HIVE Digital Technologies' subsidiary BUZZ High Performance Computing has accumulated approximately $570 million in combined estimated AI cloud contract value through two deals signed in 2026, anchoring a pivot from Bitcoin mining to GPU-based AI services across Canada.

The subsidiary is repurposing infrastructure originally built for proof-of-work hashing into advanced GPU cloud capacity. The strategic logic pairs two market conditions: surging demand for AI compute and Canadian data residency rules that require sensitive workloads for hospitals, government agencies, and researchers to remain inside the country.

BUZZ HPC operates NVIDIA GPU infrastructure across five provinces: British Columbia, Manitoba, Ontario, Quebec, and New Brunswick. The customer set spans enterprises, government agencies, researchers, hospitals, and startups.

What contracts has BUZZ HPC secured?

In June 2026, BUZZ HPC signed a three-year agreement with Bell AI Fabric valued at approximately $220 million. The deployment covers 2,304 NVIDIA Grace Blackwell GPUs allocated to Cohere Inc., the Canadian AI company.

Two months later, in August 2026, BUZZ HPC closed a five-year AI cloud services contract estimated at $350 million. BUZZ HPC has not named the customer. That arrangement provisions 2,016 NVIDIA GB300 GPUs.

The combined $570 million pipeline supports a financial target: annual recurring revenue of approximately $180 million from GPU cloud services by September 2026. BUZZ HPC reported average daily revenue above $1 million as of September 10, 2026, marking a sharp acceleration from its prior mining-only revenue profile.

How does the pivot reshape HIVE's mining business?

HIVE spent years building warehouses of power-hungry machines for Bitcoin mining. Rather than retire those sites, the company repurposes them for AI compute. BUZZ HPC expanded its commercial bench through a collaboration with data services firm ProCogia and a series of executive appointments.

The operational consequence is asset reuse without scrapping capital. Mining-specific ASICs lose direct relevance once a site pivots, but the real estate, power purchase agreements, and cooling systems retain economic value when paired with NVIDIA accelerators. Provincial site diversity also gives BUZZ HPC redundancy and proximity to research clusters.

What is the Toronto gigafactory plan?

HIVE intends to construct a 320-megawatt AI gigafactory in the Greater Toronto Area, with commissioning targeted for the second half of 2027. At full capacity the facility will house more than 100,000 GPUs. HIVE has estimated the investment at approximately CAD $3.5 billion.

The site would consolidate BUZZ HPC's domestic footprint and serve customers with strict data residency mandates. Cohere's involvement through the Bell AI Fabric agreement links HIVE to one of Canada's most prominent AI developers, with thousands of Grace Blackwell GPUs allocated to Cohere workloads.

What risks and milestones remain?

Two large contracts account for the bulk of the disclosed pipeline, and one counterparty remains unnamed. Customer concentration will draw scrutiny as BUZZ HPC scales beyond its anchor agreements, particularly if the unnamed August 2026 client represents a material revenue share.

Financing the CAD $3.5 billion gigafactory will require capital structure decisions HIVE has not yet detailed. Watchpoints for the remainder of 2026 and 2027 include the September 2026 ARR print, Toronto site progress, and any new anchor customer disclosures beyond Cohere and the August 2026 unnamed client.