Canada’s emerging strategy for the next generation of data centers is not simply a technology policy. It is an electricity policy, an industrial policy, a sovereignty policy and, increasingly, a test of federal-provincial coordination. The country has the benefit of being able to watch what is happening with its southern neighbor, as power demand far outstrips availability, and data centers become a political hot potato.

The country wants to position itself as a preferred home for artificial intelligence infrastructure, sovereign cloud capacity and high-performance computing. But the central question is no longer whether Canada has enough research talent, cold weather, land or fiber. The question is whether it can build, connect and finance enough power quickly enough to support AI-scale data centers without sacrificing affordability, reliability or climate goals.

A National Grid Strategy

That is why Canada’s new National Electricity Strategy matters to the data center industry. Ottawa has formally connected the future of AI infrastructure to a national build-out of generation, transmission, storage, interties, grid modernization and skilled labor. When announcing the new energy plan, the Rt. Hon. Mark Carney, Prime Minister of Canada said:

In broad terms, the strategy calls for doubling the capacity of Canada’s electricity system by 2050 while maintaining a predominantly non-emitting power mix. Ottawa estimates that the resulting electrification could lower total energy costs for seven in 10 Canadian households. For hyperscale data center developers, that signals opportunity. For utilities and provincial system planners, it signals pressure. And for communities, it raises an important question: where is the money coming from for the new grid?

Canada already has a head start in building energy for the future and in supporting data center development. The country has a high share of non-carbon-emitting electricity, approximately 80%, much of it from hydro and nuclear power. It has a cool climate that can reduce cooling loads. It has large land areas outside crowded U.S. data center hubs. It has deep engineering, telecom and AI talent. And it has a political argument for sovereign compute: Canadian businesses, researchers, universities and public institutions should not be entirely dependent on foreign-owned compute platforms and foreign-controlled cloud infrastructure.

But Canada’s weakness is equally clear. And those weaknesses come as no surprise to anyone familiar with AI data center development. Available grid capacity is scarce in the places developers most want to build. Provincial utilities are being flooded with connection requests. Transmission takes years to site and build. Nuclear takes longer. Hydroelectric reservoirs are geographically fixed. Wind, solar and batteries can be deployed faster, but they do not provide the 24/7 firm power profile data center operators prefer. And while natural gas can move quickly, especially in Alberta and Saskatchewan, but gas-fired AI infrastructure complicates the clean-power narrative that the Canadian government  is trying to sell.

The federal government’s answer is to treat electricity as nation-building infrastructure. The National Electricity Strategy branded as “Powering Canada Strong,” will, according to the Prime Minister’s announcement, be guided by four pillars:

1.      Build enough new infrastructure to double electricity generation

2.      Connect Canada’s fragmented provincial grids

3.      Train and retain the workforce needed for the build-out

4.      Strengthen domestic manufacturing of grid components

The plan recognizes that AI data centers are only one driver of demand. Electric vehicles, heat pumps, new housing, critical minerals, manufacturing, ports, electrified industry and northern development are all competing for the same future electrons. The plan also makes it clear that for Canada to be part of the AI driven future, the energy infrastructure must be in place, and supported by the government.

For the data center sector, the most important piece is not a single subsidy. It is the emerging financing stack. Ottawa is putting clean economy investment tax credits, the Canada Infrastructure Bank, the Canada Growth Fund, Indigenous loan guarantees and targeted clean electricity programs into the same policy frame. The Clean Electricity Investment Tax Credit is designed to support low-emitting generation, storage and transmission. The Smart Renewables and Electrification Pathways Program supports grid modernization, storage and renewables. The Canada Infrastructure Bank is positioned to bring private capital into generation, transmission, storage, district energy and clean infrastructure. The Indigenous Loan Guarantee Program has also been expanded, reflecting the growing role of Indigenous consultation, partnership, ownership and equity participation in major energy and transmission projects.

Data center power demand is too large to be solved by utility rate base expansion alone. The most politically durable model will require data center developers to help pay for the power and grid infrastructure they need. In the old model, large customers queued for power on a relatively technical first-come, first-served basis. In the new model, provinces are asking whether a project delivers economic value, strategic benefit, environmental performance, Indigenous participation, community support and grid adequacy.

Five Provinces, Five Power Models

That shift is already visible across the country. Québec, long one of North America’s most attractive power markets because of low-cost hydroelectricity, is no longer treating data centers as ordinary industrial loads. Hydro-Québec has proposed a separate rate for large data centers, with new customers above five megawatts paying substantially more than traditional large-power users. The stated goal is to ensure that energy-intensive digital infrastructure pays a price closer to the value of scarce renewable electricity. Québec expects data center electricity use to rise sharply by 2035, and its policy is designed to capture more value for the province while protecting other customers. Claudine Bouchard, President and Chief Executive Officer of Hydro-Québec said:

Ontario faces a different challenge. Its grid is anchored by nuclear power, with hydro, gas, wind, solar and storage also in the mix. The province’s system planner expects electricity demand to grow significantly through 2050, and data centers are now a much larger part of the forecast. Its plan, titled “Energy for Generations”,  is to procure new resources, extend and refurbish nuclear assets, build transmission, use energy efficiency and prepare for long-lead projects such as hydro and long-duration storage. For data center developers, Ontario offers proximity to Toronto, cloud regions, enterprise customers, universities and nuclear baseload. But the province will not have unlimited near-term capacity, and connection policy is moving toward screening projects for strategic and community benefits.

