The government is about to drop a lot of cash on AI infrastructure, but there’s no guarantee that it will all stay in Canada.
Driving the news: The federal government opened consultations on how to spend $2 billion earmarked to improve access to chips, data centres, storage, and networking needed to train and run AI — computational resources that are collectively known as ‘compute.’
Why it matters: Canada is falling behind on compute. This makes it harder for researchers and companies to develop cutting-edge AI, which also leads to brain drain as ambitious talent heads to countries with the computing power needed to develop their ideas.
- Canada ranked 23rd in AI infrastructure in the most recent Tortoise Global AI Index, while the country’s two largest supercomputers rank 123rd and 124th in the world by computing power, according to the most recent rankings.
What they’re saying: While major cloud providers like Amazon, Google, and Microsoft might see an opportunity to expand their data centre operations in Canada, Bob Beachler, VP of product at chip maker Untether AI, says the government’s priority should be spurring domestic development of the many things that go into AI compute.
-
That’s one reason cited for Canada falling behind on AI development: The feds prioritized luring multinationals here over supporting home-grown options.
- Critics have pointed to the fact that Big Tech has been able to take advantage of subsidies as a problem with other innovation programs, like the Scientific Research and Experimental Development tax credit or the $4 billion Canada Digital Adoption Program.
What’s next: There could be room for a balanced approach. Ben Davies, CIO of the Vector Institute, says Ottawa should immediately increase access to compute, then turn attention to a long-term strategy for sovereign compute — infrastructure owned by a country to ensure researchers and businesses can access it — to ensure Canada meets its economic goals.