Our CEO joined provincial leaders on stage at the Global Energy Show in Calgary. Here’s what the conversation means for the enterprises actually making AI and infrastructure decisions, from the people who run the infrastructure.
At the Global Energy Show in Calgary last week, our CEO James Beer joined a panel on the future of AI infrastructure in Canada. He was the only data centre operator on stage, alongside Saskatchewan Premier Scott Moe, Alberta Minister of Technology and Innovation Nate Glubish, data centre developer Jeffrey Gossain (managing director, Canada, for Norwegian-backed EW Bolora), and Aric Carlisle (director of engineering and project development for data centres at SLB). The session was moderated by Wish Bakshi of AQ Energy.
The headline was clear: Canada has a narrow window to lead in AI infrastructure, and its energy advantages are why it can. But the more useful question for most organizations isn’t whether Canada wins the race. It’s what the race means for the decisions you’re making right now.
The backdrop, briefly
AI runs on electricity, and the demand is staggering: estimates on stage put new North American load growth near 100 gigawatts, roughly a hundred mid-sized cities’ worth, with tens of billions of dollars in chips per gigawatt. Crucially, it’s funded by the cash flows of the world’s largest tech companies, not speculative debt. The customers and the capital already exist. Power is the constraint.
That’s Western Canada’s opening. Cheap, abundant natural gas, a cold climate that cuts cooling costs, and regulators willing to move fast. Premier Moe credited Saskatchewan’s wins to keeping projects moving “at the pace of business as opposed to the pace of government.” Glubish signalled Alberta’s own gigawatt-scale projects are close: “shovels in the ground, permits complete, financing in place,” with the first campus 12 to 18 months out. Neither province sees it as a rivalry. “There’s so much demand, there’s enough to go around,” Glubish said.
The real prize: sovereign compute for Canadians
The idea the panel kept returning to wasn’t capacity. It was sovereignty. Both leaders were explicit that landing global hyperscalers, while welcome, is not the goal in itself.
“We talked a lot about the hyperscalers, but let’s not lose sight,” Glubish said. “The home run for Canada, Alberta, and Saskatchewan is not just to bring hyperscalers here. It’s to get sovereign compute, to ensure that we have digital sovereignty.” The hyperscalers, in his words, are “just the domino to get the trend going.” What matters more is building an ecosystem “where the hardware, the software, the fibre, the electricity, the entire stack is controlled by Canadian innovators, entrepreneurs, and investors,” so Canadian data and intellectual property stay protected and Canadian companies can compete globally. Moe framed it as a duty: provinces “have a responsibility to have a very mature conversation about how we do this across Canada.”
Asked what sovereign compute actually means, given that he runs data centres for a living, James kept it simple: “Canadian companies, full stop. Canadian companies, Canadian operators, Canadian management teams, Canadian networks.” In other words, sovereignty isn’t a label. It’s Canadian organizations running their most important workloads on infrastructure that is genuinely Canadian in its people, operations, and networks.
What this means for you
Here’s the part the recap doesn’t cover, and where we’ll offer an operator’s point of view.
The compute you’ll actually use is moving closer to you, not further away. Most coverage fixates on remote gigawatt mega-campuses. They’re real and they matter. But James’s argument on stage was that the current build pattern is “unsustainable” on its own, and that inference and training workloads will converge. As AI shifts from training huge models toward running them, often in agentic applications that operate for hours, latency and proximity start to matter again. That favours well-located, urban and regional data centres “closer to the users, closer to the eyeballs.” For most enterprises, the relevant AI infrastructure won’t be a distant campus you never see. It’ll be capacity near your operations and integrated with your existing systems.
Your own teams are about to need this. Every customer we talk to is working out how to bring AI into their practice, which increasingly means bringing chips into their own environments. The practical question for most organizations over the next 18 months isn’t “should we build a campus.” It’s “is our infrastructure ready to host the AI workloads our teams will want, and where should those workloads live for the latency, data residency, and integration we need.”
Timelines will tighten before they loosen. The panel’s most sobering note came from Aric Carlisle, who warned that Western Canada already faces an electrician shortage before data centre demand is even counted. “If we don’t do that,” he said of training thousands of new tradespeople now, “we run the risk of collapsing this data centre infrastructure buildout,” while driving up costs and delaying schedules in other industries too. The takeaway for anyone planning power-dense capacity: lead times on construction and skilled trades will stretch. If you’ll need capacity in two or three years, the planning conversation is a now conversation, not a later one.
Sovereignty is becoming a real, available option. With both governments actively prioritizing Canadian-controlled compute, genuinely Canadian infrastructure for sensitive or regulated workloads is more available than it was even a year ago. If data residency matters to your organization, it’s worth weighting in your decisions rather than treating it as a nice-to-have.
Four questions worth asking yourself
If you take one thing from the panel, make it these:
- Where will our AI inference actually run, and how close does it need to be to our users and data?
- Is our current infrastructure ready to host the chips our own teams will soon want to deploy?
- What are our data residency and sovereignty requirements, and does our infrastructure genuinely meet them?
- Given tightening power and skilled-trades capacity, is our timeline realistic, and have we started planning early enough?
Why we were in the room
We don’t build gigawatt mega-campuses, and we’re not trying to. Qu’s focus is the part of this story the panel kept circling back to: Canadian operations with Candian teams, located close to the businesses that use it, built to bring AI into real enterprise environments. As Premier Moe put it, “we are truly in a moment in Canada, and we should not let it pass us by.” We agree, and we’re building for it.
If you’re working through any of the four questions above for your own organization, that’s exactly the conversation we like to have. Reach out to our team.
