Evaluating the Productivity Divergence in Artificial Intelligence Between U.S. and Canadian Markets
The frantic speed with which the artificial intelligence bonanza is rewiring the United States economy is staggering.
The U.S. now spends more to construct data centres than office buildings. Having roughly doubled imports of processors and servers from Mexico last year to US$90-billion, the U.S. is on track to double that number again this year.
Meanwhile, four tech giants alone plan to commit US$650-billion this year in capital spending. That’s almost as much as the U.S. private sector spent on all nonresidential structures last year, including factories, mines, railroads, warehouses and pipelines.
Read More On Our Daily Stock Market Reports – Geopolitical Relief Fuels Broad-Based Market Rally
The manic rollout of AI-related infrastructure and systems is juicing economic growth south of the border in profound ways. In Canada, not so much.
While developers here have announced scores of new data centre projects – both standard facilities and large scale operations necessary to power AI – over the last two years, sparking a growing public backlash over water and energy-use fears, very few construction projects have gotten under way.
Likewise, workers and businesses in Canada have been much slower to adopt AI or invest in computing technology, continuing a decades-long pattern of underinvestment that has left Canada a productivity laggard. This is all reflected on the trade front, with Canada largely sitting out the rush to secure AI-related equipment from overseas.
Last year roughly 30 per cent of U.S. real GDP growth stemmed from private investment in IT equipment and software, according to a recent report by Desjardins’ deputy chief economist Randall Bartlett. By contrast, he estimated weaker spending on tech gear by businesses in Canada meant little more than 5 per cent of economic growth in 2025 could be attributed to AI-related capital spending.
“While tech-related investment did pick up in Canada toward the end of last year, there is a lot of ground to make up to close the gap with the U.S.,” he wrote.
In April, U.S. companies spent more than US$50-billion on an annualized basis on data centre construction, a 180-per-cent increase in three years, according to the U.S. Census Bureau.
Statistics Canada has no similar detailed measure, with investment in data centre construction mixed in with all spending on transportation and utilities. The closest the official stats get to measuring data centre construction activity is in the number and value of building permits for what Statscan calls “communications buildings,” a grab bag of bygone structures such as post offices, telephone exchanges and radio stations into which data centres get lumped.
The tally is sobering. During the last 12 months, as of March, municipalities issued just $188-million worth of permits for all types of communication buildings, and that was down 52 per cent from the previous 12-month period.
Xavier Lemyre, an analyst with Statscan’s building construction division, said there simply isn’t enough activity to warrant separate data centre building permit and construction investment releases yet. That could change if activity picks up, but he said there is no threshold for when that might occur.
“We keep a close eye on trends, and if the number of permits received increases steadily over the coming months, it’s something we would consider,” he said.
Avik Dey has a front-row seat to Canada’s sluggish data centre rollout as the president and chief executive of Capital Power Corp., an Edmonton-based independent power producer. Since 2023, the company has been advocating for its Genesee Generating Station, a thermal power facility near Edmonton, to attract a large-scale AI data centre to the site.
Mr. Dey said Canada’s smaller size and the difficulty of securing semiconductors partly explains why the country is lagging, but the larger “bottleneck” for AI is access to power in Canada’s heavily regulated electricity market.
Since Alberta is the only fully deregulated power market, “Alberta has the advantage. The path to power is seamless, and there’s a path to power that actually benefits consumers and ratepayers here,” he said, pointing to the province’s ambitious goal of attracting $100-billion of data centre investment by 2030. “I believe the process will be successful, it’s just taking a longer time to get there.”
Still, even in Alberta, developers face mounting opposition to their data centre plans. Meanwhile, last week, two planned projects were thwarted shortly after the Prime Minister Mark Carney released his government’s AI strategy aimed at boosting its adoption.
