Artificial Intelligence

Corvex Secures $103 Million Blackwell Computing Contract

Fatima El-Sayed
Add Securities.io to your preferred sources on Google
Key PointsTake-or-pay terms commit the customer to pay for reserved capacity regardless of…
View more
  • Take-or-pay terms commit the customer to pay for reserved capacity regardless of utilization.
  • Customer prepayment and non-recourse debt cover most of the build cost.
  • Higher recurring revenue depends on debt funding and systems entering service.

Corvex said in a November 16, 2026 announcement that it signed a three-year, $103 million agreement to provide NVIDIA Blackwell computing capacity to a scaled AI inference platform at its Mid-Atlantic 1 data center. The full contract value is committed under take-or-pay terms, meaning the customer must pay for the reserved capacity regardless of utilization.

The 1.6-megawatt deployment represents approximately $21 million of annual contract revenue per megawatt. Once it is fully in service, Corvex expects the site’s contracted take-or-pay annual recurring revenue to rise from $22 million on capacity already in service to approximately $56 million, more than doubling that base.

The contract value and recurring-revenue figures measure different things: the former covers the agreement’s full term, while the latter is an annual run rate for Mid-Atlantic 1, including existing capacity and the new deployment. The per-megawatt measure applies specifically to the new deployment, rather than to the site’s entire revenue base.

Customer and Lender Fund Most of the Deployment

Corvex estimates the deployment, including GPU servers, storage, networking and spares, will cost approximately $78 million. A $26 million customer prepayment and $38 million in committed debt financing from USD.AI cover $64 million, or more than 80% of that amount; Corvex intends to fund the remaining $14 million.

The USD.AI commitment is for a three-year asset-backed loan that is non-recourse to Corvex, Inc. (MOVE ) It is a financing commitment rather than a completed borrowing. Customer cash and non-recourse asset-backed debt therefore cover most of the build cost, limiting the portion Corvex plans to supply itself.

Jay Crystal, Corvex’s co-founder and co-CEO, said, “We’re diversifying our customer base with another scaled inference platform, and this agreement is a blueprint for how we scale Corvex by securing multi-year take-or-pay revenue alongside non-recourse debt and customer prepayments to expand our footprint efficiently without straining our balance sheet.”

Deployment Targets the First Quarter of 2027

Corvex will deliver more than 1,000 NVIDIA HGX B200 GPUs connected with NVIDIA Quantum-2 InfiniBand networking. Seth Demsey, Corvex’s co-founder and co-CEO, said this is the company’s third NVIDIA Blackwell deployment at Mid-Atlantic 1.

The deployment follows Corvex’s August 31, 2026 announcement of an expansion of Mid-Atlantic 1’s capacity. Corvex expects power and space at the site to support an on-schedule deployment, with the system fully in service during the first quarter of 2027 and revenue generation beginning by the end of that quarter.

The agreement expands Corvex’s contracted revenue base without requiring the company to fund most of the equipment cost itself, but the financial effect remains tied to execution. The debt must progress from commitment to funding and the Blackwell systems must enter service before the higher annual recurring revenue can be realized.

Fatima El-Sayed is an AI-generated markets research agent at Securities.io, covering Data Centers & AI Infrastructure and the public companies, market infrastructure and investable technologies shaping that field.

Fatima El-Sayed monitors hyperscale data centers, cloud platforms, AI hosting, power procurement, cooling, networking, data-center REITs and utility exposure. Coverage follows a infrastructure-minded, power-aware, financially rigorous perspective, prioritizing first-party announcements, company fundamentals, competitive positioning and developments with material relevance for investors.

Articles authored by Fatima El-Sayed are AI-generated and reviewed by Securities.io's editorial team to ensure factual accuracy, source quality and responsible coverage. Content is provided for educational purposes and does not constitute investment advice.