the SPV. The chips themselves matter mainly if something goes wrong: if a customer stops paying or the SPV defaults, lenders can claim the hardware and its revenue stream. From the lender’s seat, each GPU is worth largely what its attached contract is expected to earn over the life of the loan, with resale value as the backstop.

    That has consequences beyond Sharon AI’s balance sheet. Graphics processors are a fast-moving asset class, and each new NVIDIA generation can change what older chips fetch on the secondary market. A lender relying on resale value alone would carry that risk directly. Tying security to contracted cash flows moves the central credit question from what a chip will be worth at the end of the loan to whether the customer will keep paying, which makes the strength and duration of the customer book the main driver of how cheaply an AI cloud operator can finance hardware.

    What a Fixed 9.95% Buys, and What It Leaves Out

    The 9.95% rate is fixed, so Sharon AI’s interest bill will not rise if benchmark rates do, a useful match for revenue that comes from multi-year contracts at agreed prices. The figure explicitly excludes fees, however. Arrangement and commitment fees on structured facilities can be meaningful, so the all-in cost of this borrowing is higher than 9.95% by an amount the company has not stated.

    Chief Executive James Manning said the structure is “designed to enhance return on equity.” The mechanism is simple: if the GPUs earn more than the cost of debt, borrowing lets shareholders keep the difference on a smaller equity base. The arithmetic also runs in reverse. If utilization, pricing or customer payments fall short, leverage magnifies losses for equity holders, who rank behind senior lenders. Whether the claim holds depends on contract economics Sharon AI has not published.

    The lender roster adds context. Goldman Sachs and “select large private credit funds” sit alongside Australian and Asian investors. Private credit, meaning lending by investment funds rather than banks, commonly provides asset-backed loans of this kind, and a committed facility (money promised up to a limit) gives Sharon AI drawable capacity as deployments come online.

    $356 Million Is the Opening Tranche of a 68,000-GPU Plan

    Sharon AI calls the facility “the first in an expected series of GPU financings” behind a scheduled build-out of more than 68,000 NVIDIA GPUs by mid-2027. It has not said how many of those chips this US$356 million covers, but the framing is clear: this is a repeatable structure, not the whole bill.

    Two company figures set the scale. Sharon AI says it has secured more than US$2.6 billion of institutional debt and equity over the past ten months, and that its customer offtake, meaning contracted commitments to buy compute, stands at a total contract value (TCV) above US$8.8 billion. TCV sums the full value of contracts across their entire terms, so it is not annual revenue, and its worth to a lender depends on contract length, cancellation rights and counterparty strength. Sharon AI lists hyperscalers (the largest cloud platforms), AI-native companies, government, enterprise and research organizations among its customers, without naming those behind this facility.

    The company is also building product features for regulated buyers. On September 23 it announced a collaboration with VAST Data to offer VAST DataEnclave, a confidential-computing environment in which, the companies say, customer data and model weights remain encrypted even from Sharon AI as operator, targeting banks, government agencies and other regulated organizations. If such features help win long-dated contracts from institutional customers, they feed directly into the cash flows that secure facilities like this one.

    Reporting Discipline Matters More Once Debt Is Layered In

    Lenders monitor borrowers through their financial reporting, so audit quality becomes more important as debt is added. In an 8-K filed September 24, Sharon AI disclosed that its Audit Committee approved ending HoganTaylor LLP’s engagement on September 20 and engaged Ernst & Young on September 23 for the fiscal 2026 audit, following a competitive process involving several firms.

    The filing states there were no disagreements with HoganTaylor and that its audit reports for 2024 and 2025 carried no adverse or qualified opinions. It does list one reportable event: a material weakness, meaning a significant gap in internal controls over financial reporting, identified by HoganTaylor and disclosed in the 10-K filed March 31, 2026. Moving to a Big Four auditor is a common step for a growing public company, and the filing gives no indication of a dispute. It does not say whether the material weakness has been remediated, and for a borrower planning a series of GPU-backed facilities, reliable controls over cash flows and asset records are part of what lenders are underwriting.

    Background

    SharonAI Holdings Inc. trades on Nasdaq under the ticker SHAZ and describes itself as an Australian “neocloud,” a specialist provider that rents GPU computing power for AI rather than offering general-purpose cloud services. Through its AI Factory platform and a network of technology and co-location partners (data center operators that host its equipment), it serves organizations training and running AI models, with an emphasis on sovereign infrastructure that keeps data within national borders. It reports to the SEC as an emerging growth company.

    Companies in this segment typically need large sums up front to buy GPUs before customer revenue arrives, which makes financing structure central to their growth. Sharon AI says its ambition is gigawatt-scale compute capacity across Australia, New Zealand and the broader Asia-Pacific, a gigawatt being roughly the output of a large power station.

    Sources

    Source: Sharon AI Enters Into GPU-Backed Debt Facility, Expanding Funding Flexibility for AI Factory Deployments, Sharon AI’s announcement of its US$356 million GPU-backed SPV debt facility.

    Primary sources: SharonAI Holdings 8-K filed September 24, 2026 (change in certifying accountant); SharonAI Holdings 8-K/A filed September 23, 2026; Exhibit 99.1 to 8-K/A: Sharon AI Collaborates with VAST Data on the Launch of DataEnclave; SharonAI Holdings 8-K filed September 23, 2026; Exhibit 99.1 to 8-K: Sharon AI Collaborates with VAST Data on the Launch of DataEnclave.

  • Broadcom’s Up-to-$42B Anthropic Loan Puts AI Compute Credit Risk on a Chipmaker

    Broadcom’s Up-to-$42B Anthropic Loan Puts AI Compute Credit Risk on a Chipmaker

    TL;DR · 30-second read

    The Short Version

    Broadcom, one of the world’s biggest chip makers, has agreed to lend Anthropic, the company behind the Claude chatbot, up to $42 billion, according to a financial filing. The money is for Anthropic to rent Broadcom’s own chips.

    Think of a carmaker lending you the money to lease its cars. If you stop paying, the carmaker takes the loss.

    That is the real story: some of the risk that an artificial intelligence company cannot pay its bills now sits with the company selling it the hardware.

    Broadcom has agreed to lend Anthropic up to $42 billion so the AI developer can lease Broadcom’s chips, Reuters reported in an exclusive, citing a filing. The arrangement makes Broadcom both the supplier of Anthropic’s AI computing hardware and the source of the money used to pay for it.

    The figure is a ceiling, not a fixed sum. How much Anthropic will draw, over what period, and on what terms has not been made public by either company.

    Executive Summary

    The headline number is large, but the structure is what matters. In a typical hardware deal, a chipmaker sells, gets paid, and any question about whether the buyer can afford the purchase belongs to the buyer and its lenders. Here, the chipmaker is also the lender. If the facility is drawn, Broadcom holds a claim on Anthropic’s future ability to pay, not just a completed sale.

    For Anthropic, the facility offers a way to secure large amounts of AI computing capacity without paying for it upfront. For Broadcom, it helps lock in demand for its chips with one of the most prominent AI model developers. For everyone else watching AI infrastructure spending, it is a concrete example of compute financing moving onto a supplier’s balance sheet, and it raises fair questions about collateral, pricing and how the resulting revenue should be read.

    When the Chip Vendor Becomes the Bank

    The mechanism is simple. Broadcom extends credit of up to $42 billion; Anthropic uses it to lease Broadcom chips; Anthropic repays over time from the money it earns selling AI products. Every dollar drawn becomes a dollar of exposure for Broadcom to a single borrower, and that exposure is only as good as Anthropic’s future revenue. This is what the term credit risk means here: the chance that a borrower cannot repay in full and on time.

    That risk now sits on a chipmaker’s books rather than solely with banks, bond investors or specialist equipment lessors, the parties that normally absorb it. The people affected are Broadcom shareholders and creditors, who inherit exposure to the economics of AI model development; credit analysts, who must now assess a semiconductor company partly as a lender; and Anthropic itself, whose funding for compute is tied to one supplier’s balance sheet.

    None of this makes the arrangement unsound. A loan tied to hardware may be secured by that hardware, and a large, profitable chipmaker can be a more patient lender than a bank. But the risk does not disappear because the lender is also the seller. It changes hands.

    Why Lease, and Why Let the Supplier Pay

    Leasing chips, rather than buying them, lets a customer spread the cost over the period it uses the hardware instead of paying everything upfront. For an AI developer whose computing needs are growing quickly, that matches cash going out to the revenue the chips help generate. Financing the lease through the chip supplier can also be faster than arranging a syndicated loan, and the supplier has unusual insight into the value of the hardware being financed.

    For Broadcom, the logic is demand certainty. Custom AI chips are designed over long cycles, and a customer committed through a multi-billion-dollar financing facility is a customer less likely to switch suppliers. In effect, the loan helps secure the order book. The trade-off is that a portion of Broadcom’s future chip revenue would be funded with Broadcom’s own capital.

    What Credit and Equity Analysts Will Look For

    The first question is revenue quality. When a vendor finances its own sales, the revenue it reports is only fully realized once the customer repays. Investors will want to see how Broadcom accounts for the leases and the loan, and how much of any reported AI revenue is tied to this facility.

    The second question is collateral. AI chips lose value as newer generations arrive, so hardware that secures a loan today may be worth considerably less within a few years. A lender’s protection depends on how fast the loan amortizes relative to how fast the hardware depreciates.

    The third is precedent. Customer financing by telecom equipment makers around 2000 is the case credit analysts most often cite, because losses mounted when financed customers failed. The comparison is a question to test, not a verdict: the outcome depends on the borrower’s revenue, the lender’s concentration, and the security behind the loan, none of which have been fully disclosed here.

    The Ripple Effect Beyond the Two Companies

    Chips are only one part of AI computing capacity. Leased accelerators still need data center space, cooling, networking and, above all, electricity. As reported, this facility is tied to chips, so operators and utilities serving any resulting deployment will still look to other parties for commitments on buildings and power.

    For competing chip suppliers and AI cloud providers, the deal sets a visible reference point. Customers negotiating large hardware orders may ask whether financing is on the table. One transaction does not establish an industry pattern, but it does show that supplier balance sheets are now part of how frontier AI compute gets paid for.

    Background

    Broadcom is a large US semiconductor and infrastructure software company. In AI infrastructure, it is best known for designing custom accelerator chips for major cloud and AI companies and for the networking silicon that links thousands of processors together. Custom chips are an alternative to general-purpose graphics processors, and they tie a customer to a specific design partner over multi-year development cycles.

    Anthropic, founded in 2021, builds the Claude family of AI models. Like other developers of frontier AI models, it needs enormous amounts of computing capacity to train and serve its models, which makes access to chips, and the financing to pay for them, a central business constraint.

    Sources

    Source: EXCLUSIVE: Broadcom to lend Anthropic up to $42 billion to lease its chips, filing says, Reuters’ report on Broadcom’s financing facility for Anthropic’s chip leases.

  • SanDisk and Kioxia’s $31B Through 2032 Puts AI Storage Supply on a Fab Clock

    SanDisk and Kioxia’s $31B Through 2032 Puts AI Storage Supply on a Fab Clock

    TL;DR · 30-second read

    The Short Version

    Two companies that make the memory chips inside phones, laptops and the giant computer warehouses behind artificial intelligence plan to spend $31 billion expanding their factories in Japan through 2032.

    The surprising part is the calendar. Demand from artificial intelligence can jump within months, but a chip factory takes years to build, equip and get running smoothly. So the extra storage this money buys will arrive in stages over the next six years or so, not whenever technology companies happen to need it.

    That timing helps decide how much storage costs and how fast new supply shows up.

    SanDisk and Kioxia plan to invest $31 billion to expand memory chip production in Japan through 2032, scanx.trade reported. The two companies are long-time partners in NAND flash, the type of memory chip that stores data inside solid-state drives (SSDs), smartphones and memory cards, and have jointly operated fabrication plants (fabs) in Japan since the early 2000s.

    Executive Summary

    The plan is one of the largest multi-year commitments of capital to flash memory manufacturing announced by the two partners, and it lands at a moment when AI data centers are consuming growing volumes of high-capacity storage for training data, model checkpoints and the retrieval systems that feed AI models at inference time.

    What makes the announcement analytically interesting is less the headline total than its end date. A spending horizon running to 2032 means the additional capacity will come online in phases dictated by how quickly cleanrooms can be built or refitted, manufacturing tools installed and new chip generations brought to acceptable yields. Buyers planning AI storage in the next two years will largely be served by capacity decisions already in motion, not by the later stages of this plan.

    The $31 billion figure also cannot yet be converted into a volume of storage. How the money divides between new wafer capacity and upgrades to existing lines, and when each phase produces output, will determine its real effect on supply and pricing.

    A 2032 End Date Puts AI Storage on a Fab Clock

    AI data center projects are typically planned and energized in phases measured in quarters: servers ordered, power contracted, halls fitted out. Flash memory supply moves differently. Before a new wafer of memory ships, cleanroom space has to be built or converted, deposition and etch tools have to be delivered and installed, and each new generation of 3D NAND (chips built by stacking memory cells in vertical layers) has to be ramped until enough chips on each wafer work. A plan stretching to 2032 sequences those steps across roughly six years.

    Simple arithmetic illustrates the pacing. If the $31 billion were spread evenly from now through 2032, it would amount to roughly $4 billion to $5 billion a year. That is substantial, but it is a flow, not a lump. The rate at which the money is spent, and the order in which tools arrive, determines when extra storage reaches the market. A surge in AI storage demand in 2027 cannot be met faster than the fab schedule allows, however large the total commitment.

    The people most exposed to that mismatch are cloud providers, colocation operators and enterprises sizing storage for AI clusters, along with the SSD makers who sell to them. For them, the practical implication is that flash availability is set years in advance. Long-term supply agreements and early procurement become the tools for managing the gap between how quickly AI demand can shift and how slowly memory capacity can follow.

    Why $31 Billion Does Not Translate Directly Into Capacity

    In NAND manufacturing, capital spending serves two different purposes. Some of it builds new wafer capacity: more fab space and more tools processing more silicon. Much of it funds technology transitions, moving existing lines to chips with more stacked layers so that each wafer yields more gigabytes. Both raise output, but at very different costs and on different timelines. Until the partners disclose how the plan divides between the two, the headline figure says how much will be spent, not how much storage it will produce.

    The memory industry’s history also argues for reading the total as an envelope rather than a fixed schedule. Flash has gone through repeated cycles in which capacity added during strong demand led to oversupply and falling prices. Multi-year programs in this sector have historically been paced to market conditions, and a plan with a distant end date leaves room to accelerate or slow. That flexibility is prudent for the manufacturers; for buyers, it means the plan’s effect on prices will depend on execution choices made year by year.

    Concentration in Japan Cuts Both Ways

    Keeping the expansion in Japan builds on the partners’ existing manufacturing base at Yokkaichi in Mie Prefecture and Kitakami in Iwate Prefecture, where shared infrastructure, trained workforces and supplier ecosystems lower the cost of adding capacity. It also aligns with Japan’s broader push to rebuild domestic semiconductor manufacturing.

    Concentration carries risk as well. A 2022 contamination incident disrupted output at the joint venture’s plants, a reminder that a large share of global flash supply sitting in a few locations can turn a local problem into a market-wide shortage. Large storage buyers that value supplier diversity will weigh this plan alongside capacity decisions by Samsung, SK hynix and Micron, which will shape total market supply at least as much as any single program.

    Background

    SanDisk and Kioxia, which was known as Toshiba Memory until 2019, have been partners in flash memory manufacturing since the early 2000s, sharing investment in and output from fabs at Yokkaichi and Kitakami in Japan. SanDisk was acquired by Western Digital in 2016 and re-emerged as the standalone Sandisk Corporation in February 2025, while Kioxia listed on the Tokyo Stock Exchange in December 2024.

    The NAND flash market is dominated by a handful of manufacturers, including Samsung, SK hynix and Micron alongside the SanDisk-Kioxia partnership. It is known for pronounced boom-and-bust cycles, and demand from AI data centers for high-capacity solid-state drives has become an increasingly important driver of the industry’s investment decisions.

    Sources

    Source: SanDisk, Kioxia plan $31 billion Japan memory expansion through 2032 (scanx.trade), a report on the partners’ multi-year flash memory manufacturing investment in Japan.

