$HIVE (+0%) Through its subsidiary BUZZ HPC, the company has signed a five-year contract for GPU cloud services with an “investment-grade” enterprise customer. This brings the annual target of $200 million in GPU cloud ARR within reach.
Total contract value: ~$350 million over 5 years
Additional ARR: ~$70 million per year
CapEx: ~$185 million
Customer down payment: ~$35 million (~10% of the total contract value)
Hardware: 2,016 NVIDIA $NVDA Blackwell Ultra GPUs (GB300 NVL72, Quantum-X800 InfiniBand, VAST Data Storage)
Location: Bell $BCE’s AI Fabric in Merritt, British Columbia (100% hydroelectric power)
Launch: Q4 2026, followed by ~$500,000 in daily revenue from the HPC/AI business
Performance (estimated): ~28 GB300-NVL72 racks at ~135 kW each → ~4 MW IT load or ~5 MW including cooling (HIVE does not specify the MW figure)
Revenue calculation: ~$174,000 per GPU over 5 years (~$35,000 ARR/year) = ~$14–18 million ARR/MW/year
HIVE retains ownership of the GPUs and is financing the purchase through a 0% convertible bond maturing in June 2026, as well as additional debt and equipment financing. Following the $220 million deal with Bell and Cohere in June, this is the second major cluster—and the next step away from pure crypto mining toward AI infrastructure.
Source:
