Canada’s helium reserves could become increasingly valuable as the global race to build artificial intelligence infrastructure accelerates, according to one economist.
While the biggest bottleneck facing new data centres is access to power, the growing demand for semiconductors is creating an opportunity for Canadian helium producers because “there is no substitute” for the gas in semiconductor manufacturing, says Mauri Hall, economist at TD Economics.
Hall notes that 30 per cent of the world’s helium production comes from the Ras Laffan facility in the Middle East.
However, disruptions in the region have taken the facility offline, and repairs are predicted to require a few years, he says.
“This gives Canada an opportunity, or at least Saskatchewan and Alberta, an opportunity to expand their helium production and boost the global footprint of Canadian helium in the world,” says Hall.
AI buildout driving demand
Hall’s comments come as companies continue pouring billions of dollars into artificial intelligence infrastructure, driving demand for semiconductors and the materials needed to manufacture them.
He says helium is essential to the manufacturing of semiconductors, which is a crucial component of the data-centre supply chain.
According to Global Helium, helium acts as a coolant for high-speed computing, which is the foundation of AI development.
Hall says shortages of key commodities are unlikely to stop AI projects from moving ahead because companies have already committed significant capital to the sector.
Instead, the shortages are expected to increase costs, he says.
“The bottlenecks on commodities, I think, will serve to make the build-out more expensive rather than slow it down,” says Hall.
Power remains the biggest hurdle
Although helium presents an opportunity, Hall says the biggest obstacle to expanding AI infrastructure remains electricity.
“The biggest one is power,” Hall says.
“Gaining access to the grid can take longer than expected because of the immense demand for power,” says Hall.
He explains countries such as the U.S., which require large amounts of power to meet their expansion targets, are placing enormous strain on electricity grids and are increasingly turning to alternative energy sources such as natural gas and gas-fired turbines.
However, he notes many of those turbines are sold out until 2028.
“The lead times for power connection, whether you’re looking to do it on grid or off grid, is constraining how fast these data centers can be built and energized,” says Hall.
Canada has room to grow
Hall says the U.S. has by far the largest data-centre expansion projects and capital expenditure, but continues to face supply-chain bottlenecks.
Meanwhile, China is advancing more quickly because it faces fewer regulatory constraints.
“When you have an economy that is top down rather than a political system that’s bottom up, like a democracy. They can bypass regulations more easily,” says Hall.
Canada also plans a massive boost in data centre power. Hall references a document prepared for the federal artificial intelligence minister, which says Canada has roughly 337 megawatts of active AI data center capacity alongside over 20 gigawatts (20,000 megawatts) in projects “under planning or development”.
Hall says although the initial focus had been on low-emission power, that approach has shifted in some regions, particularly Alberta, where a one-gigawatt gas-powered Meta data centre is planned.
“I think Alberta has pulled back some of that specifically for data centers, so as to keep pace with the AI buildout,” says Hall.
Despite the challenges, Hall says Canada is well positioned to expand its own AI infrastructure because of its energy resources.
He points to Alberta, where construction has begun on a large Meta data centre powered by natural gas.
“Because of the natural resources, especially in the prairies, we can have a build out the size of say the meta factory or the meta facility of a gigawatt, which is roughly the amount of power used by the City of Edmonton,” he explains.
“There is certainly room for growth in Canada in the AI sector,” Hall says.
He emphasizes this also won’t put pressure on consumers, as the availability of these natural resources means meeting energy demand in this case does not automatically lead to higher energy costs for consumers.
