Federal and provincial governments could combine their processing power purchases to encourage the growth of Canadian-owned and -operated AI infrastructure, according to a new analysis from the University of Toronto’s Munk School of Global Affairs & Public Policy. It also called for Canada to ensure it doesn’t rely on foreign-controlled AI models, either by supporting Toronto-based Cohere or pooling compute with other middle powers to develop open-source systems. (The Logic)

Talking point: The report’s authors, Sean Mullin and Jaxson Khan, were key architects of the Liberal government’s $2-billion Sovereign AI Compute Strategy, first unveiled in the 2024 federal budget. “Sovereignty” has taken on considerably more importance since then, as the Trump administration seeks to ensure the U.S. dominates the AI industry and other countries try to avoid relying on technology that either Washington or Beijing could turn off. Canada can’t go it alone, the report says, in part because it lacks homegrown AI chipmaking. The authors say Ottawa should work with allies on hardware and advancing the capabilities of AI models, while defending domestic buying policies in USMCA negotiations.