BUZZ to Build 320 MW AI Gigafactory in Greater Toronto Area

BUZZ targets 320 MW AI compute campus in Toronto-Waterloo corridor

HIVE Digital Technologies’ subsidiary BUZZ High Performance Computing has moved forward with plans for a large-scale AI data centre campus in the Greater Toronto Area, announcing a contiguous site with a 320 megawatt power allocation and land purchases totalling roughly 25 acres. The company said the project is designed to host more than 100,000 graphics processing units at full build-out and represents an estimated capital investment of about CAD 3.5 billion, with a target to bring initial capacity online in the second half of 2027.

The announcement positions the site as one of the largest AI-focused deployments proposed in Canada to date. BUZZ framed the initiative as part of a broader push for domestic compute capacity, arguing that onshore infrastructure is a strategic asset for research institutions, enterprises and public sector workloads that require data residency and low-latency access to large-scale GPU clusters.

Project basics and company context

According to company filings and statements, BUZZ paid approximately CAD 46 million for a main parcel of about 21 acres and an additional CAD 12 million for an adjacent four-acre parcel. The combined property benefits from a 320 MW allocation and is being developed as a vertically integrated supercomputing platform supporting Kubernetes, bare metal and other managed services for AI workloads.

HIVE and BUZZ also highlighted their existing footprint: HIVE reports over 850 MW of power under management globally, combining operating capacity and projects in development, and BUZZ operates AI infrastructure across multiple Canadian provinces and other jurisdictions. The firm currently has thousands of GPUs in operation and an earlier 70 MW site in New Brunswick among its national platform.

Why Toronto-Waterloo matters

The chosen location sits inside the broader Toronto-Waterloo innovation corridor, a cluster of universities, startups and enterprise demand for AI services. Low-latency links to research centres and engineering talent are central to the project’s rationale. For companies and institutions that prioritize data residency, onshore compute capacity can reduce reliance on foreign cloud providers and shorten networks for inference and training.

Industry context and implications

Large, dedicated AI campuses are becoming strategic assets worldwide as model sizes and inference workloads grow. Building domestic capacity responds to several converging trends: growing enterprise adoption of generative AI, concerns about data sovereignty, and competition for limited GPU supply. BUZZ’s plan highlights all three.

GPU demand and procurement — Hosting 100,000+ GPUs would materially increase demand for high-end accelerators in the region. Procurement remains a critical risk for any large AI build-out. Global supply of datacentre-grade accelerators has tightened at times in recent years, and lead times can affect deployment schedules and capital plans.

Grid and energy considerations — BUZZ says the facility will be built to operate on Ontario’s low-carbon grid and will use closed-loop cooling and other efficiency measures. Large AI campuses create concentrated electricity demand and typically require close coordination with utilities for transmission upgrades, interconnection agreements and potential demand management. Ontario’s generation mix includes significant nuclear and hydro resources, which can reduce carbon intensity compared with gas-heavy grids, but the addition of hundreds of megawatts of load still has technical and regulatory consequences.

Competition and market positioning — Hyperscalers and cloud providers already operate in Canada and in some cases offer domestic regions. BUZZ’s pitch is sovereignty and vertical integration: to provide onshore physical infrastructure under Canadian control for customers that cannot or prefer not to colocate with global cloud platforms. Whether enterprises and public-sector customers will shift workloads from existing providers depends on pricing, services, regulatory incentives and trust in local operators.

Permitting, timelines and execution risks

Large-scale data centre projects typically face a complex checklist: land use approvals, grid interconnection, construction logistics, equipment supply and labour. The company has indicated a timeline targeting the second half of 2027 for initial online capacity, but such schedules are sensitive to permitting processes, supply chain constraints for servers and accelerators, and potential grid upgrade lead times. BUZZ’s own statements note that timelines and costs are forward-looking and subject to risks.

Policy and economic impact

From a policy perspective, the project touches on national discussions about industrial-scale AI infrastructure. Several governments have signalled interest in expanding domestic compute for strategic and economic reasons. A major Canadian facility could support local research and startups, create construction and permanent operations jobs, and anchor related services in the region if market demand materializes.

However, public benefits will depend on how capacity is allocated and priced. If the campus primarily serves private enterprise and multinational customers, the economic spillovers will differ from a model that prioritizes public research access or subsidized academic usage. The announcement did not specify any formal public-private partnerships or government funding.

Bottom line

BUZZ’s proposed 320 MW AI campus in the Greater Toronto Area is a notable step toward expanding Canada’s domestic compute capacity. The plan aligns with broader industry moves to secure onshore GPU resources and to offer larger, more efficient GPU pools for enterprise and research workloads. Execution will hinge on GPU procurement, utility interconnection and construction delivery, and the final market impact will depend on how the facility is priced and integrated into Canada’s existing cloud and research ecosystem.

Observers in the data centre, AI and public policy communities will be watching whether the project meets its 2027 target and how it affects local compute markets, grid planning and the balance between hyperscaler-owned and locally owned AI infrastructure.

Disclosure: The information in this article is based on company disclosures and public statements. Projections and timelines cited by the company are forward-looking and subject to change.