The federal government wants Canadian industry players to submit proposals to build new data centre projects, as part of its commitment to increasing compute capacity for AI.
As part of its federal budget commitments, Innovation, Science and Economic Development Canada (ISED) is calling for Canadian companies and consortia to submit proposals for building “sovereign, large‑scale AI data centres” with capacity greater than 100 megawatts (MW).
“This initiative aims to strengthen Canada’s AI sovereignty, drive economic growth, support Indigenous participation, and maximize the use of Canadian partners and supply chains.”
While a standard data centre generally requires between five and 10 MW of power, one AI “hyperscale” data centre typically demands more than 100 MW, according to the International Energy Agency (IEA). For context, one megawatt-hour (MWh) could power 300 homes for a day.
Submissions for the feds’ new data centre projects opened on Jan. 15 and will stay open until Feb. 15. That’s an accelerated timeline similar to the feds’ previous AI Compute Access Fund, which covers compute costs for small businesses.
“This initiative aims to strengthen Canada’s AI sovereignty, drive economic growth, support Indigenous participation, and maximize the use of Canadian partners and supply chains,” AI minister Evan Solomon wrote in a LinkedIn post Tuesday.
Solomon’s ministry said it would “enter into memoranda of understanding (MOU)” with the proponents of selected projects, but did not indicate if financial support would be available. BetaKit has reached out to Solomon’s office for comment.
The government said it will prioritize projects that show a clear path to completion, “include Indigenous participation, minimize environmental impacts and maximize the use of Canadian partners and supply chains.” Other listed criteria include “sovereignty factors,” energy considerations, and project cost.
The submission form includes questions asking if the proposed project will be majority-controlled by a Canadian organization, and if all of the data processed will be domiciled in Canada.
Data centre energy consumption is expected to reach 1,400 terawatt-hours by 2030, roughly four percent of global power demand, according to consulting firm McKinsey. This increased demand is driven largely by training energy-hungry AI models.
The federal government has echoed industry and researchers in saying that Canada needs to grow its computing capacity without relying on US providers, as generative AI usage puts pressure on demand. It committed $2 billion toward AI infrastructure through its Canadian Sovereign AI Compute Strategy in the final months of former Prime Minister Justin Trudeau’s tenure. The initiative aims to increase access to computing power for Canadian companies and researchers, and includes a $300-million fund for small companies.
The feds’ new Major Projects Office has also pledged to build a Canadian sovereign cloud, though experts have cautioned that an entirely Canadian supply chain may be difficult to achieve. At the AI conference ALL IN in September, Solomon said the government’s priority is ensuring that Canadians’ “key sensitive data” is hosted securely and is “free from coercion.”
Feature image courtesy Unsplash. Image by Kevin Ache.