The market for artificial intelligence infrastructure is moving through a phase in which access to GPUs alone is no longer enough to differentiate a provider. NVIDIA accelerators remain the essential currency of the AI infrastructure market, but customers increasingly evaluate the systems surrounding those GPUs: power availability, cooling, networking, storage, cluster reliability, deployment speed, security and the software required to keep expensive accelerators continuously productive.

That is the environment in which Corvex, Inc., a relatively new AI infrastructure company focused on GPU-accelerated computing for AI training and inference, is attempting to establish itself. The operating company was founded in 2024 and commercially launched in 2025. In March 2026, Corvex completed an all-stock merger with Movano, providing the combined company with a Nasdaq listing under the ticker MOVE. Prior to the merger, Corvex and Movano raised $40.2 million to support expansion of the AI infrastructure business.

What makes Corvex interesting is not its current scale. Measured against CoreWeave, Nebius, Lambda, Crusoe or Nscale, Corvex remains small. The more relevant question is whether its engineering model can bring new generations of NVIDIA infrastructure online quickly inside existing data centers while adding security capabilities attractive to AI developers, regulated industries and government customers.

That positioning was reinforced by an August 4 announcement involving NVIDIA Blackwell infrastructure.

The Blackwell Agreement

Corvex's August 4 announcement was not a new purchasing agreement with NVIDIA. The company said it had signed a multi-year agreement with an undisclosed “leading AI company” to provide NVIDIA Blackwell GPU clusters connected through NVIDIA Quantum-2 InfiniBand. Corvex separately participates in the NVIDIA Cloud Partner program. Seth Demsey, Co-Founder and Co-Chief Executive Officer of Corvex, Inc, said:

The customer agreement expanded an earlier commitment and includes dedicated high-speed storage and CPUs in addition to GPUs. Corvex initially delivered capacity during the first quarter and continued deployment through the second and third quarters. By its Aug. 14 earnings update, the company said the multi-year Blackwell agreement had been fully delivered. Corvex reported approximately $22 million in contracted annualized recurring revenue from compute that was live and accepted by customers. The expansion was financed through debt, customer prepayments and cash on hand rather than additional equity issuance.

But the most noteworthy aspect of the project may be the deployment itself.

Corvex installed high-density, liquid-cooled NVIDIA HGX B200 systems inside an existing air-cooled data center and says it commissioned the capacity approximately two weeks after the equipment arrived. Instead of rebuilding the facility around a central liquid-cooling system, Corvex worked with Lenovo to use Lenovo Neptune liquid-to-air cooling technology. Liquid removes heat from the servers and transfers it to the existing air-cooled facility infrastructure.

The cluster uses Lenovo ThinkSystem systems equipped with NVIDIA HGX B200 GPUs, NVIDIA Quantum-2 QM9700 InfiniBand for GPU traffic, NVIDIA Spectrum SN5600 switches for storage networking and SN2201 switches for management traffic. Corvex says the design allowed it to place high-density Blackwell infrastructure into the existing facility without a conventional facility-wide liquid-cooling conversion. Lenovo, in a case study of the deployment, contrasts the approximately two-week commissioning period with what it describes as typical data center upgrade timelines of seven to 12 months or more.

That has significance beyond a single cluster. Power availability and suitable data center capacity increasingly constrain GPU deployment. If the approach proves repeatable, liquid-to-air cooling could allow some existing air-cooled facilities with sufficient power and other supporting infrastructure to accommodate higher-density AI systems without first undergoing a full central liquid-cooling conversion.

For Corvex, that creates a straightforward time-to-revenue strategy: identify available powered capacity, adapt the thermal architecture, install current-generation accelerators and place contracted workloads on the infrastructure as quickly as possible.

Corvex Is Building More Than a GPU Rental Business

Corvex describes its core products as AI Factories, GPU clusters, confidential computing and Token Factory, an inference service that entered closed alpha during 2026. Its infrastructure can be delivered as bare metal or managed Kubernetes and in single-tenant, multi-tenant or on-premises configurations.

Its AI Factory concept integrates accelerators, networking, storage, power, cooling and systems software rather than treating GPUs as isolated cloud instances. Current supported hardware includes NVIDIA H200, B200 and B300 systems.

The more unusual part of the Corvex offering is confidential computing.

Corvex has deployed confidential computing on NVIDIA HGX B200 systems using NVIDIA GPU Confidential Computing, Intel Trust Domain Extensions and remote attestation. The architecture is intended to protect information not only while stored or transmitted but while actively being processed. Corvex’s Secure Model Weights technology is designed so that encryption keys can be released only after the CPU and GPUs have passed hardware attestation. Model weights can then be decrypted inside protected GPU memory instead of becoming visible to the cloud infrastructure operator.