Alberta represents the biggest point of contention in Canada’s data center strategy. The province is aggressively pursuing data center development with its Artificial Intelligence Data Center Strategy. It has abundant natural gas, large land parcels, a deregulated power market, experienced energy developers and political leaders actively courting AI infrastructure. That makes it attractive to data center operators that care most about speed to power. Alberta has also promoted “bring your own generation” models, where data center developers pair facilities with dedicated generation rather than relying entirely on the public grid. But Alberta’s electricity system is much more carbon-intensive than Québec, British Columbia, Manitoba or Ontario. The same feature that makes it attractive for development, potential  large AI build-outs powered primarily by natural gas, would undercut Canada’s claim that its data centers can run on some of the cleanest power in the world.

Saskatchewan illustrates another version of the opportunity. Bell Canada’s planned 300-megawatt AI data center in the Rural Municipality of Sherwood near Regina is a major signal that large-scale AI infrastructure can move beyond the traditional Toronto-Montreal-Calgary corridor. The project combines domestic telecom infrastructure, sovereign compute ambitions, hyperscale tenants, fiber partnerships, Indigenous procurement participation and closed-loop cooling. It also shows why power availability is now the deciding factor in site selection. At 300 megawatts, a single facility becomes a grid-planning event, not merely a real estate development.

British Columbia, meanwhile, is trying to prioritize power among competing industrial demands. Data centers are arriving at the same time as mining, LNG, manufacturing, forestry, hydrogen and electrification projects. The province has moved toward limiting and screening certain high-load uses, including data centers and cryptocurrency mining, so that scarce clean electricity is allocated to projects with the strongest public benefit.

This seems to be a preview of the future for these industrial scale projects in Canada, Not every data center proposal will be treated equally. Projects that bring jobs, Canadian ownership, Indigenous participation, waste heat reuse, grid investment, efficient cooling and low-carbon power will have a stronger case than projects that merely ask for a large block of electricity.

Transmission and Nuclear: The Long Game

As we see so often, transmission is the least glamorous but most important part of the plan. Canada’s power system is fragmented by provincial boundaries. Hydroelectric strength in Québec, Manitoba and British Columbia does not automatically solve load growth in Alberta, Saskatchewan, Ontario or the Maritimes. Interties can allow provinces to share surplus energy, balance variable renewables, reduce duplication and improve reliability. The government is now pushing a more coordinated intertie strategy, including projects such as Alberta-British Columbia, Alberta-Saskatchewan, Saskatchewan-Manitoba and Atlantic Canada connections. The Saskatchewan-Manitoba concept is especially important because it could move large amounts of low-carbon power across the Prairies, where data center growth and industrial electrification would otherwise lean heavily on gas.

Nuclear is another pillar for the technology development. Canada is presenting itself as a “Tier One” nuclear nation, with uranium resources, CANDU experience (a Canadian designed heavy water pressurized reactor design in use worldwide), a skilled workforce, nuclear regulators and active small modular reactor development. Ontario’s Darlington SMR program is a first-mover project, and the federal government has now set out a broader nuclear strategy that contemplates up to ten new large reactors, with two under construction by 2035 and more in planning by 2040. As always, especially with the AI market,  the timing is critical. AI data center developers want power in two to five years. Large reactors are more likely to shape the 2035-and-beyond market. SMRs may arrive earlier, but their commercial and financing models are still being proven.

The Problem of Competing Timelines

As we have seen, the mismatch between data center timelines and power-sector timelines is the heart of the problem. AI infrastructure can be planned, financed and built faster than major generation and transmission assets. Canada’s plan will succeed only if electricity planning becomes anticipatory rather than reactive. Waiting until data center interconnection queues explode is too late. The next generation of data centers may need to function as dispatchable digital infrastructure, reducing load when the grid is stressed and ramping up when clean power is abundant.

This very top-down approach being proposed by the Canadian government is likely to run into some long existing roadblocks. Provinces control much of the electricity system. Developers will go where power is available, not necessarily where it is cleanest. Unless federal incentives, procurement rules and provincial connection policies reward low-emissions power, Canada risks building a two-track AI infrastructure market: clean but capacity-constrained provinces on one side, faster but higher-emitting gas-backed provinces on the other.

If the needs of the rate payers, the AI developers, industry, and the political infrastructure cannot be aligned, the country will face a familiar Canadian problem: vast potential slowed by jurisdictional fragmentation, permitting delays, underbuilt transmission, and political disputes over who pays. The next generation of data centers will be built where power is available, clean enough, affordable enough and fast enough.

Canada’s new strategy is a recognition that the AI race will be won not only in laboratories and chip fabs, but also on transmission corridors, hydro reservoirs, nuclear sites, gas plants, storage projects and utility flexibility. The announcements and governmental directions announced over the last few months are a good start, but they are just that, a starting point.  In the AI era, compute is power. Canada’s challenge is to build both.

Keep pace with the fast-moving world of data centers and cloud computing by connecting with Data Center Frontier on LinkedIn, following us on X/Twitter and Facebook, as well as on BlueSky, and signing up for our weekly newsletters using the form below.