AIBRIEF
BUZZ HPC, HIVE's AI infrastructure unit, signs the five year deal with an unnamed 'investment grade enterprise customer' who pre pays only ~10%; HIVE retains ownership of the hardware.
HIVE Digital Technologies, a TSX- and NASDAQ-listed operator that runs AI-infrastructure-as-a-service through its BUZZ HPC subsidiary, is putting up roughly $185 million of its own capital to deliver 2,016 NVIDIA Blackwell Ultra GPUs to an unnamed "investment-grade enterprise customer" under a five-year, $350 million GPU cloud services deal, the company said Monday. The customer is pre-paying only about 10 percent of the contract value, or roughly $35 million, before hardware arrives (HIVE press release).
The deployment, at HIVE's Bell AI Fabric facility in Merritt, British Columbia, will use NVIDIA's current-generation GB300 NVL72 rack-scale systems, a rack-scale unit that pairs 72 Blackwell Ultra GPUs with NVIDIA Quantum-X800 InfiniBand networking and VAST Data high-performance storage. The site runs on renewable hydroelectric power and uses closed-loop liquid cooling designed to eliminate ongoing water consumption.
Once the cluster comes online in Q4 2026, the deal is expected to add about $70 million in annualized revenue, lifting BUZZ HPC's total annualized run-rate to roughly $180 million. HIVE targets $500,000 in daily HPC and AI revenue once the new hardware is fully deployed, and the company retains ownership of the NVIDIA infrastructure after the contract ends.
The counterparty is not named in the press release, and the deal's revenue, deployment, and utilization figures remain contingent on Q4 2026 hardware delivery. HIVE's announcement, distributed via Proactive Investors, is the only cited source.