On June 4, Manitoba Premier Wab Kinew nixed a large-scale data centre planned for Île-des Chênes, southeast of Winnipeg, saying, “There’s a big threat to the environment and not much benefit to the economy.” Then in Ontario, Hamilton’s committee of adjustment rejected a plan to partition land on the city’s waterfront for an AI data centre.
Canada isn’t sitting out the data centre building boom entirely. But for all the hype and attention the myriad of new construction announcements receive, that’s not yet translated into many shovels in the ground, according to Vancouver-based Aterio, which tracks the growth of the data centre industry.
Canada needs to be realistic about what its data centre goals are, said Viet Vu, manager of economic research at the Dais, a public policy think-tank at Toronto Metropolitan University, who cautioned against “using the U.S. as a yardstick to compare us against on AI.”
That’s because the frothy rollout south of the border is banking entirely on future gains from AI that may not unfold as hoped, which could leave local governments and electricity rate payers exposed if projects fail.
A bigger concern, he said, is the growing antipathy toward AI in Canada, with global surveys finding Canadians’ trust in the technology is significantly below the global average.
“Data centres are the physical embodiment of the visceral fear employees and people have about automating their jobs away in the long run,” he said. “For Canadians to embrace AI, that fear needs to be tackled.”
For Sal Guatieri, a senior economist at Bank of Montreal, the concerning AI gap between Canada and the U.S. isn’t so much data centre construction, which he said makes a relatively small contribution to GDP growth. It’s the yawning disconnect between business investment in innovative technologies between the two countries.
“We’re not investing as heavily in AI systems as U.S. companies are and that underinvestment in information technologies has been a longstanding issue for Canada all the way back to the tech boom of the late 1990s and beyond,” said Mr. Guatieri.
The investment gap has widened dramatically since 2023, with spending on computer equipment surging as a share of GDP in the U.S., while nudging only modestly higher in Canada.
It’s likely too soon to tell whether higher U.S. investment levels related to AI are behind that country’s productivity boom. Growth in U.S. productivity has been rising at its fastest pace in two decades, but that’s been the case for five years, well before AI hysteria began.
Still, given Canada’s poor track record, the widening gulf in business investment doesn’t bode well.
The first quarter of the year did hint at improvement for Canada, said Mr. Guatieri. Business spending on computers and software combined climbed by nearly 8 per cent from the year before, the fastest pace since 2022, despite the uncertainty from tariffs that is leading chief financial officers to keep a tight grip on spending.
“Investment in high-tech equipment would likely be even stronger if not for the trade war,” he said. “The AI boom is adding some resilience to the U.S. expansion but providing only a very mild tailwind here in Canada, though certainly any tailwind is welcome.”
As much as intense AI capital spending is pushing U.S. economic activity higher, the country heavily relies on others to supply most of the computer processors and servers for all those data centres, and that is creating trade distortions that complicate the U.S. growth picture.
But powering up an AI data centre and keeping it running requires more than chips. And it’s here where another stark economic gap between the U.S. and Canada lies.
In a recent study, Michael Waugh, an economist and monetary adviser at the Federal Reserve Bank of Minneapolis, used AI to analyze the true scale of the computer equipment, batteries, electrical switchgear, copper wiring and hundreds of other products required for the U.S.’s AI rollout.
Taken together, those products amounted to 23 per cent of all U.S. imports last year, he found. Absent the AI boom, he estimated, the U.S. goods trade deficit would have been roughly smaller by US$200-billion, or 16 per cent.
When The Globe and Mail applied the same screen to Canadian imports, using a list of the 20 largest AI imports published by Mr. Waugh, the result was dramatically different. While Canadian and U.S. imports of AI-related equipment grew at roughly the same pace for most of the last decade, in early 2024, U.S. imports exploded, while Canadian import growth stayed the same. (Mr. Waugh reviewed and supported The Globe’s analysis.)
It’s another reminder that when it comes to AI’s effect on the economy, the U.S. is in a world all its own.
With a report from Dexter McMillan
This article was first reported by The Globe and Mail
