Canada Needs a Sovereign Intelligence Fund
PM Carney built the factory financing. Now Canada needs to build the factories of intelligence.
PM Carney announced the Canada Strong Fund this week. $25 billion to start, a Crown corporation structure, and an explicit invitation for Canadians to invest alongside the government.
I think it’s a great move. Long overdue. And I think the next move is even more important.
Jensen Huang has been making a point lately that I can’t stop thinking about. Chips and data centers are the factories of the future. Not metaphorically. Literally. They take in electricity and produce intelligence, sold by the token, consumed at the rate of billions per second.
Tokens are units of cognitive labor. As models get better, more of what used to be human work moves into the factory. Translation, writing, analysis, support, design, increasingly engineering. None of it disappears. It just shifts location.
Here is what worries me. Almost none of those factories are in Canada. The dominant ones are owned by a handful of US companies running on chips fabricated in Taiwan. When a Canadian business uses AI to do the work that used to require five employees, the productivity gain is real, but the rent flows to Nvidia, OpenAI, Microsoft, Google, Anthropic. Local wages compress. Foreign profits expand. Our trade balance on intelligence will start to look like an oil importer’s balance on oil.
A sovereign wealth fund is a good start. But the most important project of national interest right now is the factory itself. Canada should be building and operating token factories at national scale. Hydro-rich provinces, cold climate, stable rule of law, abundant land. The natural advantages are obvious. The financing structure has been the missing piece.
Call it a Sovereign Intelligence Fund. A Crown corporation with a single mandate: build and operate the compute infrastructure that produces tokens for Canadian use. Canadian companies live on top of that base layer, taking the raw output and curating it into products the world wants to buy. That is the part of the value chain we are good at. We will not get to play that game for long if we do not own the factory underneath.
The early pieces are in motion. The Cohere and CoreWeave facility in Cambridge is online. The federal AI compute strategy has $2 billion committed. The instinct is right. The scale is wrong. $2 billion is a rounding error compared to what a single US hyperscaler spends on data centers in a quarter.
The next move out of Ottawa should be ambitious. Tens of billions in scale, not single-digit billions. An explicit mandate to be the wholesale supplier of Canadian-controlled tokens to Canadian businesses. And a retail product that lets ordinary Canadians own a slice of it, just like the Canada Strong Fund.
We have the chance to collectively own the factories of the next economy. We should take it.

