Axe Compute Signs $1.5 Billion GPU Deal With Customer Unnamed
Data Center Dynamics reported a five-year $1.5 billion compute capacity agreement for Axe Compute, with 9,200 Nvidia B300 GPUs planned in the US and the customer still unnamed.
Axe Compute's $1.5 billion compute capacity agreement creates a larger AI infrastructure obligation than the newly renamed company has yet proved in revenue.
Data Center Dynamics reported that the five-year contract covers a US cluster of 9,200 Nvidia B300 GPUs for an unnamed customer.
Axe will operate the cluster itself and lease the capacity, so infrastructure delivery is part of the contract rather than only access to third-party compute.
The deal also moves the company away from the lighter model set out in its first annual report, where it had presented distributed GPU access rather than ownership of dedicated infrastructure.
Prepayments Fund A Larger Buildout
The new contract and other July agreements are expected to bring Axe more than $534 million in aggregate prepayments during the next month.
The company put its 2026 signed contract value above $3 billion after the latest award.
Those up-front payments are meant to sit beside project-level financing for the remaining buildouts.
Revenue is scheduled to start this month and rise as more deployments come online, but the public record ties that growth to clusters that still have to be delivered.
Christopher Miglino, Axe Compute's chief executive, framed the prepayment structure as a way to reduce reliance on equity financing.
His quoted remarks also put the latest agreement inside the company's Axe Build portfolio and compared the 2026 signed-contract total with where the business stood earlier in the year.
Annual Filing Shows A Business Pivot
The March 2026 annual report records the company's move into compute after earlier corporate identities including Skyline Medical, Precision Therapeutics and Predictive Oncology.
That filing describes a distributed network model using GPU capacity for artificial intelligence and high-performance computing workloads, sourced mainly through the Aethir network.
A June filing presents a more capital-intensive path.
The risk section identifies dedicated AI infrastructure clusters, infrastructure ownership, equipment life-span and depreciation as issues attached to the pivot.
The same filing places the company's largest contract at that point in a single data centre with 4.8MW of dedicated power.
It also records revenue concentration risk among a small number of customers and contractors for the foreseeable future.
The planned B300 cluster is not just a resale commitment.
Dedicated equipment ownership can bring depreciation, replacement and useful-life questions into a provider's economics before customer revenue has been fully recognised.
The filing also identifies execution risk around constructing, deploying and maintaining infrastructure clusters, leaving the prepayment model dependent on turning signed capacity into operating systems.
Q1 Results Put Scale Ahead Of Current Revenue
Axe Compute's Q1 2026 results released in May put quarterly revenue below the prior-year level, while compute revenue remained a small part of the business.
Total operating costs were $3.5m and the net loss was $7.7m.
The company targets a Q3 2026 customer deployment that, once live, would generate $21m per quarter in revenue.
That target gives the new GPU contract a clear execution gap: customer identity, site location and delivery schedule remain outside the public record.
