
Axe Compute reported third-quarter revenue of $7.2 million, more than double the prior quarter’s total, as the company shifted its focus entirely to AI infrastructure following the sale of its legacy medical business.
The company also posted its first quarter of positive operating and net income, although both figures included gains outside its core compute operations. Executives said the next phase of growth depends on bringing contracted GPU clusters online and converting customer prepayments into revenue.
Revenue rose 125% from $3.2 million in the second quarter. All third-quarter revenue came from Axe Compute Access, the company’s shared-capacity offering; none came from its dedicated Build projects, Chief Financial Officer Jeremy Yaukey-Witter said.
Profit included asset gains
Operating income was $6.8 million, compared with an operating loss of $16.4 million in the second quarter. The latest result included a $10.5 million gain on digital assets, Yaukey-Witter said.
Net income was $11.5 million, compared with a $17.2 million net loss in the prior quarter. It included a $4.6 million unrealized gain on equity securities and a convertible note received in the sale of the Helomics business.
Adjusted EBITDA, a non-GAAP measure, was a loss of $2.3 million for the quarter. Yaukey-Witter said the company revised its definition of the measure to exclude unrealized gains and losses on the securities and note received in the Helomics sale, in addition to adjustments for digital-asset marks.
Axe ended September with $147.7 million in cash and $686.1 million in total assets. Customer prepayments of $569.3 million funded much of the balance-sheet expansion; those payments are recorded as contract liabilities and are recognized as revenue as services are delivered, the CFO said.
Four clusters in delivery
President Kyle Okamoto said Axe has more than 18,000 NVIDIA B300 GPUs in delivery across four dedicated AI infrastructure projects in the U.S. and Europe. The company still targets full deployment of its signed book across the fourth quarter of 2026 and into 2027, he said, with revenue beginning as each cluster completes customer acceptance and enters service.
- Georgia: A 2,304-GPU cluster is energized, provisioned and undergoing customer acceptance.
- Sweden: A cluster with more than 2,000 GPUs is being commissioned.
- Large U.S. cluster: A project with more than 9,000 GPUs is in manufacturing and site readiness.
- U.S. expansion: A cluster with more than 4,600 GPUs for an existing customer is also in manufacturing and readiness, with some equipment already on site.
Chair and Chief Executive Officer Christopher Miglino said Axe acquired the Georgia cluster as of Sept. 30 and paid off its asset-backed loan with cash on hand. The customer extended its contract to five years through 2031, and the company expects approximately $352.6 million in revenue over the contract’s life.
Axe still owes a deferred purchase price to Duos for the Georgia acquisition. The obligation had a face value of $42.9 million and was carried at $36.6 million, net of discount, at quarter-end. It is secured by the cluster and due by Sept. 30, 2027, or earlier upon refinancing, Yaukey-Witter said.
Contracts and pipeline expand
Miglino said the company had signed approximately $3 billion in contract value over the preceding seven months, including previously announced U.S. and European build agreements. He said none of the more than $3 billion in build-project contract value had been recognized in third-quarter revenue.
Separately, Okamoto said Axe signed $266 million in new contract value across eight agreements since its second-quarter call, comprising four new customers and four expansions or renewals. The company received $64.5 million in prepayments against those contracts. Some revenue from the agreements began in the third quarter, with the rest expected to activate through the fourth quarter and early first quarter, he said.
Okamoto said the company’s qualified pipeline rose to $14.1 billion from $5.9 billion in the prior quarter, covering 118 opportunities. He cautioned that the pipeline does not represent signed contracts and that not every opportunity will close. Axe considers an opportunity qualified when it knows the customer’s technical requirements, decision-makers, timing and commercial parameters.
Asked about customer transparency, Miglino said nondisclosure agreements often prevent Axe from identifying customers. He said most customers are enterprises, while some smaller Access agreements are with AI-native companies.
Miglino also said he expects hourly GPU rental prices to rise as global customers seek U.S. capacity, despite additional data-center space and power coming online. Okamoto said the company’s approach to inference workloads emphasizes locating infrastructure close to customers while maintaining a distributed footprint.
About Predictive Oncology (NASDAQ:POAI)
Predictive Oncology Inc. is a biotechnology company focused on applying artificial intelligence and machine learning to cancer drug discovery. Its approach combines computational tools with tumor samples and related biological and drug-response data to help researchers assess potential treatments and identify promising drug candidates. The company’s PEDAL platform is designed to analyze these data and model how tumors may respond to different compounds. Through its Helomics-related operations, Predictive Oncology has also offered laboratory services and access to tumor-based research resources for pharmaceutical and biotechnology partners.
