The world’s largest technology companies will spend more than $600 billion on AI infrastructure in 2026. Amazon, Microsoft, Google and Meta will each cross the $100-billion mark in annual capital spending this year, nearly six times what these companies spent in 2022, the year before ChatGPT flipped the tech industry table.
All of that money needs to go somewhere. It needs land, power and stability. And right now, a surprising number of those cheques are being written with Canada’s name on them.
This country is making a credible play to become the backbone of the West’s AI infrastructure. It has abundant clean energy, a cold climate, political stability and the rule of law. Canada looks increasingly like an attractive solution for hyperscalers scrambling to find places where they can plug in.
Globally, the AI arms race has created an infrastructure crisis. Every major cloud provider wants to build as fast as possible, but the single biggest bottleneck is electricity.
Data centres follow a simple rule: if you can power it, they will come. The big hyperscalers prioritize sites that can deliver large, continuous megawatts quickly, with fibre connectivity and clear permitting processes.
In the United States, community opposition has blocked or delayed an estimated $98 billion worth of data centre projects. Grids in multiple states are straining under the load while residents are pushing back. Ireland was an early mover in the space but today it watches as data centres consume more than 20 per cent of the country’s electricity.
Then there’s Canada.
Right now there are 337 data centres operating in Canada, with primary hubs in Toronto, Montreal, Calgary and Vancouver. More than 60 new facilities sit in various stages of development. That represents over two gigawatts of planned capacity. The country’s data centre market is on pace to double from $21 billion to $43 billion by 2030.
The projects pouring in to Canada are enormous. In Toronto, Cohere and CoreWeave are building what will become Canada’s largest AI facility, backed by $240 million in federal funding and powered by NVIDIA GPUs. In Alberta, Synapse Data Centre announced a $10-billion, one gigawatt campus in the town of Olds, which would be 10 times large than any existing data centre in the country.
British Columbia is making moves as well. Bell launched its AI Fabric initiative to build 500 megawatts of hydro-powered AI compute across six facilities in the BC interior. Telus has partnered with NVIDIA to build a sovereign AI factory in Rimousky, Que., powered by hydroelectricity. Calgary now holds more than 25 per cent of the nation’s future data centre power capacity. The map is filling out fast.
The reasons Canada is attractive to hyperscalers looking for a home for their data centres are straightforward. Canada has abundant clean hydroelectricity capacity. Nealry 70 per cent of the nation’s electricity is generated by hydro, making it one of the cleanest in the world. Quebec alone generats more than 99 per cent of its power from renewables. British Columbia is close behind at 98 per cent. This is massive selling point for AI companies with environmental awareness written into their operating plans.
Canada also has an idea climate for data centres. The country is cold for half the year. This is a legitimate competitive advantage, because cold air reduces the need for energy-intensive cooling systems, which means lower operating costs and better energy efficiency.
Add in a highly educated workforce, political stability, strong rule of law and an orderly society, and Canada becomes a very attractive investment. Montreal, Toronto and Vancouver are home to some of the most respected AI research institutions on the planet, meaning the people who build and train these models already live here.
The federal government sees the potential. Ottawa launched its Sovereign AI Compute Strategy in 2024 and commited $2 billion over five years. The government has open calls for proposals seeking data centre projects exceeding 100 megawatts, with selected proponents entering memoranda of understanding with the federal government. More tellingly, the Canada Infrastructure Bank was given the green light to invest direclty in AI infrastructure projects.
There are some regional catches, however. Alberta’s electric system operator capped large load data centre connections at 1,200 megawatts through 2028, and all of that capacity was snapped up almost immediately. Quebec, despite massive hydro reserves, just proposed doubling electricity rates for large data centres to $0.13 per kilowatt-hour. The province wants to make sure the rest of the population does not end up subsidizing tech giants. It’s a fair concern, but it make the pitch a little less attractive.
Community opposition in Canada is as real as in the United States. Residents in Olds, Alta., learned about the massive Synapse proposal in January and immediately raised concerns about water use and noise polution. A Kineticor proposal in Rocky View County also faced pushback. These are legitimate concerns about the impact of industrial-scale facilities on small communities.
Another concern is the lack of jobs these centres provide. Once the construction is completed the average data centre runs with very little need for human oversight. A few people for security and maintenance is all that is needed, so these projects don’t bring many jobs or other boosts to local economies.
Still, Canada has something genuinely rare in the global AI race: clean power, political stability, cold air, talent and room to grow. Very few countries can offer all of these at once.