Canada has spent years telling the world it is an artificial intelligence pioneer. Now comes the harder part: proving that research leadership can survive contact with the real economy.
Prime Minister Mark Carney’s new National Council on Artificial Intelligence is supposed to help make that transition. The group brings together researchers, entrepreneurs, investors, executives and public-sector figures to advise Ottawa on AI adoption, safety, governance and sovereign infrastructure.
Its remit is broad because Canada’s problem is broad: inventing AI is no longer enough. Nations need the compute, companies, capital, skills and regulatory confidence to deploy it.
The council sits inside the government’s AI for All strategy, launched in June. Ottawa wants AI adoption to rise from just over 12% to 60% by 2034, while generating an additional US$200 billion in economic growth and creating 250,000 AI-related jobs over five years.
Those numbers make this more than another advisory panel. They turn AI policy into an industrial strategy.
Canada has obvious advantages. Yoshua Bengio, one of modern AI’s foundational researchers, sits on the council alongside figures including Sanja Fidler, Valérie Pisano, Michelle Zatlyn and Ajay Agrawal.
Canada has universities, start-ups and established technology companies with deep AI expertise. What it has struggled with, like many innovation economies, is converting intellectual leadership into companies with global scale.
That is where sovereignty becomes more than political branding. AI systems depend on expensive infrastructure, powerful chips, data centres, energy and access to models. A country that produces researchers but depends heavily on foreign infrastructure can find itself renting the foundations of its digital economy.
Carney framed the challenge starkly: prosperity and sovereignty will belong to countries able to build, adopt and govern AI on their own terms. That ambition explains why the council will examine sovereign AI infrastructure alongside adoption, national champions and safety.
The risk is that ‘AI sovereignty’ becomes another slogan attached to spending programmes without answering difficult questions. How much domestic compute does Canada need? Which technologies must remain under Canadian control? Should government procurement favour domestic AI companies? How can Ottawa support national champions without insulating weak businesses from competition?
Digital Transformation Canada, launched in September, could become an important part of that equation. By modernising federal services while using government purchasing power to help Canadian digital companies scale, Ottawa can potentially act as both customer and catalyst.
But councils do not build infrastructure or create competitive companies. They advise governments that must choose where to place capital, accept technological risk and occasionally tolerate failure.
Canada’s next AI test therefore is not whether it can assemble impressive experts. It can. The test is whether their advice produces decisions fast enough for a technology moving faster than government usually does.
The AI race will not wait for another consultation. For Canada, that means turning research prestige into deployed systems and globally competitive businesses. Success will be measured not in strategies published, but in who owns the infrastructure, who captures the value and whether Canadians trust the systems reshaping their lives.
