HIVE Digital Technologies subsidiary plans a CAD $3.5 billion data center capable of housing over 100,000 GPUs, marking a major shift in Canada's compute sovereignty
Canada has spent years outsourcing its AI compute needs to facilities south of the border and overseas. That chapter is getting a very expensive rewrite.
BUZZ HPC, a wholly owned subsidiary of HIVE Digital Technologies, announced plans to develop a 320 MW AI infrastructure facility in the Greater Toronto Area. The project carries an estimated price tag of roughly CAD $3.5 billion and is designed to house more than 100,000 GPUs at full build-out.
What’s actually being built #
The facility is being described as an “AI gigafactory.” A 320 MW utility capacity is enormous. To put that in perspective, a typical large data center runs somewhere in the range of 30 to 50 MW. This is roughly six to ten times that size.
BUZZ HPC is building around liquid-cooled GPU infrastructure. The facility will leverage NVIDIA GPUs, and the software orchestration layer relies on Kubernetes and Slurm, two widely used systems for managing large-scale compute clusters.
BUZZ HPC isn’t going it alone. The project involves partnerships with Bell Canada, Dell Technologies, and VAST Data. Bell brings telecom and fiber infrastructure expertise. Dell handles the hardware side. VAST Data specializes in high-performance storage systems.
HIVE’s bigger compute play #
HIVE Digital Technologies has been in the high-performance computing business since 2017. The company is an NVIDIA Cloud Partner, and BUZZ HPC currently supports a pipeline of approximately 130,000 GPUs across various Canadian sites. The GTA gigafactory is an expansion of an existing operational footprint that has previously included sites in New Brunswick, Manitoba, and British Columbia.
The broader strategy is about scaling liquid-cooled GPU capabilities across Canada. The GTA location gives access to fiber connectivity, relatively affordable power by global standards, and proximity to AI research institutions including the Vector Institute and the University of Toronto.
When Canadian AI infrastructure sits in another country, operators are subject to that country’s regulations, pricing, and capacity constraints. The GTA facility is designed to serve government, enterprise, and research clients as a domestic alternative.
What this means for investors #
A CAD $3.5 billion capital investment is a significant bet. HIVE is positioning itself not as a hyperscaler competitor but as a domestic infrastructure provider serving government, enterprise, and research clients.
The existing pipeline of approximately 130,000 GPUs across Canadian sites provides some revenue foundation, but a project of this magnitude will test the company’s capital allocation discipline. Investors should watch how HIVE finances the gap between the CAD $3.5 billion investment and its current revenue base, and whether that comes through debt, equity, strategic partnerships, or phased construction.
Building and filling a 320 MW facility with over 100,000 GPUs requires coordination across power procurement, construction timelines, hardware supply chains, and customer acquisition. GPU supply, while loosening compared to the extreme shortages of 2023, remains competitive, and construction delays in data center projects are common.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our