  • Bitdeer’s $1.7B Malaysia Offtake Shows AI Cloud Earns ~9x Colocation per Megawatt

    Bitdeer’s $1.7B Malaysia Offtake Shows AI Cloud Earns ~9x Colocation per Megawatt

    TL;DR · 30-second read

    The Short Version

    Bitdeer, a company best known for bitcoin mining, has signed customers for most of a new artificial intelligence computing center in Malaysia. The site draws about as much electricity as a small town, and those customers are expected to pay more than $1.7 billion over five years.

    The surprise is the rate. For the same amount of electricity, this business brings in roughly nine times what Bitdeer earns simply renting building space and power to a tenant in Norway.

    The catch: Bitdeer generally has to supply the costly computer chips itself, and some of its borrowing costs 13% a year.

    Bitdeer AI, the AI cloud unit of Nasdaq-listed Bitdeer Technologies Group (BTDR), announced off-take commitments covering more than 70% of its 21.7-megawatt (MW) A201 data center in Malaysia, with more than $1.7 billion in expected revenue over five years, Investing.com and MarketScreener reported. An off-take commitment is a customer’s contractual promise to buy a set amount of capacity over a fixed term.

    The deal follows a 6-K filed September 29, in which Bitdeer told US regulators that Bitdeer AI had already sold out the roughly 9.5MW A102 site in Malaysia in August on five-year commitments, and that its Norwegian subsidiary had signed a separate colocation lease worth about $4.7 billion over 16 years.

    Executive Summary

    On its face this is a contract announcement: roughly 15MW of AI computing capacity in Malaysia, sold forward for five years. The more useful number is what it implies per megawatt. Spread evenly, $1.7 billion over five years on about 15MW comes to roughly $22 million of revenue per megawatt per year.

    Bitdeer’s own filings provide an unusually clean comparison. Its Tydal, Norway colocation deal, disclosed in the same 6-K, is expected to earn an average of $2.4 million per IT megawatt per year. The same company, selling power-backed capacity two different ways, is booking roughly nine times more revenue per megawatt when it sells finished AI compute rather than space, power and cooling.

    That gap is the story. It shows where the revenue in AI infrastructure sits for a company with power but no hyperscale customers of its own, and it also shows the trade-off: the higher top line comes with GPU purchases, shorter terms and far less of the landlord-style margin Bitdeer projects in Norway.

    The Math: Roughly $22 Million per Megawatt-Year

    Start with the disclosed figures. More than 70% of 21.7MW is at least 15.2MW. More than $1.7 billion over five years is at least $340 million a year. Dividing one by the other gives roughly $22 million per contracted megawatt per year. Because both figures are floors (more than 70%, more than $1.7 billion), the true rate could move modestly in either direction, but the order of magnitude holds.

    Now set that beside Tydal. In its 6-K, Bitdeer said the Norway colocation lease runs at an average of about $202 per kilowatt per month, or $2.4 million per IT megawatt per year, with electricity reimbursed separately by the tenant. Colocation means renting out a powered, cooled building; the tenant brings its own servers. AI cloud means renting out the computing itself, usually by the GPU-hour or under reserved contracts, with the provider owning the servers. Comparing $22 million with $2.4 million, Malaysia generates roughly nine times the revenue per megawatt.

    Two caveats keep this honest. Bitdeer has not said whether 21.7MW refers to IT load (power delivered to servers) or gross site capacity, and AI cloud pricing typically bundles electricity into the service. Neither caveat closes a ninefold gap. The bulk of the difference is the hardware and software layer sitting on top of the power.

    Revenue Density Is Not Margin

    The operational conclusion runs the other way from the headline number. At Tydal, Bitdeer estimates a net operating income margin of about 90%, because a landlord’s main costs are the building and its upkeep. At A201, Bitdeer AI must supply the accelerators, networking and storage that generate the $22 million, and that equipment is where most of the capital in an AI cloud goes. A five-year commitment tries to match contract length to the working life of that hardware, so the question for margins is how much of each megawatt’s revenue goes toward paying off the servers before the contract ends.

    That makes financing the real constraint on how far this model scales. Bitdeer’s interim financial statements show $373.5 million of principal on its asset-collateralized loans from BIT Group due within twelve months of June 30, 2026, at an effective interest rate of 13.0% for the first half of the year, plus $375 million of 5.00% convertible notes issued in February. Buying GPUs for revenue this dense can pay back quickly, but only if the contracted customers keep paying for the full term.

    Two Business Models, One Power Portfolio

    Bitdeer is now running both models side by side. In Norway, it is a landlord: 121 IT MW leased to one tenant for 16 years, backed by about $1.3 billion in letters of credit, a form of bank guarantee. In Malaysia, it is a cloud operator: small sites of 9.5MW and 21.7MW sold on five-year terms. The first model trades revenue per megawatt for duration and credit quality; the second trades duration for density.

    Against a total portfolio the company puts at roughly 3.0 gigawatts, including 742MW of existing and pipeline capacity in Milam County, Texas, the Malaysian sites are a small slice by power. They matter as a price signal. For a former bitcoin miner deciding how to use each additional megawatt, the choice is between a long, low-risk lease and a short, capital-heavy compute business earning several times more. Bitdeer’s filings suggest it intends to keep doing both.

    Background

    Bitdeer Technologies Group is a Cayman Islands company listed on the Nasdaq under the symbol BTDR. Its business spans bitcoin self-mining, cloud hash rate, mining-rig sales, hosting and, more recently, AI cloud services, all of which appear as separate revenue lines in its interim financial statements. Like several large bitcoin miners, it controls substantial power capacity, which it now says totals about 3.0 gigawatts across sites in the US, Norway, Bhutan, Ethiopia, Malaysia and elsewhere.

    AI workloads need that same scarce resource, grid power, which is why miners have been repurposing capacity for AI. They generally do so in one of two ways: leasing powered buildings to AI tenants as colocation landlords, or buying GPUs and selling compute directly. Bitdeer is doing both, with a large 16-year lease in Norway and smaller five-year AI cloud contracts in Malaysia.

    Sources

    Source: Bitdeer AI secures $1.7 billion in Malaysia data center contracts (Investing.com): coverage of Bitdeer AI’s off-take commitments for more than 70% of its 21.7MW A201 Malaysia data center.

    Primary sources: Bitdeer Technologies Group Form 6-K, filed 2026-09-29 (Report of Foreign Private Issuer); Exhibit 99.1 to 6-K filed 2026-09-29: unaudited interim consolidated financial statements; Exhibit 99.2 to 6-K filed 2026-09-29: Recent Developments.

  • AMD’s $8.2B World Labs Deal Makes AI Model Research a Chip Design Input

    AMD’s $8.2B World Labs Deal Makes AI Model Research a Chip Design Input

    TL;DR · 30-second read

    The Short Version

    AMD, one of the biggest makers of the chips that run artificial intelligence, is buying World Labs, a young company whose software can build realistic, explorable three-dimensional worlds from a sentence, a photo or a video.

    AMD is paying about $8.2 billion, all in its own shares rather than cash. World Labs’ co-founder, the prominent researcher Fei-Fei Li, becomes AMD’s chief scientist.

    The bet behind the price is that to design tomorrow’s chips, a chipmaker needs to understand tomorrow’s artificial intelligence from the inside.

    Advanced Micro Devices has agreed to acquire World Labs Technologies, a developer of spatial-intelligence AI models, for approximately $8.2 billion, to be paid entirely in AMD common stock. In an 8-K filed September 28, AMD disclosed that it signed the merger agreement on September 26. The price is subject to customary adjustments, and the number of shares is not yet fixed: it will be set by AMD’s 10-day volume-weighted average share price ending two trading days before closing.

    World Labs builds models that create, reconstruct and simulate interactive 3D environments from text, images and video, and it develops technology for robot learning and simulation. Benzinga reported that co-founder and CEO Fei-Fei Li will join AMD as executive vice president and chief scientist. AMD shares were up 0.08% at $608.38 in after-hours trading on Monday.

    Executive Summary

    AMD is paying $8.2 billion in stock for a research team and its models, not for a chip product, a factory or a customer base. AMD’s stated rationale is explicit. CEO Lisa Su said that building compute platforms for the next generation of AI “requires a deep understanding of how models are evolving,” and that World Labs’ insight will be used to “develop the hardware, software and systems that will power the next generation of AI and strengthen the open AI ecosystem.”

    That framing matters to anyone who builds or buys AI infrastructure. Accelerator roadmaps are set years ahead, so a chipmaker is always betting on what the workloads of the future will look like. AMD is now paying for a first-hand view of one frontier class of workload: 3D world generation, simulation and robot training. The move puts AMD on the same path Nvidia has taken, from selling silicon toward owning more of the software and model layers that generate demand for it.

    What AMD has not yet shown is how that insight turns into products, what World Labs earns today, or when the deal closes. The strategic logic is coherent. The return on $8.2 billion is still undemonstrated.

    Why a Chipmaker Is Buying a Model Lab

    An AI accelerator is the specialised processor that trains and runs AI models. Designing one means committing years in advance to specific choices: how much memory sits next to the compute, how fast chips talk to each other, which numerical formats get dedicated hardware. Each choice is a wager on what models will need by the time the silicon ships. Get it wrong, and a chip can be fast on paper but poorly matched to the software customers actually run.

    AMD’s announcement argues directly that owning model research reduces that risk. Su’s statement ties the acquisition to understanding “how models are evolving” and then feeding that understanding into hardware, software and systems. On that reading, the $8.2 billion is mainly a price for an internal source of workload signal. World Labs’ team, led by Li as AMD’s new chief scientist, becomes an in-house customer that pushes AMD’s platforms with frontier models and reports back where they fall short.

    The choice of World Labs points to a specific category of demand. Generating and simulating interactive 3D environments, and training robots inside those simulations, is a different workload from text chatbots. If AMD believes spatial and physical AI will be a meaningful share of future data center compute, owning a lab at that frontier gives its chip architects earlier and more direct evidence than waiting for customers to report problems. That affects AMD’s accelerator roadmap first, and then the cloud providers and data center operators who decide which hardware to deploy.

    Following Nvidia Up the Stack, With an ‘Open’ Twist

    Nvidia’s lead in AI has never rested only on chips. It has paired hardware with a deep software layer and has released its own models and tools for robotics and simulation, the same territory World Labs works in. The more of that stack a vendor controls, the more it shapes which workloads exist and how well they run on its hardware. AMD’s purchase moves it in the same direction, from supplying accelerators toward influencing the demand for them.

    AMD is framing the move differently, as a way to “strengthen the open AI ecosystem.” That fits AMD’s long-running pitch as the more open alternative, built around its open-source ROCm software stack. The phrase also raises a question AMD has not yet answered: whether World Labs’ models will be released openly, optimised for AMD hardware, or kept as internal research tools. Each option sends a different signal to AMD’s customers.

    That signal matters because AMD’s biggest accelerator buyers include cloud providers and AI developers who build their own models. A chip supplier that also owns a well-known model lab walks a line between learning from the frontier and competing with the customers it sells to. Nvidia has managed this tension so far. How AMD manages it will depend on choices it has not yet disclosed.

    An All-Stock Price Tag With a Moving Share Count

    Paying in stock rather than cash lets AMD make an $8.2 billion acquisition without drawing down its balance sheet. The cost falls instead on existing shareholders, through dilution. According to the 8-K, the share count will be set by AMD’s volume-weighted average price over the 10 trading days ending two days before closing. As a rough illustration only, at Monday’s after-hours price of about $608, $8.2 billion would work out to roughly 13.5 million shares. The actual figure will move with the stock.

    The shares will be issued privately to World Labs’ stockholders under Section 4(a)(2) and Regulation D exemptions, not through a public offering. That is a standard route for paying the owners of a privately held company in stock. The market’s first reaction was muted: a 0.08% after-hours move. That suggests investors see the deal as manageable in size relative to AMD, with the payoff still to be proven.

    That caution seems justified on the available disclosures. The announcement includes no World Labs revenue, no product integration plan and no timeline. The case for the price rests on Su’s strategic argument and on the calibre of the team. Both may prove sound, but neither is yet a financial result.

    Background

    AMD, headquartered in Santa Clara, California, makes CPUs, graphics processors and Instinct accelerators for AI data centers, where it is the main challenger to Nvidia. AMD has built its data center AI pitch around competitive hardware paired with ROCm, its open-source software stack, and presents itself as the open alternative in a market where Nvidia’s hardware-plus-software ecosystem dominates.

    World Labs is a privately held AI company co-founded by Fei-Fei Li, a Stanford computer scientist known for creating ImageNet. The company focuses on spatial intelligence: models that generate and simulate interactive 3D worlds, and tools for training robots in simulation. That field sits where AI research meets robotics, and a growing share of it depends on large-scale data center compute.

    Sources

    Source: AMD to Buy AI Firm World Labs in $8.2 Billion Deal (Benzinga): AMD agrees to acquire spatial-intelligence company World Labs in an all-stock deal, with Fei-Fei Li joining as chief scientist.

    Primary sources: AMD Form 8-K filed 2026-09-28 (Item 3.02, Unregistered Sales of Equity Securities); AMD to Acquire World Labs to Advance the Future of AI Compute (GlobeNewswire).

  • Oracle’s 3.5% Drop: A $664B AI Backlog Is Only as Good as Its Build Dates

    Oracle’s 3.5% Drop: A $664B AI Backlog Is Only as Good as Its Build Dates

    TL;DR · 30-second read

    The Short Version

    Oracle, the business software giant, holds about $664 billion in signed contracts to rent out computing power for artificial intelligence. It only gets paid once the buildings that house those computers are finished and switched on.

    A huge New Mexico site built for OpenAI faces local objections over water and air, and a setback on the gas pipeline meant to power it. Banks have struggled to resell $18 billion of loans for the site at full price, and Oracle’s shares fell 3.5% on fears of delay.

    The lesson: a promise of future business is only as good as the construction schedule behind it.

    Oracle shares fell 3.5% on concerns about data center delays, TradingView reported. The pressure centers on Project Jupiter, a New Mexico data center that is part of Oracle’s AI infrastructure expansion with OpenAI. The Financial Times reported, as relayed by Reuters, that about $18 billion of loans tied to the project were quoted at 89 to 91 cents on the dollar, and that banks had struggled to distribute the debt to other investors as concerns about Oracle’s leverage grew.

    Jupiter faces local opposition over water and air quality, and a proposed natural-gas pipeline to serve it has suffered a regulatory setback. Insider Monkey, summarizing Oracle’s latest figures, set those problems against a remaining contract backlog of $664 billion, fiscal 2026 capital spending of $55.7 billion and $125.3 billion of borrowings as of August 31.

    Executive Summary

    Oracle has no shortage of AI demand. Its remaining performance obligations, meaning revenue it has contracted but not yet earned, reached $664 billion, up $209 billion year over year. Cloud infrastructure revenue rose 121% to $7.4 billion in the first quarter of fiscal 2027. The question the market is now asking is when that backlog becomes cash. That depends on physical sites being permitted, powered and energized on schedule.

    Project Jupiter shows how that dependency can bite. Opposition over water and air quality and a setback on a gas pipeline have raised questions about the site’s development. The $18 billion of debt behind it is being quoted below face value. S&P downgraded Oracle in July to one notch above junk, and Oracle’s spending already runs well ahead of its cash generation: free cash flow was negative $23.7 billion in fiscal 2026. At that point, schedule risk at a single campus becomes a balance-sheet question.

    This matters beyond Oracle. The episode shows that for a heavily committed AI builder, investors and lenders are pricing delivery dates, not just contract totals. It also shows who else carries that risk: lenders holding the project debt, the customer waiting for capacity, and investors in Oracle’s securities.

    A Backlog Is a Delivery Schedule, Not a Bank Balance

    Remaining performance obligations, often shortened to RPO, are revenue a company has signed contracts for but not yet recognized. Recognition happens as the service is delivered. For a cloud provider renting out AI computing, delivery means racks of GPUs running in an energized building. Oracle’s $664 billion of RPO dwarfs the $11.6 billion of cloud revenue it booked in its latest quarter. Most of that backlog therefore sits in future periods, and it cannot arrive faster than the capacity that delivers it.

