Canada's AI strategy bets on adoption, not frontier models

Ronni Holmvig Strøm · 2026-06-05

On June 4, Mark Carney launched AI for All, Canada's national AI strategy, in Toronto. The press release leads with the big figures. Canada projects $200 billion in

On June 4, Mark Carney launched AI for All, Canada's national AI strategy, in Toronto. The press release leads with the big figures. Canada projects $200 billion in economic growth, 250,000 new jobs, a $500-million fund for Canadian companies, and a public supercomputer. The headline ambition is to raise national AI adoption from just over 12% to 60% by 2034.

The strategy says something unusual for a G7 government. Canada is skipping the model race to close the gap between a capability that already exists and a country that has not yet picked it up.

Adoption Is the Strategy

Most national AI plans of the last three years read as industrial policy for frontier capability: compute, talent, a sovereign lab, a seat at the table where the largest models get built. AI for All has some of that, including the supercomputer and an expanded Canadian AI Safety Institute. The center of gravity sits elsewhere.

Canada's own framing is that it has "world-class talent and one of the fastest-growing digital sectors in the G7" while remaining "among the slowest countries to adopt AI at scale." The bottleneck it names is distance, the gap between what the models can already do and what Canadian firms and hospitals actually use.

So the spending follows use. A literacy initiative targets one million post-secondary students; a Missions Program leads with a flagship health mission in diagnostics and patient care; up to 90,000 work placements go to young people. Each puts an existing system into more hands rather than building a new one. It is the rare national plan that bets on deployment.

Trust Is the Bottleneck Carney Named

The strategy's three principles lead with trust, ahead of opportunity and sovereignty. A survey released ahead of the launch showed a public confidence gap, and Minister Evan Solomon said the plan would be built around closing it.

Adoption is not downstream of capability. It is downstream of whether a nurse or a small-business owner believes the output enough to act on it. Canada is treating that belief as infrastructure and funding it accordingly. The plan frames its online safety regime and transparency requirements as adoption enablers rather than brakes, which is the correct read.

The Guardrails Arrive Next

The privacy and online-harms legislation meant to support the strategy has not landed yet. Federal NDP leader Avi Lewis told The Tyee the plan leans heavy on business adoption and light on worker protection, and hundreds rallied in Vancouver last month against new data-centre projects. These are the questions the legislative text gets to answer, and getting them right is how a country earns the trust the strategy is built on.

The 12% is the country today. The 60% is the wager, and the legislation decides it.