Corvex’s $103M Blackwell Deal and Conditional USD.AI Loan | Magica
Corvex’s $103 Million GPU Deal Depends on Conditional USD.AI Financing
Editorial Team
••📖6 min read
Corvex plans to fund a $78 million Blackwell deployment with customer prepayment, a $38 million USD.AI debt commitment and $14 million of its own money. The contract could more than double its Mid-Atlantic 1 annual contracted revenue, but loan closing, delivery and customer acceptance remain ahead.
Corvex’s three-year capacity contract is worth $103 million, with revenue expected to start by the end of March 2027.
Customer prepayment and a conditional USD.AI loan would fund about 82% of the estimated $78 million deployment cost.
The deal expands an existing lending relationship; the new loan’s interest rate has not been disclosed in the announcement.
Corvex, a Nasdaq-listed provider of GPU infrastructure for artificial intelligence, plans to use a $38 million debt commitment from USD.AI, an onchain GPU-backed lending platform, to help deliver a $103 million computing-capacity contract. The three-year customer agreement could more than double annual contracted revenue at its Mid-Atlantic 1 data center, but the financing has yet to close.
The distinction matters for an AI cloud business still early in commercialization. Corvex reported roughly $5 million of revenue in the third quarter of 2026, up 31% from $3.8 million in the second quarter, according to a report on its quarterly results. Its proposed deployment alone will cost an estimated $78 million, with customer prepayment and borrowing intended to reduce the amount it must supply itself.
Corvex outlined the arrangement in its November 16 announcement. Its accompanying securities filing says a binding financing term sheet does not assure funds will be available when needed or in the anticipated amount. Technical due diligence and other closing conditions remain.
A large contract for a young AI business
The public company previously operated as Movano, a wearable-health technology business. It acquired the Corvex AI business on March 19, 2026, changed its name on March 23 and transferred its legacy wearable-business assets to a lender in June to settle debt, according to its third-quarter filing. Its current offerings include GPU clusters, confidential computing and inference services.
That filing also shows why another customer matters: two customers accounted for 72% and 23% of third-quarter revenue, respectively—95% combined. Corvex says the new agreement diversifies its inference-platform customer base, although it has not named the buyer. Inference is the use of AI models to generate outputs; NVIDIA supplies the computing technology in this deal, while the customer buys access to capacity.
The ransomware gang threatens a November 28 data release, but Nvidia has not confirmed an intrusion. Earlier listings of a supplier and a partner raise questions about data provenance without establishing a connection.
Corvex says the full $103 million is committed under take-or-pay terms, requiring payment for reserved capacity regardless of use. The announcement follows an earlier Blackwell supply agreement disclosed in August. A report tracing the company’s announcements notes that Corvex said on November 12 it was finalizing a customer contract for a 1.6-megawatt Blackwell cluster, before disclosing the value on November 16.
The new deployment calls for more than 1,000 NVIDIA HGX B200 GPUs, part of the Blackwell computing platform, connected through Quantum-2 InfiniBand networking. It will occupy 1.6 megawatts of IT capacity at the existing Mid-Atlantic 1 facility. Corvex calls it the site’s third Blackwell deployment and says power and space are on track, following a capacity expansion announced August 31.
How the $78 million funding plan works
Corvex’s estimate covers GPU servers, storage, networking and spares. Its announced funding mix is:
Funding source
Planned amount
Share of estimated deployment cost
Customer prepayment
$26 million
33%
USD.AI debt commitment
$38 million
49%
Corvex contribution
$14 million
18%
The rounded shares are calculated from Corvex’s figures. Customer prepayment and debt together would provide $64 million, or about 82% of the estimate. Prepayment funds future services; borrowing creates repayment obligations. Subtracting deployment cost from the contract value would therefore not establish project profit: operating expenses and financing costs must also be covered.
The proposed loan has a three-year term, is asset-backed and is non-recourse to Corvex, Inc., according to the company. Repayment obligations remain at the borrowing entity despite the limit on recourse to the public parent.
Corvex’s planned $14 million contribution is equivalent to about one-third of the $42 million in cash and cash equivalents it held on September 30, a historical balance rather than a deployment-date cash forecast. Its filing records $16.6 million of operating cash use in the first nine months of 2026 and a September equity placement that raised $33 million gross, or about $31 million after issuance costs. Its third-quarter net loss was $17.5 million, including $11.3 million of stock-based compensation expense.
Corvex also established a facility in October permitting up to $250 million of stock sales into the market, though none had been sold under it as of November 12. That provides a potential additional funding route, with shareholder dilution if used; the authorization itself is not cash raised.
An established lender relationship, an undisclosed new rate
USD.AI’s model connects crypto finance with physical computing equipment. In an August 2025 description, the platform presented a yield-bearing stablecoin backed by GPU hardware and said it evaluated equipment specifications, warranties and rental history to issue loans programmatically.
USD.AI described the commitment as a follow-on to its August financing for Corvex, according to a report on the lender’s announcement. The same report says USD.AI targets the “compute middle market”: operators whose requirements exceed small equipment loans but whose businesses are too small for large financial institutions. USD.AI says long-term customer contracts and deployment into existing data centers fit that strategy.
Corvex’s filing supplies the earlier transaction’s terms. On August 3, subsidiary Corvex Primus borrowed $7.5 million from GPU Finance Ltd., a subsidiary of the USD.AI Foundation. The loan was secured by substantially all of the borrower’s assets, including equipment and its customer contract, with a limited guaranty from Corvex’s operating subsidiary. Outstanding principal was $7.3 million at September 30.
