The race for AI dominance isn’t just about code; it’s about processing power. As the ambition of the industry soars, there’s never been more demand for the infrastructure to support it.

As a result, compute has become the hot commodity in AI. It’s why tech giants and asset managers are proposing and building data centres that consume as much power as entire neighbourhoods. It’s also why companies like Nvidia, whose GPUs are used to help run and train AI models, are soaring in value.

In Canada and across the world, governments and businesses are trying to grab a piece of the compute build out, which has become a shorthand for the wider competition to own AI.

That need for processing power stretches from academic researchers and the earliest of early-stage startups to tech giants and governments. And if you read the headlines, supply is limited and demand is soaring.

But not every company needs the same type of compute, or the same amount. And infrastructure providers have different visions of how to deliver it, from mammoth data centres proposed in Alberta to upstart neoclouds selling AI services. Then there’s the question of sovereignty, a term seemingly everyone’s using differently. 

As Canada tries to secure its place in the AI race, how much compute capacity does it need, and can it build it? And why should people and businesses care about sovereignty? Watch the video below for an in-depth discussion tackling all the big questions.