    That is the mechanism behind the headline. A signed contract fixes the size of the opportunity but not its timing. When a site like Jupiter meets opposition over water and air quality, or a pipeline it needs for power hits a regulatory setback, the contract still stands. But the revenue stream shifts right on the calendar. The 3.5% share decline TradingView attributed to delay concerns came despite a backlog that grew by $209 billion in a year. For Oracle, the market is weighing build dates alongside contract totals, and it is treating the delivery schedule as the variable that determines what the backlog is worth today.

    The effect is sharpest for a company whose costs are already committed. A backlog funded out of spare cash flow can absorb a slipped date. A backlog funded with borrowed money pays interest on the gap.

    Costs Run on a Calendar; Revenue Waits for Power

    Oracle’s spending has moved far ahead of its cash generation. Capital expenditures, the money spent on buildings, servers and equipment, rose to $55.7 billion in fiscal 2026 from $21.2 billion a year earlier. The result was negative free cash flow of $23.7 billion. First-quarter fiscal 2027 capex alone was $28.5 billion, and Oracle expects full-year spending to exceed the prior year’s. The company raised $43 billion of debt and $5 billion of equity in fiscal 2026 and expects about $40 billion of additional debt and equity financing in fiscal 2027.

    Leases add a second layer. As of August 31, Oracle carried $34.6 billion of recognized operating lease liabilities plus $288 billion of additional data center lease commitments expected to begin between fiscal 2027 and fiscal 2029. Those commitments give Oracle capacity, but they also fix a schedule of obligations. The margin data already shows the strain: Oracle’s cloud and software margin percentage declined in the latest quarter as higher infrastructure expenses supported cloud growth. When a site is late, costs such as financing, early lease payments and pre-purchased equipment can arrive before the capacity produces revenue. That squeezes margin further.

    What Jupiter’s Discounted Debt Signals Down the Chain

    A loan quoted at 89 to 91 cents on the dollar means a holder selling $100 of it would receive roughly $89 to $91. Banks often underwrite large project loans intending to sell much of them on. When distribution stalls, they hold more of the risk themselves, and the discount tells future lenders what compensation the market now requires. Two separate concerns are bundled into that price. One is site-level: permits, water, air quality and a gas pipeline. The other is sponsor-level: Oracle’s leverage, which S&P flagged by downgrading the company in July to one notch above junk, the lowest rung of investment grade.

    Those risks reach beyond Oracle’s shareholders. Lenders holding the Jupiter loans are exposed to the site’s schedule. OpenAI, as the customer the expansion serves, depends on the capacity arriving when planned. Investors in Oracle’s debt are exposed to the financing needs any delay would enlarge. A Reuters Breakingviews commentary ran under the headline “Oracle freakout exposes fragile data center boom.” The evidence here, though, is one project at one company with an unusually leveraged balance sheet. Whether other builders face the same repricing is a question this episode raises but does not settle.

    The Offsets Are Real, but They Also Run on Delivery

    There is a genuine counterweight. Oracle said $75 billion of its large AI contracts involve customers prepaying for GPUs or supplying the GPUs themselves. That substantially reduces the capital Oracle must raise for AI data centers, and it shifts part of the chip bill to the parties that want the capacity. Cloud infrastructure revenue more than doubling in a quarter also shows that backlog is converting.

    Even so, customer-funded chips still need a building, power and cooling to go into. The customer funding eases the equipment side of the ledger. It does not remove the permitting and power risk that has put Jupiter under scrutiny. The balanced reading is that Oracle’s demand is well documented and its financing model is adapting. What the market is now testing is whether the build schedule can keep pace with the contracts.

    Background

    Oracle built its business on database and enterprise software and has spent recent years turning Oracle Cloud Infrastructure into a major supplier of AI computing, competing with the largest cloud providers, often called hyperscalers. Its fiscal year ends May 31. Its proxy statement filed September 25 lists Safra Catz, Clayton Magouyrk and Michael Sicilia each as a principal executive officer during fiscal 2026, reflecting a change at the top during the year.

    AI data centers pack thousands of power-hungry GPUs, the specialized chips used to train and run AI models, into buildings that need large, dependable electricity supplies and cooling. That makes power availability, water use and local permitting central to how fast capacity can be delivered. Project Jupiter in New Mexico, part of Oracle’s expansion with OpenAI, is one of the sites where those constraints now intersect with the company’s financing.

    Sources

    Source: Oracle’s AI Expansion Faces a New Test as Data Center Costs Rise (Insider Monkey), on the pricing of Project Jupiter’s $18 billion financing and Oracle’s spending, debt and backlog. Also cited: Oracle’s Stock Falls 3.5% on Data Center Delay Concerns (TradingView) and Breakingviews – COMMENTARY: Oracle freakout exposes fragile data center boom (Reuters).

    Primary sources: Oracle Corporation DEF 14A proxy statement, filed with the SEC on September 25, 2026.

  • Hut 8’s $1.07B Revolver Puts Bank Credit, Not Cash, Behind AI Grid Deposits

    Hut 8’s $1.07B Revolver Puts Bank Credit, Not Cash, Behind AI Grid Deposits

    TL;DR · 30-second read

    The Short Version

    Hut 8, a company that builds the huge, power-hungry buildings that run artificial intelligence, has lined up a $1.07 billion line of credit from 12 banks.

    The surprising part is what much of it can be used for. Before a building goes up, power companies and equipment makers want deposits. This deal lets a bank’s written promise to pay stand in for that cash, so the company’s own money is not sitting idle.

    It is a reminder that building for artificial intelligence costs serious money long before any computer is switched on.

    Hut 8 Corp. (Nasdaq, TSX: HUT) announced on September 28, 2026 that it has closed a $1.07 billion, four-year senior secured revolving credit facility provided by a syndicate of 12 lenders. J.P. Morgan was lead left arranger, bookrunner and administrative agent; Citi, Goldman Sachs and Morgan Stanley were joint lead arrangers and joint bookrunners.

    Borrowings are priced at the Secured Overnight Financing Rate (SOFR) plus 150 to 200 basis points depending on Hut 8’s debt-to-market-capitalization ratio, starting at SOFR plus 175. The facility carries a letter-of-credit sublimit of $1.07 billion intended to cover interconnection deposits and obligations to utilities and equipment vendors.

    Executive Summary

    A revolving credit facility, or revolver, works like a corporate credit card with a very high limit: the company can draw, repay and draw again for the life of the agreement. Hut 8’s new $1.07 billion revolver sits at the parent company level and is secured by company assets, and Hut 8 says it can be drawn as needed and repaid without prepayment penalties.

    The detail that sets this facility apart is its letter-of-credit sublimit, which equals the full $1.07 billion. A letter of credit is a bank’s guarantee that it will pay a counterparty if the borrower does not. Hut 8 says it will use these to satisfy collateral requirements during site development, including grid interconnection deposits and obligations to utilities and equipment vendors, reducing the need to post cash.

    The facility is designed to complement, not replace, the $7.5 billion of non-recourse, investment-grade project financing Hut 8 has already arranged for construction at its River Bend and Beacon Point AI data center campuses. The revolver covers the earlier, riskier stretch of a campus’s life; the project debt takes over once a site is further along.

    The Letter-of-Credit Line Is the Real Story

    Headlines naturally focus on the $1.07 billion total, but the operationally significant number is the letter-of-credit sublimit set at the same $1.07 billion. Before an AI data center campus can be built, a developer typically has to secure its place in a utility’s interconnection process, the queue and engineering studies that determine when and how a site connects to the grid, and reserve long-lead equipment. Hut 8’s release says both utilities and equipment vendors impose collateral requirements at this stage.

    Posting that collateral in cash means money that could fund construction instead sits in escrow for months or years. A letter of credit gives the utility or vendor the same assurance, backed by a bank, while Hut 8 keeps its cash. Because the sublimit matches the full facility, the entire commitment can in principle be deployed as guarantees rather than loans, though in a typical revolver any letters of credit issued reduce what remains available to borrow. The release does not say how much has been issued so far.

    For Hut 8, this is a capital-efficiency measure that directly targets the pre-construction phase. For the utilities and equipment makers on the other side of these obligations, it means their security comes from a syndicate of large banks rather than the balance sheet of a single developer.

    Two Layers of Debt for Two Stages of a Campus

    CFO Sean Glennan described the logic plainly: the revolver lets Hut 8 fund projects through development while it determines the optimal timing and structure for long-term, non-recourse financing as they de-risk. Non-recourse project debt is repaid from a specific project’s cash flows, and lenders generally cannot pursue the parent company if the project fails. That kind of financing is usually available only once a project has contracts, power and a construction path in place.

