Axe Compute, trading under the ticker AGPU, has secured its largest contract to date, a 36-month enterprise agreement valued at approximately $260 million. The deal involves supplying 2,304 of Nvidia's latest B300 GPUs, part of the Blackwell Ultra series, along with high-speed storage solutions to an unnamed enterprise client. Deployment is scheduled to commence in the third quarter of 2026 at a U.S.-based Tier 3 data center, a facility classification denoting fully redundant systems that allow maintenance without service interruption.
Strategic Shift Amid Minimal Legacy Revenue
The substantial contract arrives at a critical juncture for the Pittsburgh-based firm, which recently reported total revenue of just $125,284 for the full year 2025. This revenue was entirely derived from its former drug-discovery segment, a legacy of its previous identity as Predictive Oncology. The company confirmed it has not yet recognized any revenue from its new AI infrastructure business line, making this agreement a pivotal test of its strategic pivot.
CEO Christopher Miglino positioned the deal as a significant market signal. He stated that enterprises are increasingly moving away from tailoring their AI ambitions to the constraints of major cloud vendors and are instead willing to pay a premium for dedicated, U.S.-based infrastructure to avoid queues for shared cloud capacity.
Payment Structure and Execution Risks
The contract follows a "take-or-pay" structure, requiring the customer to pay for the full booked capacity regardless of usage. Payment terms include an initial deposit, a prepayment, and subsequent monthly advance charges. The agreement spans three years and includes options for renewal.
However, Axe Compute has cautioned investors that the $260 million figure is not guaranteed. The final value is contingent on contract terms, and the company faces several execution risks. These include potential supply-chain disruptions, delays in bringing the data center facility fully online, challenges with enforceability, customer follow-through, and intense competitive pressures, all of which could reduce the final payout or slow revenue recognition.
Market Context and Asset-Light Model
The announcement coincides with a widespread scramble for AI computing capacity. Firms like CoreWeave and Nebius have been linked to multibillion-dollar compute deals, highlighting the sector's growth. Analysts at Morgan Stanley have noted a shift in the computing bottleneck toward CPU and memory as AI applications evolve from generation to autonomous action, though demand for GPUs remains strong.
Axe Compute operates an asset-light marketplace model, distinct from capital-intensive cloud giants. The company does not own data centers or GPU hardware outright. Instead, it acts as a broker, which limits upfront capital requirements but also makes its operations dependent on third-party providers and their infrastructure availability.
Path to Revenue and Previous Disclosures
Miglino has previously emphasized that driving revenue and growth are the top priorities for 2026. Prior to this major deal, on April 1, the company disclosed it had secured roughly $12 million in signed agreements with over 20 enterprise customers, projecting an estimated $835,000 in monthly recurring revenue once those systems are operational. The CEO stressed at the time that this backlog was a concrete business milestone, not merely a marketing achievement.
The core challenge for Axe Compute now is the operational execution of bringing 2,304 high-end B300 GPUs online as planned and successfully converting a signed contract into recognized, booked revenue—a crucial step for a company whose market valuation will likely hinge on proving its AI infrastructure model can scale profitably.
