Table of Contents
Craig Tavares, president and COO at BUZZ HPC, told RCR Wireless News that Canada is one of HIVE’s growth market, where the company is developing sovereign AI infrastructure anchored by abundant clean hydroelectric power, political stability, and strong demand from both domestic enterprises and allied nations seeking trusted compute outside the U.S. hyperscaler ecosystem
In sum – what to know:
Capacity expansion underway – HIVE is increasing AI data center capacity in Canada from 4 MW to 16.6 MW, adding a new site in British Columbia with further expansion planned.
GPU deployment scaling – The company targets over 4,000 GPUs in the near term and up to 6,000 longer term, supporting AI cloud growth.
Growth tied to colocation – Expansion is enabled through existing data center partnerships, with capacity secured but ongoing costs tied to GPU deployment and operations.
HIVE Digital Technologies is expanding its AI cloud infrastructure footprint in Canada, increasing its liquid-cooled data center capacity across Manitoba and British Columbia.
Through its BUZZ High Performance Computing (HPC) unit and an existing colocation partnership, the company is scaling from 4 MW of critical IT load in Manitoba to a total of 16.6 MW across two provinces. The expansion includes a new facility in British Columbia, adding 5 MW of immediate capacity, with an option to increase capacity by a further 7.6 MW from 2027.
The additional capacity is intended to support high-density AI workloads. The British Columbia site is expected to accommodate approximately 2,000 next-generation GPUs, complementing around 2,000 GPUs planned or deployed at the existing Manitoba site. In total, HIVE is targeting a near-term deployment of more than 4,000 GPUs across its Canadian operations.
The company outlined a broader expansion path tied to its colocation agreement, which could enable more than 6,000 GPUs to be deployed in Canada over time. The Manitoba site currently supports 504 GPUs, with remaining capacity available for additional installations.
HIVE said the expanded colocation capacity has already been secured through prior deposits made in 2025, meaning no additional capital expenditure is required to access the contracted data center space. However, costs related to GPU procurement, installation, and ongoing operations remain ongoing.
HIVE is targeting 6,000 GPUs deployed in Canada and an annualized run-rate revenue of $200 million from its high-performance computing segment by the end of its fiscal year in March 2027. The company also indicated that long-term enterprise GPU contracts are expected to operate at relatively high margins, reflecting demand for dedicated AI compute infrastructure.
“One of our growth market is Canada, where we’re developing sovereign AI infrastructure anchored by abundant clean hydroelectric power, political stability, and strong demand from both domestic enterprises and allied nations seeking trusted compute outside the U.S. hyperscaler ecosystem. Beyond that, we evaluate new markets against three criteria: access to low-cost renewable energy, supportive regulatory frameworks for data sovereignty, relationships with local government and stakeholders and proximity to underserved demand. Latin America and select markets in the Nordics all score well on that matrix,” Craig Tavares, president and COO at BUZZ HPC, told RCR Wireless News.
“Our goal is to continue to develop AI factories, which are the new engines of the digital economy: purpose-built infrastructure where power, data, and accelerated compute come together to train, fine-tune, and run the AI systems that will shape every industry. They matter because AI will not be built or scaled in one place alone; every region needs trusted, high-performance infrastructure to support its own innovation, protect its data, strengthen competitiveness, and ensure it is not left behind in the next industrial era,” he added.
As demand for AI infrastructure grows, enterprises are also facing increasing pressure to manage risks tied to data location, regulation, and control. According to a blog post by Ron Babin, adjunct research advisor at IDC, AI sovereignty is becoming a central concern for organizations operating across multiple jurisdictions. The concept refers to control over the full AI stack, including data, models, and the underlying compute infrastructure.
IDC notes that governments are increasingly asserting authority over how AI systems are developed and deployed within their borders. This shift is forcing CIOs to move beyond purely technical considerations and treat AI as a broader strategic and regulatory issue. Organizations must align AI deployment with local laws while maintaining flexibility across different markets, according to IDC.
The firm highlights a range of risks associated with AI adoption, including exposure to foreign regulations, data and intellectual property loss, supply chain dependencies, and rising infrastructure costs. These factors are pushing enterprises to rethink reliance on centralized cloud platforms, particularly when sensitive workloads are involved.
To address these challenges, IDC recommends a more distributed and controlled approach to AI infrastructure. This includes balancing global and local providers, securing sensitive data within defined boundaries, and adopting hybrid architectures that allow workloads to run closer to where data is generated. As a result, demand for regionally controlled, high-performance compute infrastructure is expected to increase.
This trend is also reflected in moves by major hyperscalers to address sovereignty requirements more directly. Amazon Web Services (AWS) previously announced the general availability of its European Sovereign Cloud, a dedicated cloud environment fully located within the European Union and designed to operate independently from other AWS regions.
The offering is physically and logically separated, with operations, infrastructure, and governance fully based in the EU. AWS said the platform is aimed at governments and enterprises handling sensitive data, providing strict controls around data residency, operational independence, and security.