    The revolver fills the gap before that point. Hut 8’s $7.5 billion of fully amortizing project financing for River Bend and Beacon Point, meaning debt that is paid down to zero over its life rather than refinanced, handles construction. The parent-level facility handles the earlier stage, when a site is still acquiring interconnection rights and equipment slots. Treating development as a recurring activity with a standing funding line, rather than financing each site from scratch, is the structural shift here.

    That 12 lenders, led by four of the largest US banks, committed to a four-year corporate facility suggests they are comfortable underwriting Hut 8’s development activity on an ongoing basis. It is one facility at one company, however, and it does not by itself establish how lenders view former bitcoin miners as a group.

    A Borrowing Cost Linked to the Share Price

    The pricing grid ties Hut 8’s margin to its consolidated total debt-to-market-capitalization ratio. In practical terms, the company’s borrowing cost moves with both its debt load and its stock market value. If Hut 8 adds project debt or its shares fall, the ratio rises and the margin can step up toward SOFR plus 200 basis points; if its market value grows faster than its debt, the margin can move toward 150.

    This is worth noting because the release frames the facility as non-dilutive, meaning it raises money without issuing new shares. That is accurate for the borrowing itself, but the cost of that borrowing remains exposed to equity markets. Hut 8’s own forward-looking statements list declining equity valuations and tightening lending standards among the financing risks it faces.

    The secured structure is also a reminder of where the company stands today. Hut 8 describes an investment-grade corporate profile as something it continues to pursue. Investment-grade companies often borrow on an unsecured basis, so this facility reflects progress on access to capital rather than arrival at that goal.

    Background

    Hut 8 Corp., listed on Nasdaq and the Toronto Stock Exchange, describes itself as an energy infrastructure platform integrating power, digital infrastructure and compute for energy-intensive workloads such as AI, high-performance computing and ASIC compute, the specialized chips used in bitcoin mining. Like several companies that grew up in bitcoin mining, Hut 8 has leaned on its experience securing large amounts of power to move into AI data center development.

    Its flagship AI projects are the River Bend and Beacon Point campuses, backed by $7.5 billion of fully amortizing, non-recourse, investment-grade project financing. The new revolver adds a corporate-level funding layer for the development work that precedes that project-level debt.

    Sources

    Source: Hut 8 Expands Corporate Liquidity with $1.07 Billion Senior Secured Revolving Credit Facility, Hut 8’s September 28, 2026 announcement of a four-year revolver with a full-size letter-of-credit sublimit.

  • Riot Clears a $200M Bitcoin Loan Before Its 191 MW AI Lease Tests the Balance Sheet

    Riot Clears a $200M Bitcoin Loan Before Its 191 MW AI Lease Tests the Balance Sheet

    TL;DR · 30-second read

    The Short Version

    Riot Platforms, a company that runs huge warehouses of computers that mine bitcoin, has paid off a $200 million loan from the crypto exchange Coinbase about seven months early, with no penalty.

    To get that loan, Riot had set aside about half of its bitcoin as a guarantee, the way a house backs a mortgage. That bitcoin is no longer tied up.

    The timing is the interesting part. Riot recently signed a 20-year deal to rent a large, power-hungry part of its site to an artificial intelligence company, and getting that space ready will take money.

    Riot Platforms (Nasdaq: RIOT) repaid all outstanding principal and accrued interest on its fully drawn $200 million credit facility with Coinbase Credit on September 21 and terminated the agreement, according to an 8-K filed with the SEC on September 25. The Energy Mag reported that the repayment carried no early termination fee, that Coinbase’s commitment to lend further ended, and that its security interests in Riot’s pledged bitcoin, USDC and cash were released at the same time.

    As of June 30, Riot had pledged 5,821 bitcoin, valued at about $340.7 million, roughly 51% of its 11,380-bitcoin holdings, to secure the loan. The facility carried a fixed 6.15% annual rate and was not due until April 20, 2027.

    Executive Summary

    Riot has retired a bitcoin-backed loan that, at its fixed 6.15% rate, cost about $12.3 million a year in interest, and in doing so lifted a lender’s claim on what was, at last disclosure, about half of its bitcoin treasury. The repayment came one month after the facility’s early termination fee window closed on August 21, so Riot exited about seven months ahead of maturity at no penalty.

    The move matters beyond crypto treasury management because of what Riot is building. In August, the company announced a 20-year lease for 191 megawatts of computing capacity at its Rockdale campus with an unnamed artificial intelligence developer, which it said was expected to generate about $9.1 billion in revenue over the initial term. Riot now approaches the delivery of that capacity with no Coinbase debt outstanding and no lender holding a lien on its bitcoin under this agreement.

    What Riot has not said is how it intends to fund the AI build, or where the cash for the repayment came from. The payoff widens the company’s options; it does not by itself reveal which option Riot will take.

    The Balance Sheet Riot Carries Into a 191 MW AI Build

    A megawatt is a unit of electrical power, and data center capacity is quoted in megawatts because power, not floor space, is what limits how many AI servers a site can run. A 191 MW lease is utility-scale: Riot has committed to deliver computing capacity on that scale to a single AI tenant for 20 years, against expected revenue of about $9.1 billion over the initial term. Riot has not disclosed what readying that capacity will cost or how it will be financed, but converting mining infrastructure into AI-grade data halls typically requires substantial capital before lease revenue arrives.

    That is the context in which the Coinbase payoff reads as more than housekeeping. While the facility was in place, Riot carried a $200 million liability due April 2027, roughly $12.3 million a year in interest, and a lien over pledged assets that stood at 5,821 bitcoin on June 30. Pledged coins cannot be freely sold or offered to a different lender. With the security interests released, those assets are back under Riot’s unrestricted control, where they can be held, sold, or used to back new financing sized for a data center build rather than a crypto credit line.

    The test the headline refers to is still ahead: delivering 191 MW on schedule for a single counterparty is a capital and execution exercise, and lenders, the AI tenant and shareholders will all be watching how Riot funds it. The repayment does not answer that question. It means Riot enters it without this particular claim on its largest liquid asset.

    Why Repay 6.15% Debt Seven Months Early

    The timing lines up closely with the agreement’s own terms. The early termination fee applied only through the four-month anniversary of the facility’s original April 21, 2026, maturity date, a window that closed on August 21. Riot repaid on September 21, one month later, and paid nothing beyond principal and accrued interest. At a 6.15% fixed rate on $200 million, the roughly seven months of remaining term would have carried on the order of $7 million in interest, a simple arithmetic estimate rather than a disclosed figure.

    The facility’s history shows why Riot may have been content to close it. When first opened in April 2025 as a $100 million line, it priced at the greater of the federal funds target range’s upper limit or 3.25%, plus 4.5 percentage points, implying a minimum of 7.75%. Doubling it the following month cost a one-time $1 million fee. The April 2026 amendment fixed the rate at 6.15% and extended maturity, but it remained a loan secured by a volatile asset from a crypto-native lender.

    There is a trade-off. Coinbase’s commitment to provide further loans also ended, so Riot gave up an established borrowing relationship, and the cash used to repay is no longer available for other purposes. Whether that is a sound exchange depends on what the money would otherwise have earned and on what financing Riot lines up next.

    Bitcoin Collateral Is Cheap Until the Price Moves

    On June 30 figures, the $200 million loan equaled about 59% of the reported $340.7 million value of the pledged coins. Loans secured by an asset as volatile as bitcoin generally leave the borrower exposed to its price: a sharp decline can erode the collateral cushion and pressure the borrower to add assets or reduce debt. For a company whose strategy increasingly rests on long-dated data center contracts, removing a financing arrangement tied to bitcoin’s price reduces one source of balance-sheet volatility.

    That observation should not be stretched into a verdict on the wider sector. This is one company clearing one facility, and it says nothing definitive about how other miners pursuing AI hosting will finance themselves. What it does show is a practical point for anyone evaluating a miner’s AI ambitions: the size of its bitcoin treasury matters less than how much of that treasury is already spoken for.

