Key Takeaways
- The United States is experiencing an unprecedented AI‑driven investment surge: private spending on IT equipment and software accounted for roughly 30 % of U.S. real GDP growth last year, and tech giants plan to commit about US$650 billion in capital spending this year—nearly equal to the entire U.S. private‑sector outlay on non‑residential structures.
- Data‑centre construction in the U.S. now exceeds office‑building construction, with imports of processors and servers from Mexico hitting US$90 billion in 2023 and on track to double again in 2024.
- In contrast, Canada’s data‑centre rollout remains sluggish: only $188 million in building‑permits for “communications buildings” (a proxy for data centres) were issued in the 12 months ending March 2024, down 52 % year‑over‑year, and actual construction has barely begun despite numerous announcements.
- Canadian businesses invest far less in AI‑related technology—estimated to contribute only about 5 % of 2025 economic growth—reflecting a long‑standing underinvestment in information technology that has left Canada a productivity laggard.
- Power access is a critical bottleneck; Alberta’s fully deregulated electricity market gives it an advantage, while the rest of Canada’s heavily regulated grid hampers rapid data‑centre development.
- Public opposition over water and energy use, coupled with low trust in AI, is slowing projects; several high‑profile proposals (e.g., in Manitoba and Ontario) have been rejected or stalled.
- Trade‑data analysis shows that AI‑related imports now represent 23 % of all U.S. goods imports, shrinking the U.S. trade deficit by an estimated US$200 billion; Canada’s import growth in the same category has remained flat, underscoring the divergent AI trajectories of the two economies.
The Scale of the U.S. AI Investment Boom
The artificial intelligence frenzy is reshaping the U.S. economy at a breathtaking pace. Last year, private investment in IT equipment and software drove roughly 30 % of U.S. real GDP growth, according to a Desjardins report. Meanwhile, imports of processors and servers from Mexico have surged to US$90 billion—about double the 2022 level—and are projected to double again in 2024. Four of the largest technology firms alone intend to spend US$650 billion on capital expenditures this year, an amount that rivals the total U.S. private‑sector spending on all non‑residential structures (factories, mines, railroads, warehouses, pipelines) in 2023. This flood of spending is not limited to chips; it encompasses the full ecosystem of AI infrastructure, from data‑centre construction to ancillary electrical and mechanical components.
Data‑Centre Construction Outpacing Office Buildings
A striking symptom of the AI wave is that the United States now spends more to build data centres than to construct office buildings. The U.S. Census Bureau reported that, in April 2024, annualized spending on data‑centre construction exceeded US$50 billion—a 180 % increase over three years. By comparison, investment in traditional office space has stagnated. This shift reflects the massive scale of compute power required to train and run large AI models, which demands sprawling facilities filled with servers, cooling systems, and backup power.
Canada’s Announcements vs. Actual Construction
Canada has not been idle on the publicity front: over the past two years, developers have unveiled scores of new data‑centre projects, ranging from standard facilities to hyperscale campuses intended to support AI workloads. Yet, tangible progress on the ground remains minimal. Statistics Canada’s closest proxy for data‑centre activity—building permits for “communications buildings” (a catch‑all that includes post offices, telephone exchanges, and radio stations)—showed just $188 million in permits issued over the 12 months ending March 2024, a 52 % decline from the previous year. Xavier Lemyre of Statscan’s building‑construction division noted that the current level of activity does not yet merit a separate data‑centre‑specific release, though the agency will monitor for sustained growth.
Underinvestment in AI Technology and Productivity Gaps
Beyond bricks and mortar, Canadian businesses lag in adopting AI‑related computing equipment. Desjardins’ deputy chief economist Randall Bartlett estimated that weaker spending on tech gear in Canada means AI‑related capital could account for only about 5 % of economic growth in 2025, compared with the U.S.’s 30 % contribution. This pattern echoes a decades‑long trend of underinvestment in information technology that has left Canada a productivity laggard relative to its southern neighbour. Sal Guatieri, a senior economist at BMO, emphasized that the real gap lies not in data‑centre construction per se—which contributes a modest share to GDP—but in the broader reluctance of Canadian firms to invest in innovative AI systems, a shortfall that dates back to the late‑1990s tech boom.
Power Access as a Critical Bottleneck
One of the most significant structural hurdles for Canada is access to electricity. Avik Dey, CEO of Capital Power Corp., points out that Alberta’s fully deregulated power market offers a seamless path to securing the large, reliable electricity supplies required by AI data centres. In contrast, the rest of Canada operates under a heavily regulated grid, where securing new power connections can be protracted and costly. Alberta has even set an ambitious target of attracting US$100 billion in data‑centre investment by 2030, leveraging its regulatory advantage to lure developers seeking predictable, market‑based power pricing.
Public Opposition and Environmental Concerns
Even where power is available, projects face mounting resistance. Concerns over water consumption for cooling and the strain on local energy supplies have sparked public backlash in several provinces. In June 2024, Manitoba Premier Wab Kinew cancelled a large‑scale data‑centre proposal for Île‑des Chênes, citing environmental risks and limited economic benefit. Shortly thereafter, Hamilton’s committee of adjustment rejected a plan to partition waterfront land for an AI data centre. These decisions illustrate a growing aversion that is not merely NIMBY‑ism but reflects deeper anxieties about AI’s societal impact, particularly fears of job automation.
Trust Deficit and the Cultural Dimension
Viet Vu of the Dais think‑tank warns against using the U.S. as a blanket benchmark for Canada’s AI aspirations, noting that the American boom is predicated on expectations of future AI‑driven gains that may not materialize. More importantly, global surveys reveal that Canadians trust AI significantly less than the global average. Vu argues that data centres have become physical symbols of the visceral fear that automation will erode employment, and that addressing this trust gap is essential for any meaningful AI adoption. Without a shift in public perception, investment and construction are likely to remain subdued.
Trade Implications: AI‑Related Imports and the U.S. Deficit
A deeper look at trade data underscores the divergence between the two economies. Economist Michael Waugh of the Federal Reserve Bank of Minneapolis found that AI‑related imports—processors, servers, batteries, switchgear, copper wiring, and hundreds of ancillary goods—accounted for 23 % of all U.S. goods imports in 2023. He estimated that, absent the AI boom, the U.S. goods‑trade deficit would have been roughly US$200 billion smaller (about 16 % lower). When The Globe and Mail applied the same methodology to Canadian imports using Waugh’s list of the top 20 AI‑related goods, the result showed stagnant growth: while U.S. imports exploded in early 2024, Canadian import growth remained flat. This contrast highlights how the United States is effectively absorbing a disproportionate share of the global AI supply chain, whereas Canada remains largely on the sidelines.
Outlook and Policy Considerations
The first quarter of 2024 offered a glimmer of improvement for Canada: business spending on computers and software rose nearly 8 % year‑over‑year, the fastest pace since 2022, despite lingering tariff‑related uncertainty. Guatieri suggests that, absent trade‑war headwinds, investment in high‑tech equipment would be even stronger. Nevertheless, the prevailing consensus among experts is that Canada must confront several intertwined challenges—streamlining power‑market regulation, alleviating water‑and‑energy concerns, rebuilding public trust in AI, and fostering a culture of bold corporate investment in emerging technologies—if it hopes to narrow the yawning AI‑driven growth gap with the United States. Without such steps, the country risks continuing to watch the AI‑fuelled economic surge from a distance, benefiting only marginally from the technological tide that is lifting its neighbour.