RegulatoryHIVE

BUZZ HPC lands $350 million AI cloud deal as Canadian GPU capacity builds

Demand for dedicated AI compute infrastructure in Canada is moving from announcement to signed paper. HIVE Digital Technologies (TSX: HIVE, NASDAQ: HIVE) disclosed through its subsidiary BUZZ High Performance Computing Inc. on…

By Owen Gallagher·August 20, 2026·二〇二六年八月二十日·2 min read

Key takeaways

  • HIVE Digital Technologies, through subsidiary BUZZ HPC, signed a five-year GPU cloud services agreement worth approximately $350 million in total contract value with an unnamed investment-grade enterprise customer, disclosed August 17.
  • The deal adds roughly $70 million in annualized revenue and calls for 2,016 NVIDIA Blackwell Ultra GPUs in GB300 NVL72 rack-scale systems deployed in Q4 of this calendar year at the Bell AI Fabric facility in Merritt, British Columbia.
  • HIVE expects deployment capital expenditures of approximately $185 million, partially offset by the customer's upfront deposit of roughly $35 million (about 10% of contract value).
  • The agreement raises BUZZ HPC's total contracted annualized revenue to approximately $180 million and supports HIVE's year-end target of $200 million for the GPU cloud unit.
  • Full revenue depends on hardware deployment and acceptance, and the $185 million capex must be financed partly through equipment debt and proceeds from HIVE's June 2026 zero-coupon convertible bond.

Demand for dedicated AI compute infrastructure in Canada is moving from announcement to signed paper. HIVE Digital Technologies (TSX: HIVE, NASDAQ: HIVE) disclosed through its subsidiary BUZZ High Performance Computing Inc. on August 17 that it had locked in a five-year GPU cloud services agreement worth approximately $350 million in total contract value with an unnamed investment-grade enterprise customer, adding roughly $70 million in annualized revenue to BUZZ HPC's book.

The physical specifics are precise. The contract calls for delivery of 2,016 NVIDIA Blackwell Ultra GPUs configured in NVIDIA GB300 NVL72 rack-scale systems, interconnected via NVIDIA Quantum-X800 InfiniBand networking and backed by VAST Data storage. That hardware is expected to be delivered and deployed in Q4 of this calendar year at the Bell AI Fabric facility in Merritt, British Columbia, a site running on 100% renewable hydroelectric power with closed-loop direct liquid cooling. HIVE says it expects capital expenditures for the deployment, including hardware and service warranties, to total approximately $185 million. The customer's upfront deposit of roughly 10% of total contract value, approximately $35 million, provides some early offset against that outlay.

Where the deal sits in the cycle

The agreement brings BUZZ HPC's total contracted annualized revenue to approximately $180 million, composed of roughly $35 million in active, realized revenue today and approximately $145 million in contracted revenue the company expects to come online through Q4 2026. HIVE's year-end target for the GPU cloud unit is $200 million of annualized revenue, per CEO Aydin Kilic. Once the Merritt cluster reaches full deployment, HIVE expects combined HPC and AI daily revenue of approximately $500,000.

The macro read-through runs to the Canadian sovereign AI conversation. BUZZ HPC is building at a moment when domestic enterprises and policymakers are actively seeking compute capacity that sits on Canadian soil and runs on clean power. Executive Chairman Frank Holmes cited approximately 400 MW of capacity available for Tier III data center development in Canada and said the company believes it could bring more than 120,000 GPUs online over the next two years. That is a management projection, not a contracted figure.

Bell Canada and VAST Data are the named infrastructure partners. NVIDIA's reference architecture anchors the cluster design.

The caveat is the one the warehouses always surface first: the customer remains unnamed, full revenue does not begin until the hardware is deployed and accepted, and the $185 million capex commitment must be financed in part through equipment debt and proceeds from HIVE's June 2026 zero-coupon convertible bond. Until the Merritt cluster is operational and generating that projected $500,000 daily, the $350 million sits on paper.

Related reading

Source · 來源

sec.gov

Share · 分享

Frequently asked

Who is the customer in the $350 million deal?

The customer is an unnamed investment-grade enterprise; HIVE did not disclose its identity.

Where and when will the GPU cluster be deployed?

The hardware is expected to be delivered and deployed in Q4 of this calendar year at the Bell AI Fabric facility in Merritt, British Columbia, which runs on 100% renewable hydroelectric power with closed-loop direct liquid cooling.

What hardware does the contract include?

It calls for 2,016 NVIDIA Blackwell Ultra GPUs configured in NVIDIA GB300 NVL72 rack-scale systems, interconnected via NVIDIA Quantum-X800 InfiniBand networking and backed by VAST Data storage.

How much daily revenue is expected once the cluster is fully deployed?

Once the Merritt cluster reaches full deployment, HIVE expects combined HPC and AI daily revenue of approximately $500,000.

How many GPUs does HIVE project it could bring online?

Executive Chairman Frank Holmes said the company believes it could bring more than 120,000 GPUs online over the next two years, a management projection rather than a contracted figure.