Canada’s reliance on foreign cloud providers and advanced computing hardware leaves its artificial intelligence sector exposed to geopolitical risks, according to a new University of Toronto report released March 10.
The assessment, announced in a Munk School of Global Affairs and Public Policy news release, identifies those two areas as the most critical vulnerabilities in the infrastructure supporting Canadian AI. It also raises concerns about whether data centres located in Canada are governed solely by Canadian laws.
Written by Munk School senior fellows Sean Mullin and Jaxson Khan, Sovereign by Design: Strategic Options for Canadian AI Sovereignty examines seven layers of AI technology against five dimensions of digital sovereignty.
The authors argue that protecting Canadian control over AI does not require withdrawing from international markets. Instead, their framework focuses on reducing exposure to foreign pressure while maintaining access to global technology and preserving Canadian legal authority over data and infrastructure.
Cloud access and hardware supply pose risks
According to the report, Canada depends heavily on foreign providers for the advanced computing capacity needed for AI. Many data centres operating within Canada may also fall under other countries’ laws, creating potential geopolitical exposure.
The report separately identifies a high concentration of suppliers in computing hardware. The authors say the resulting technological dependencies are too extensive for any one country to address on its own.
Rather than recommending one universal solution, the assessment sets out policy options that vary according to the sensitivity of data, costs and the institutions involved.
“The debate about AI sovereignty too often gets stuck between two extremes: build and buy everything domestically or accept total dependence on foreign providers,” Khan said in the news release.
Canadian strengths offer options
The report also identifies assets Canada can draw on, including clean energy that attracts data centre investment, leading AI research talent and a growing group of Canadian-owned infrastructure providers.
It notes that Canada has one of the few commercial foundation model companies outside the United States and China. Open-source models and cooperation among like-minded middle powers are presented as other ways to reduce dependence while retaining technological capability.
The authors place their assessment against a backdrop of hundreds of billions of dollars in global AI investment expected over the next several years. They argue that decisions about computing infrastructure made now will have lasting consequences for the technology sector.
Produced through the Munk School’s AI Competitiveness Project, the research draws on interviews with senior government officials, industry leaders and international experts, alongside analysis of international approaches to digital sovereignty. The school has made the report and an accompanying policy briefing note available online.
