HIVE Digital Signs $225 Million GPU Deal With Bell AI Fabric

HIVE Digital Technologies has signed a three-year GPU cloud contract with Bell AI Fabric, a subsidiary of Bell Canada, valued at approximately $225 million. The company projects roughly $75 million in annual recurring revenue once the hardware is operational — a figure that works out to about $6.25 million per month, roughly $205,000 per day.

The agreement runs through HIVE's BUZZ High Performance Computing arm and covers the deployment of 2,304 NVIDIA GB200 GPUs at a facility in Merritt, British Columbia. The site draws on renewable hydroelectric power. Deliveries are expected in late 2026 or early 2027.

How is the deal financed?

The GPU cluster carries a price tag of approximately $165 million. To fund it, HIVE secured a $129 million fixed-rate loan finalized in late October 2026, carrying 9% interest over three years. At 9%, that loan costs HIVE roughly $11.6 million per year in interest alone before any principal repayment.

The structure is deliberate. The loan term matches the contract term: three years of customer payments align against three years of debt service. The $129 million loan covers most, though not all, of the cluster cost, leaving HIVE to finance the balance while Bell's payments carry the debt burden.

The critical detail is ownership. When the contract expires, HIVE retains the NVIDIA GB200 cluster outright, with the related debt repaid — provided the plan holds.

What is not said matters here. The $75 million revenue projection depends entirely on the cluster being operational on schedule. A delivery delay would push revenue back while the loan clock keeps running. The late-2026 to early-2027 delivery window is the near-term checkpoint investors should watch.

Where does this fit HIVE's broader strategy?

The Bell agreement is HIVE's second major GPU contract of 2026. In August, the company signed a five-year GPU cloud deal worth approximately $350 million, expected to add roughly $70 million in annual recurring revenue. Taken together, HIVE has signed GPU agreements totaling approximately $600 million in contract value across the year.

The scoreboard as of September 30, 2026:

• BUZZ HPC's contracted annual recurring revenue: approximately $185 million
• Year-end GPU cloud ARR target: approximately $200 million
• Gap remaining: roughly $15 million, with one quarter to close it

HIVE operates data centers in Canada, Sweden, and Paraguay, and continues Bitcoin mining operations alongside the expanding high-performance computing business. That dual model carries its own risk: the company remains exposed to both compute-demand cycles and mining economics.

Is the structure as sound as it looks?

On paper, the deal is largely self-funding. Matched loan and contract terms, a utility-grade customer in Bell, and full hardware ownership at expiry form a defensible structure. The caveats are operational, not financial design.

Revenue does not begin until hardware is running. Interest accrues regardless. And contracted ARR is a booking figure, not collected cash — the $185 million becomes real revenue only as clusters come online. HIVE's execution on deployment, not its contract signings, will determine whether the $200 million year-end target is met.

The delivery window of late 2026 to early 2027 is the test. If the Merritt cluster arrives on schedule, HIVE enters 2027 owning hardware, servicing debt from a telecom-grade customer, and holding a second signed revenue stream. If it does not, the loan matures on the same date either way.