Power and Placements: Can Canada Support OpenAI's Infrastructure Ambitions?
AI-generated image · Bay Street Wire
OpenAI managing director George Osborne eyes Canada for potential data center growth, but the hardware reality depends on energy and stability.
Q: What is OpenAI's current stance on developing data center infrastructure in Canada?
A: As BetaKit first reported, OpenAI managing director George Osborne stated during a conversation with Bloomberg AI reporter Rachel Metz at the ALL IN conference in Montréal that Canada is an "obvious place to look" for potential projects. While Osborne clarified that the company has not made any final decisions, he noted that Canada is a stable jurisdiction with a "can-do attitude" and the necessary energy resources.
Q: What specific factors make Canada attractive for the massive thermal and power loads required by next-gen AI clusters?
A: BetaKit reports that Aidan Gomez, the CEO of Cohere, highlighted three primary physical advantages: Canada's landmass, its climate, and its energy availability. These factors contribute to Gomez's view that Canada has the potential to evolve into a "superpower" in the data center sector.
Q: Is there existing evidence of large-scale compute capacity being deployed in the country?
A: Yes. BetaKit reports that Bell announced plans on Monday to quadruple the intended compute capacity of a data center development located in Saskatchewan. This project, valued at approximately $50 billion, is positioned to become the largest data center in Canada.
Q: How is OpenAI engaging with the Canadian government to facilitate these developments?
A: BetaKit notes that OpenAI has signed onto a non-binding data centre development framework established by the federal government, joining nearly two dozen other hyperscalers. Additionally, Osborne praised the National AI Strategy’s focus on adoption and the efforts of Prime Minister Mark Carney to attract international investment.
Q: What are the primary risks or requirements OpenAI is considering before committing to these costly builds?
A: Because data centers are expensive, long-term investments, Osborne told Bloomberg AI reporter Rachel Metz that OpenAI requires placement in jurisdictions that are stable and offer a predictable future.