.jpg?length=800&name=Image%20(34).jpg)
.jpg?length=400&name=Image%20(34).jpg)
Our CEO joined provincial leaders on stage at the Global Energy Show in Calgary. Here’s what the conversation means for the enterprises actually making AI and infrastructure decisions, from the people who run the infrastructure.
At the Global Energy Show in Calgary last week, our CEO James Beer joined a panel on the future of AI infrastructure in Canada. He was the only data centre operator on stage, alongside Saskatchewan Premier Scott Moe, Alberta Minister of Technology and Innovation Nate Glubish, data centre developer Jeffrey Gossain (managing director, Canada, for Norwegian-backed EW Bolora), and Aric Carlisle (director of engineering and project development for data centres at SLB). The session was moderated by Wish Bakshi of AQ Energy.
The headline was clear: Canada has a narrow window to lead in AI infrastructure, and its energy advantages are why it can. But the more useful question for most organizations isn’t whether Canada wins the race. It’s what the race means for the decisions you’re making right now.
The backdrop, briefly
AI runs on electricity, and the demand is staggering: estimates on stage put new North American load growth near 100 gigawatts, roughly a hundred mid-sized cities’ worth, with tens of billions of dollars in chips per gigawatt. Crucially, it’s funded by the cash flows of the world’s largest tech companies, not speculative debt. The customers and the capital already exist. Power is the constraint.
That’s Western Canada’s opening. Cheap, abundant natural gas, a cold climate that cuts cooling costs, and regulators willing to move fast. Premier Moe credited Saskatchewan’s wins to keeping projects moving “at the pace of business as opposed to the pace of government.” Glubish signalled Alberta’s own gigawatt-scale projects are close: “shovels in the ground, permits complete, financing in place,” with the first campus 12 to 18 months out. Neither province sees it as a rivalry. “There’s so much demand, there’s enough to go around,” Glubish said.
The real prize: sovereign compute for Canadians
The idea the panel kept returning to wasn’t capacity. It was sovereignty. Both leaders were explicit that landing global hyperscalers, while welcome, is not the goal in itself.
“We talked a lot about the hyperscalers, but let’s not lose sight,” Glubish said. “The home run for Canada, Alberta, and Saskatchewan is not just to bring hyperscalers here. It’s to get sovereign compute, to ensure that we have digital sovereignty.” The hyperscalers, in his words, are “just the domino to get the trend going.” What matters more is building an ecosystem “where the hardware, the software, the fibre, the electricity, the entire stack is controlled by Canadian innovators, entrepreneurs, and investors,” so Canadian data and intellectual property stay protected and Canadian companies can compete globally. Moe framed it as a duty: provinces “have a responsibility to have a very mature conversation about how we do this across Canada.”
Asked what sovereign compute actually means, given that he runs data centres for a living, James kept it simple: “Canadian companies, full stop. Canadian companies, Canadian operators, Canadian management teams, Canadian networks.” In other words, sovereignty isn’t a label. It’s Canadian organizations running their most important workloads on infrastructure that is genuinely Canadian in its people, operations, and networks.
What this means for you
Here’s the part the recap doesn’t cover, and where we’ll offer an operator’s point of view.
The compute you’ll actually use is moving closer to you, not further away. Most coverage fixates on remote gigawatt mega-campuses. They’re real and they matter. But James’s argument on stage was that the current build pattern is “unsustainable” on its own, and that inference and training workloads will converge. As AI shifts from training huge models toward running them, often in agentic applications that operate for hours, latency and proximity start to matter again. That favours well-located, urban and regional data centres “closer to the users, closer to the eyeballs.” For most enterprises, the relevant AI infrastructure won’t be a distant campus you never see. It’ll be capacity near your operations and integrated with your existing systems.
Your own teams are about to need this. Every customer we talk to is working out how to bring AI into their practice, which increasingly means bringing chips into their own environments. The practical question for most organizations over the next 18 months isn’t “should we build a campus.” It’s “is our infrastructure ready to host the AI workloads our teams will want, and where should those workloads live for the latency, data residency, and integration we need.”
Timelines will tighten before they loosen. The panel’s most sobering note came from Aric Carlisle, who warned that Western Canada already faces an electrician shortage before data centre demand is even counted. “If we don’t do that,” he said of training thousands of new tradespeople now, “we run the risk of collapsing this data centre infrastructure buildout,” while driving up costs and delaying schedules in other industries too. The takeaway for anyone planning power-dense capacity: lead times on construction and skilled trades will stretch. If you’ll need capacity in two or three years, the planning conversation is a now conversation, not a later one.
Sovereignty is becoming a real, available option. With both governments actively prioritizing Canadian-controlled compute, genuinely Canadian infrastructure for sensitive or regulated workloads is more available than it was even a year ago. If data residency matters to your organization, it’s worth weighting in your decisions rather than treating it as a nice-to-have.
Four questions worth asking yourself
If you take one thing from the panel, make it these:
Why we were in the room
We don’t build gigawatt mega-campuses, and we’re not trying to. Qu’s focus is the part of this story the panel kept circling back to: Canadian operations with Candian teams, located close to the businesses that use it, built to bring AI into real enterprise environments. As Premier Moe put it, “we are truly in a moment in Canada, and we should not let it pass us by.” We agree, and we’re building for it.
If you’re working through any of the four questions above for your own organization, that’s exactly the conversation we like to have. Reach out to our team.
Paul M
Paul Miedzik is Senior Manager of Marketing at Qu Data Centres, with extensive experience in enterprise cloud and digital infrastructure across the Canadian tech sector.