    Background

    Bitcoin miners run large fleets of specialised computers that consume enormous amounts of electricity, which means they have spent years securing sites, power connections and cooling at scale. Those same assets are what AI developers need to run their servers, and several miners have begun leasing capacity to AI tenants alongside, or instead of, mining. Many miners also hold bitcoin on their balance sheets and have used it as collateral for loans.

    Riot Platforms is one of the larger US-listed miners. It took out a bitcoin-backed loan from Coinbase Credit in April 2025, expanded it to $200 million a month later, and refinanced it to a fixed 6.15% rate in April 2026. In August 2026 it announced its 191 MW, 20-year AI lease at Rockdale, marking a significant step into data center leasing.

    Sources

    Source: Riot Repays Coinbase Loan, Ends $200 Million Bitcoin-Backed Facility — report on Riot’s early repayment and termination of its Coinbase credit facility and the release of pledged collateral.

    Primary sources: Riot Platforms Form 8-K filed September 25, 2026.

  • WhiteFiber’s 83 km, 136 Tbps Link Bets Fiber Can Pool AI Power Across Sites

    WhiteFiber’s 83 km, 136 Tbps Link Bets Fiber Can Pool AI Power Across Sites

    TL;DR · 30-second read

    The Short Version

    WhiteFiber, a company that rents out computing power for artificial intelligence, is now selling a way to make two data centers about 50 miles apart work as one giant computer. The two buildings are joined by a bundle of fiber-optic cables.

    Why it matters: the computers that train artificial intelligence are often limited by how much electricity a single building can get. WhiteFiber’s idea is to combine spare electricity at several existing buildings instead of building one enormous new site.

    The top speed is still being tested, and no customers have been named yet.

    StorageReview reported on September 25 that WhiteFiber (WYFI) has made WhiteFiber Continuum commercially available. Continuum is the productized version of Project Redwood, a two-site GPU supercluster the company first detailed in July. It links two HITRUST-certified QTS data centers 83 kilometers apart over 12 Zayo dark fiber strands, with DriveNets supplying the Ethernet network fabric and WEKA’s NeuralMesh supplying shared storage and memory.

    WhiteFiber rates the architecture at 136 Tbps of aggregate bandwidth, which it describes as 170 wavelength channels of 800 gigabits per second each, with 0.9 milliseconds of guaranteed round-trip latency. The July research result measured 111.2 Tbps on part of the fiber spectrum. Full-spectrum testing to confirm the guaranteed commercial specifications is underway, and the company has submitted patent applications for the implementation.

    Executive Summary

    WhiteFiber is selling a way to make GPUs in two separate buildings behave like a single cluster. GPUs are the chips that train and run AI models, and large training jobs have normally been confined to one building because the chips have to exchange data constantly. Continuum stretches that boundary to 83 kilometers by carrying the traffic over leased, unlit fiber that WhiteFiber lights itself, which the industry calls dark fiber.

    The launch matters because of what WhiteFiber says the design is for. The company pitches Continuum as a way to get past the power and space ceiling of a single campus. Enterprises would pool GPUs across sites, and telecom and metro facilities with spare power could contribute capacity to one logical cluster. If that works at the stated performance, the useful unit of AI capacity is no longer one building with enough electricity. It is any set of buildings connected by enough fiber.

    The headline number still needs confirmation. The 136 Tbps figure is a design rating that depends on additional wavelengths coming online and on full-spectrum testing now in progress. The figure WhiteFiber has actually measured is 111.2 Tbps, taken in July on part of the spectrum.

    Fiber as a Way Around the Single-Site Power Ceiling

    AI capacity is increasingly set by power: how many megawatts a single building or campus can draw. A megawatt is a million watts, and a large GPU campus uses many of them. A company that has filled its power allocation at one site usually faces a choice between waiting for more grid capacity and building a new campus from scratch. WhiteFiber offers a third option. Continuum lets enterprises pool GPUs across sites and add overflow training and inference capacity without a separate greenfield deployment. It also lets telecom and metro facilities with spare power and fiber join the same logical cluster.

    The design choice that makes this plausible is what crosses the link. According to WhiteFiber’s product page, both sites carry live traffic at the same time and only model gradients cross the 83 km connection. Gradients are the incremental updates each group of GPUs computes during training. The bulk of the work and data stays local to each site, and the two halves synchronize their learning over the fiber. A single job can span both sites, or the cluster can be split into independent workloads that burst from one site to the other as demand shifts.

    Fiber does not create power. It lets power that already exists in two places be used by one workload. The groups most affected are operators whose sites have stranded capacity, such as telco central offices and metro facilities with power to spare but too little of it to host a full cluster. Fiber owners like Zayo, whose dark fiber becomes part of the compute fabric, stand to gain as well. Enterprises that have maxed out a campus are the third group. Whether this changes where AI capacity gets built will depend on how much of a real training job’s performance survives the split, and WhiteFiber has not yet published that comparison.

    What 136 Tbps Establishes, and What It Does Not Yet

    The internal arithmetic holds together. 170 wavelength channels at 800 gigabits per second each come to 136 terabits per second. Dense wavelength-division multiplexing sends many separate colors of light down the same fiber strands, each carrying its own data stream. Adding wavelengths is how the design scales from the 111.2 Tbps measured in July to the 136 Tbps now rated.

    The status of the number matters to buyers. WhiteFiber says the design reaches 136 Tbps after additional wavelengths come online, and that full-spectrum testing is underway to confirm the guaranteed commercial specifications. Enterprises evaluating Continuum should treat 111.2 Tbps as the demonstrated figure and 136 Tbps as the contracted target until test results are released.

    The latency figure tells a different story. WhiteFiber guarantees 0.9 milliseconds round trip, which it says is within 8% of the physical limit for light in fiber over 83 km. There is little left to optimize, so distance itself is the fixed cost of the architecture. Longer links between sites add latency in proportion to distance. The 83 km result shows the approach works at metro-to-regional range, and it says nothing yet about sites much farther apart.

    Resilience and Data Boundaries as the Enterprise Pitch

    Continuum is also sold as a resilience and compliance product. WhiteFiber says neither site is a single point of failure, so training runs are designed to continue if one location goes down. For regulated industries, it says sensitive workloads can stay within their originating jurisdiction while failover and pooled compute happen across locations. HITRUST is a security and privacy certification framework widely used in healthcare, and the certification of both QTS facilities is aimed at that audience.

    CEO Sam Tabar framed the product around both themes, saying it is for enterprises that need AI compute to “hold up under compliance requirements, and not break when a single site has a problem.” These are credible goals for a two-site design. They are also claims about behavior under failure, and no failure test results have been released.

    A Partner Stack With a Patent Claim on Top

    Continuum is assembled from named partners. QTS provides the facilities, Zayo the fiber, DriveNets the Ethernet-based network fabric, and WEKA the shared storage and memory layer. WEKA CEO Liran Zvibel said NeuralMesh lets GPUs that sit many kilometers apart “operate as if the data were local.” DriveNets said its fabric is built to maximize GPU utilization across scale-across superclusters. Using Ethernet rather than proprietary interconnects for the inter-site fabric keeps the design within a widely supported networking standard.

    WhiteFiber calls Continuum the first commercially available distributed GPU supercluster architecture and has submitted patent applications for the implementation. Submitted applications are not granted patents. Because most of the stack comes from third-party vendors, WhiteFiber’s lasting advantage will likely come from its integration and operating know-how, and from customer contracts once it signs them.

    Background

    WhiteFiber, which trades under the ticker WYFI, provides AI infrastructure. In July it detailed Project Redwood, a research effort that connected GPUs in two data centers 83 km apart and measured 111.2 Tbps across the link. Its partners are well-established infrastructure suppliers. QTS operates data centers, Zayo owns and leases fiber networks, DriveNets builds Ethernet networking for AI clusters, and WEKA makes high-performance data storage software.

    Large AI models have traditionally been trained inside a single building. The GPUs doing the work exchange data constantly, and longer distances add delay. As single campuses run up against limits on how much power and space they can secure, operators have looked for ways to spread one workload across several sites. Continuum is WhiteFiber’s commercial answer to that problem.

    Sources

    Source: WhiteFiber Continuum Goes Commercial, Scaling Its 83 km Two-Site GPU Supercluster Design to 136 Tbps (StorageReview.com): WhiteFiber makes its two-site, 83 km GPU supercluster architecture commercially available with a rated 136 Tbps of inter-site bandwidth.

