How Canada could improve its AI sovereignty

The government might need to step up its game if Canadian startups are going to keep up in the AI race.

What happened: Industry Minister François-Philippe Champagne signed a letter of intent with Nvidia. Details have not been released, but Champagne said in an X post that the government and chip maker would “explore opportunities” to create AI computing power in Canada.

Zoom out: The announcement fits into a larger conversation about AI sovereignty. Promoted by the likes of Nvidia CEO Jensen Huang, it’s an idea that nations who want the business and academic benefits of AI should secure their own chips, data centres, and other computing resources. Ideas for how to do this vary:

• Building fabrication plants for chips, though few governments (or private companies, for that matter) can stomach the billions of dollars needed to build a new factory.
   
• Funding more public supercomputers, similar to the ones that already exist at five Canadian universities. AI pioneer Yoshua Bengio has suggested this would keep Canada from losing its spot as an AI research leader.
   
• Creating a government-backed consortium where public and private organizations combine their buying power to get computing resources more efficiently.
   
• Treating computing power as public infrastructure, similar to telecommunications or utilities, something currently in the works in China.

Catch-up: Ottawa seems to realize that computing power should be a priority, signing a sharing agreement with the U.K. last week. Minister Champagne also met with his EU counterpart to begin work on the Canada-EU Digital Partnership, which includes addressing disruptions in semiconductor supply chains.

Why it matters: If the government delivers on its promise for more localized computing power, it could become more accessible and affordable to AI startups, who otherwise buy from multinational cloud providers. It could also keep supply chain issues from slowing down work in a field that demands rapid innovation.

• Some AI startups spend 80% of their funding on computing.

Big picture: Nvidia has an obvious stake in AI sovereignty. It won’t be able to keep up its red-hot revenue surge forever, especially once tech companies finish stocking up on chips, so governments may be Nvidia’s next source of growth.