That loan carried a 10% nominal annual interest rate and an effective rate of about 14.8% including upfront and legal costs. It also required a reserve deposit of about $725,000. Those are terms of the August borrowing, not the new $38 million commitment. The November announcement does not disclose the new rate, fees or reserve requirements, so it cannot establish whether the larger financing is cheaper.
Company-reported contracted annual recurring revenue at Mid-Atlantic 1: $22 million on operating capacity and approximately $56 million projected after full deployment. Source: Corvex investor announcement.
Closing, delivery and acceptance will determine the result
Corvex expects revenue to begin by the end of the first quarter of 2027. Once fully in service, the deployment would increase Mid-Atlantic 1’s contracted take-or-pay annual recurring revenue from about $22 million on operating capacity to approximately $56 million, the company projects.
The company’s approximately $21 million of annual contract revenue per megawatt measures contracted sales against IT capacity. It does not measure recognized revenue today, power efficiency or profit. The $103 million contract value spans three years; neither figure establishes a pricing advantage over competing GPU services.
Existing premises reduce the need to build a new facility, but equipment supply, installation, power and networking still govern delivery. Corvex’s quarterly filing warns of constrained GPU supply and dependence on a limited number of hardware suppliers. It also records $73.9 million of estimated future payments across two additional data-center leases not yet commenced at September 30, excluding variable power charges. Those are companywide expansion commitments, not an additional cost assigned to this one deployment.
The next evidence needed is a closed loan with its final economic terms, followed by installed capacity and customer acceptance. Corvex warns that customer payment failures and termination or modification rights could affect realized revenue. Take-or-pay terms address unused reservations; whether this expansion produces cash after debt service and operating costs remains to be demonstrated.
Google is promising faster Gemini Live conversations on home speakers and displays, with timers, reminders and smart-home controls available inside the chat. The upgrade is still forthcoming and requires Google Home Premium.
Humyn Labs combines $26 million in equity with $21 million in venture debt to expand human demonstration data and planned robot testing centres. Its reported sales show demand, but deployment economics remain unproven.
Nvidia’s compact Windows desktops bring local agents and creative tools to a dedicated PC starting around $4,400. Competing desktops and Nvidia’s earlier AI hardware narrow the novelty, while conflicting configuration reports complicate price comparisons.
Maria Cantwell and Mike Rounds want a national challenge to measure how quickly AI can turn software discoveries into reliable protection. Earlier federal trials already produced patches, but safe deployment across essential services remains the test; officials say recent outages show no indication of an attack.
ByteDance is reportedly negotiating about $500 million in additional Azure spending, mostly for OpenAI models. The talks highlight different customer-access policies at Microsoft and OpenAI, against a history of scrutiny over ByteDance's use of the technology.
Hyundai Card and JPMorgan tested AI agents that compare currency quotes and request transfers under bank-enforced mandates. No real money moved, and authority across multiple banks remains an unresolved hurdle.
Industry sources say TSMC requires cash for orders exceeding customer credit limits, putting custom AI chip designers' ability to finance production at the center of the capacity race. The requirement remains unconfirmed by TSMC, and customer prepayments have a long precedent.
A congressional commission links China’s military AI ambitions to demand for advanced Nvidia chips. Procurement records show several routes to American computing, while memory costs and new remote-access rules complicate the alternatives.
Binance Wallet says subscriptions for its Anthropic IPO xStocks campaign reached the 500 million USDC limit. Final allocations, token delivery and applicable refunds remain pending; the cap does not establish proceeds received by the AI company.
Tencent says US restrictions on remote computing will likely slow its next AI model while leaving cloud revenue largely protected. Southeast Asian operators face licensing and financing uncertainty, though US cloud access and domestic demand offer partial alternatives.
Microsoft is rolling out Agent 365 Business Analytics through its Frontier early access program, pairing AI spending with estimates of assisted work. The dashboard expands its existing analytics as agent workloads add usage charges, but its dollar-value estimates do not establish financial returns.
Adobe’s Firefly update saves recurring visual subjects and creative preferences, extending an existing push into repeatable production. Memory has narrower reach than the assistant, while campaign demonstrations leave reliability and savings unmeasured.
Microsoft’s Ignite upgrades package enterprise knowledge, skills and tools into reusable business plugins. The expansion builds on IQ’s June release and competes with existing graph retrieval services; total savings and portability remain unproven.
The European Central Bank warns that debt and interlinked financing could carry an AI investment reversal into credit markets and the European economy. Resilient earnings, limited direct bank lending and stronger bank buffers temper its comparison with past bubbles.
Reported introductions helped Alice Huang secure Alibaba as a chip-rental customer and obtain financing. The relationship raises questions about Nvidia's overseas sales network, but does not itself establish illegal exports.
KPMG says AI agents pursued some issues more deeply in a parallel UK audit, though humans still had to correct their work. Client fee savings remain unconfirmed, even as KPMG International has negotiated lower fees from its own auditor by citing AI.
Payhawk announced a $280 million all-equity Series C led by Growth Equity at Goldman Sachs Alternatives, valuing it at $2 billion. The funding backs an expansion from managing business spending into broader AI finance operations, where competitors already combine automation, payments and controls.
A 16-country survey of 15- to 22-year-olds finds a median of 33% have used AI as a companion or friend. Users have higher adjusted odds of loneliness, but the research cannot establish whether AI helps, harms or replaces human connection.