  • Cipher’s $5.2B Barber Lake Extension Puts a Price on an AI Site’s Second Decade

    Cipher’s $5.2B Barber Lake Extension Puts a Price on an AI Site’s Second Decade

    TL;DR · 30-second read

    The Short Version

    Cipher Digital, a company that started out mining bitcoin, has signed a second tenant for its giant computer warehouse in Colorado City, Texas. That tenant will not move in for about ten years.

    The tenant is described only as a leading artificial intelligence company. It has promised roughly $5.2 billion in rent for that later decade, so the site now has 20 years of signed leases instead of 10.

    Why it matters: an artificial intelligence company is locking up a building a decade in advance. The catch: Cipher pays the first $359.3 million of any building cost overruns.

    Cipher Digital (Nasdaq: CIFR) announced on September 25, 2026 that it has extended the contracted term of its Barber Lake data center in Colorado City, Texas, from 10 years to 20. In an 8-K filed the same day, Cipher disclosed that its subsidiary amended its existing lease with Fluidstack, an AI cloud provider. Alongside the amendment, Cipher signed a binding commitment with an unnamed “leading AI lab” to lease the facility for a further ten years once the Fluidstack term ends. Cipher expects the new lease to add about $5.2 billion of contracted revenue, raising the facility’s total from $3.8 billion to more than $9 billion.

    The amendment also set a revised phased delivery schedule, with data halls expected between the fourth quarter of 2026 and the first quarter of 2027. It also established a cost-sharing framework under which Cipher absorbs the first $359.3 million of costs above the original budget. 24/7 Wall St. reported that Cipher’s shares rose 4% on the news, with TeraWulf and Applied Digital each gaining 2%.

    Executive Summary

    Cipher has added a second decade of contracted rent to a facility whose first data halls have not yet been handed over. The new ten-year lease will be a separate contract with a different tenant, the unnamed AI lab. According to Cipher’s press release, it will carry economic terms “substantially consistent” with the Fluidstack lease. The roughly $5.2 billion it adds exceeds the $3.8 billion the original lease carried, and Cipher did not need a new site to get it.

    This matters because the largest unknown in owning a single-tenant data center is usually what the building is worth after the first lease ends. At Barber Lake, that residual value is now contracted rather than assumed. There is a cost attached. Cipher took on tenant-requested design changes, a revised delivery schedule, and the first $359.3 million of any overruns above the initial budget.

    For investors and industry participants, the deal offers a concrete benchmark: what a committed second term can be worth on an AI campus already under construction. It also raises fair questions. The $9 billion headline is a nominal sum stretched over two decades, the second tenant is unnamed, and Cipher has not disclosed the construction budget.

    The Second Decade Now Has a Price Tag

    Much of the risk in owning a data center built for one customer sits in what happens when that customer’s lease ends. A facility laid out around one tenant’s hardware, power density and floor plan may need costly refits before a new occupant moves in, or may sit partly empty while one is found. For that reason, lenders and investors usually treat residual value, meaning what the asset is worth after the contracted term, conservatively. The concern is sharper for AI facilities, where chip generations and rack designs change quickly.

    Cipher’s deal turns that uncertainty into a contract. The AI lab’s commitment is binding and covers the ten years after Fluidstack’s term. Cipher’s 8-K says the new lease will carry economic terms substantially consistent with the existing one. The result is roughly $5.2 billion of revenue attached to years 11 through 20 of a facility whose first rent is not expected until the fourth quarter of 2026. A tenant is effectively reserving capacity about a decade in advance. One reasonable reading is that the lab places a high value on securing an identified, already-developing site now rather than competing for one later.

    This is the mechanism behind the headline. A company like Cipher, which moved from bitcoin mining to leasing capacity to AI tenants, has two broad ways to grow contracted value. It can build more energized capacity, or it can lengthen the contracted life of capacity it is already building. This deal uses the second approach, and Cipher has not said it involves additional capacity. One transaction does not establish an industry pattern. It does show that, at this site, contract duration was worth more in nominal terms than the entire original lease.

    Why “Past $9 Billion” Overstates Today’s Value

    Contracted revenue is a sum of future rent, not a present value. The $5.2 billion only begins once Fluidstack’s ten-year term ends. With first rent expected in late 2026, that is roughly a decade away, and it then runs for another ten years. At any realistic cost of capital, a dollar of rent received in the 2040s is worth considerably less than a dollar today. The $9 billion total is accurate as a contractual figure. It should not be read as $9 billion of near-term cash flow.

    The figure also carries two decades of counterparty exposure. Cipher’s return depends on the AI lab still being willing and able to pay ten to twenty years from now, in a sector where demand, technology and tenant finances can change substantially over that span. Cipher has not disclosed any guarantee, deposit or other credit support tied to the new lease.

    There is also an arithmetic question. If the terms are substantially consistent with the first lease, it is not obvious why the second decade adds about $5.2 billion while the first carries $3.8 billion, roughly 37% more. Rent escalators, the redesigned facility, or the cost-recovery rent could each contribute, but Cipher has not broken the figure down. Chief Executive Tyler Page said the deal underscores “the durability of demand for hyperscale computing capacity.” That claim is well supported for this campus. As evidence about the wider market, it remains a single data point.

    The $359.3 Million Cost of Securing the Term

    The amendment came with tenant-modification change orders, which are redesigns requested by the occupant, and a revised phased delivery schedule. Individual data halls are now expected between the fourth quarter of 2026 and the first quarter of 2027, and rent starts hall by hall as each is delivered. Cipher says it remains on track with the revised schedule.

    The cost framework is where the trade-off becomes visible. Cipher bears the first $359.3 million of costs above the initial budget. Beyond that, the tenant reimburses 50% over the full twenty-year term as additional rent, calculated to give Cipher a contracted rate of return. Cipher therefore carries the first tranche in full and, by implication, the other half of anything above it. Part of the price of the second decade is Cipher taking on construction-overrun risk today. Because Cipher has not disclosed the initial budget, outsiders cannot tell whether $359.3 million is a remote cushion or a likely spend.

    For the people who design and build these facilities, the operational lesson is clear. AI tenants’ requirements are still changing while construction is under way. At Barber Lake, the landlord that accommodated those changes, and absorbed the first layer of their cost, came away with a longer contract.

    A Benchmark for Converted Miners

    Cipher, TeraWulf and Applied Digital each built businesses around cryptocurrency mining or hosting before turning to AI and high-performance computing (HPC) tenants. HPC is the dense, power-hungry computing used to train and run large AI models. Much of these companies’ appeal to AI customers has rested on sites with large grid connections, which can take years to secure.

    Barber Lake gives the group a reference point. The commercial question shifts from how many megawatts a company can energize to how many years of rent it can lock in on what it energizes. Investors may reasonably ask peers whether their first-term leases carry renewal commitments or follow-on tenants. Neither TeraWulf nor Applied Digital disclosed a comparable arrangement alongside Cipher’s announcement, so their gains reflect read-across rather than new contracts of their own.

    Background

    Cipher Digital, formerly Cipher Mining, built its business on bitcoin mining before repositioning as a developer and operator of industrial-scale data centers for high-performance computing tenants. Barber Lake in Colorado City, Texas is its flagship campus of this kind. The original lease with Fluidstack, an AI cloud provider, covered ten years and $3.8 billion of contracted revenue.

    Several former miners, including TeraWulf and Applied Digital, have taken a similar path. They lease sites with large existing or planned power connections to AI customers who need capacity faster than new grid connections can typically be secured. Investors increasingly judge these companies by the size, duration and quality of their contracted leases.

    Sources

    Source: Cipher Digital Jumps 4% as Barber Lake Lease Extension Lifts Contracted Revenue Past $9B; TeraWulf and Applied Digital Gain 2% (24/7 Wall St., via Yahoo Finance), market coverage of Cipher’s Barber Lake lease extension and peer share moves.

    Primary sources: Cipher Digital Inc. Form 8-K filed September 25, 2026; Cipher Digital press release (Exhibit 99.1): Cipher Digital Expands Barber Lake Lease Term to 20 Years, Increasing Revenue to Over $9 Billion; Applied Digital Form DEF 14A filed September 25, 2026.