Skip to news

Bell Quadruples Saskatchewan AI Campus to 1.2 Gigawatts

Bell’s planned Saskatchewan expansion would make it one of Canada’s largest AI infrastructure projects, but its scale depends on power, permits and customers.

By THE COLDAI TIMES deskPublished 3 min read479 words

Bell Canada and Saskatchewan officials have announced a non-binding plan to expand a planned AI data center near Regina from 300 megawatts to as much as 1.2 gigawatts, potentially creating one of Canada’s largest dedicated computing hubs. The project is tied to more than C$50 billion in estimated capital investment, with the expansion unveiled September 14 at Canada’s national investment summit.

What changed

The original facility, announced earlier this year, was designed around 300 megawatts of capacity. Bell’s new memorandum of understanding adds up to 900 megawatts in phases, bringing the possible total to 1.2 gigawatts. Bell and the Saskatchewan government say the project could support thousands of construction, operations, logistics and maintenance jobs, while establishing the province as the headquarters of Bell’s national AI infrastructure business.

The power model is as important as the computing plan. The first 300 megawatts would remain connected to Saskatchewan’s grid, while the additional capacity is expected to rely on partner-developed natural-gas generation paid for by Bell. Bell says the proposed facilities would use closed-loop cooling, avoiding municipal water consumption. The expansion remains subject to commercial agreements, permits, approvals and environmental assessments.

The Canadian Press reported that the enlarged project could represent the largest private-sector investment in Saskatchewan’s history. Government and company materials frame it as a sovereign-AI initiative: computing capacity owned and operated in Canada, under Canadian rules, for customers that may include businesses and public-sector organizations seeking alternatives to U.S.-based infrastructure.

Why it matters

The announcement shows how the AI infrastructure race is moving beyond chips and models into electricity procurement, industrial construction and national data policy. A 1.2-gigawatt campus would require an energy system comparable to that of a sizable metropolitan area, making the project a test of whether new generation can be built quickly enough to satisfy AI demand without shifting costs or reliability risks onto other users.

It also illustrates the appeal of “sovereign compute.” Canada has strong connectivity, abundant energy resources and close ties to U.S. technology markets, but much of its cloud and AI capacity is controlled by foreign providers. Bell’s proposal could give Canadian institutions more leverage over data residency and access to advanced computing, although sovereignty will depend on the hardware, software and customers ultimately operating at the site.

What remains uncertain

The headline figures describe a pathway, not a completed build. The memorandum is non-binding, and the extra 900 megawatts depends on permits, financing, power generation and customer commitments. Bell has not publicly disclosed the full tenant roster, the precise construction timetable or the economics of the proposed generation assets.

The project must also win local acceptance. Questions about emissions, land use, transmission, water, tax treatment and whether permanent employment will match the investment headline are likely to intensify as approvals proceed. For now, Saskatchewan has secured a major strategic bet on AI infrastructure—but not yet a finished AI economy.

Related stories