The approach addresses an emerging infrastructure-security question. A model developer using third-party GPU capacity may be placing valuable model weights inside infrastructure controlled by another organization. Hardware-enforced confidential computing is designed to reduce that exposure by extending protection to data while it is being processed, rather than relying solely on encryption at rest and in transit or on contractual and administrative controls.

That gives Corvex an opportunity to position security as part of the infrastructure product rather than simply a compliance requirement. Demsey told DCF that customers are coming to Corvex in part for a neocloud-style computing environment combined with greater emphasis on securing customer workloads. That security focus also helps explain where Corvex is attempting to differentiate within the rapidly expanding neocloud market.

A Different Bet in the Neocloud Market

Corvex is entering a market in which larger neocloud providers increasingly operate at hyperscale. CoreWeave, for example, reported $2.08 billion in first-quarter 2026 revenue, a $99.4 billion revenue backlog, more than 1 GW of active power and more than 3.5 GW of contracted power. The company has also moved rapidly onto NVIDIA's technology roadmap, completing what it described in June as the industry's first operational bring-up and validation of Vera Rubin NVL72.

Lambda has likewise built its strategy around large dedicated AI environments while developing substantially deeper infrastructure-financing capacity, including a $1 billion senior secured credit facility announced in May and a $926 million GPU infrastructure financing in August. Nebius, meanwhile, reported more than 3.5 GW of contracted capacity after the first quarter and is building toward more than 5 GW of NVIDIA systems by 2030 under a partnership that includes a $2 billion NVIDIA investment.

Corvex is not competing with those companies on scale. Its nearer-term opportunity is narrower: dedicated and customized deployments, security-sensitive workloads and projects where quickly adapting existing powered data center capacity may matter more than access to hundreds of megawatts of new infrastructure. Its Token Factory inference service suggests ambitions farther up the software stack, but that offering remains early.

Corvex’s Opportunity and Its Risk

Corvex remains an early-stage company financially. Second-quarter 2026 revenue was $3.8 million, while net loss attributable to common shareholders was $12.8 million. Cash and cash equivalents stood at $21.7 million as of June 30. As of Aug. 14, Corvex reported approximately $22 million in contracted annualized recurring revenue from compute that was already live and accepted by customers. The company said all current AI Platform and services revenue comes from fixed-term contracts rather than spot GPU rentals.

Corvex's latest quarterly filing also underscores the other side of the growth equation. The company reported significant operating losses and negative operating cash flow since inception and said it expects net losses to continue for the foreseeable future as it invests in expanding its AI cloud business. Corvex nevertheless said its existing cash position should fund projected operating requirements for at least the following 12 months.

That fixed-contract model can also support infrastructure financing. Corvex has used customer prepayments alongside equipment-backed debt and its own capital, potentially reducing the amount of GPU capacity it must fund speculatively before securing customers.

But scale remains the central challenge. The AI infrastructure market is large enough to support both hyperscale AI clouds and specialists. Enterprises, government agencies and model developers may increasingly require environments optimized around security, sovereignty, dedicated capacity and customized deployment instead of simply purchasing generic GPU instances.

The Blackwell deployment matters because it provides evidence that Corvex can address one of the more difficult requirements of next-generation GPU infrastructure—high thermal density—inside an existing facility not originally designed around a central liquid-cooling system.

If the approach proves repeatable, it could broaden the pool of existing powered data center capacity capable of supporting some liquid-cooled AI deployments. That opportunity will still depend on power, electrical distribution, structural capacity, networking and other facility-level requirements beyond cooling.

The more compelling long-term Corvex story, therefore, may not be that it becomes another CoreWeave or monster neocloud provider. Trying to win that race purely through GPU count and gigawatts would place the company against competitors with vastly greater capital.

A more targeted strategy is emerging: compete as an engineering-driven AI infrastructure provider for customers that need new NVIDIA systems quickly, cannot wait for greenfield capacity, require dedicated infrastructure or place an unusually high value on protecting models and data from the infrastructure operator.

The Blackwell deployment gives Corvex tangible evidence that it can bring high-density AI infrastructure online quickly under the right facility conditions. Its confidential-computing work offers a second potential point of differentiation.

The harder test is whether both can scale. Corvex now needs to demonstrate that two-week deployments and security-focused AI infrastructure can become a repeatable, capital-efficient business rather than a series of technically impressive projects. If it can, Corvex may find a useful position in the AI infrastructure market—not by matching the largest neoclouds GPU for GPU or gigawatt for gigawatt, but by making existing data center capacity work for customers whose priorities are speed, dedicated infrastructure and security.

At Data Center Frontier, we talk the industry talk and walk the industry walk. In that spirit, DCF Staff members may occasionally use AI tools to assist with content. 

Keep pace with the fast-moving world of data centers and cloud computing by connecting with Data Center Frontier on LinkedIn, following us on X/Twitter and Facebook, as well as on BlueSky, and signing up for our weekly newsletters using the form below.