Canada Invented Modern AI and Watched Others Commercialise It

Canada has a strong claim to being where modern artificial intelligence was developed, and an equally strong record of watching the economic value migrate south.

The research origins:

Canadian institutions sustained neural network research through decades when it was academically unfashionable, funded consistently by national research programmes when few others would.

That patience produced results. Researchers based in Toronto and Montreal were central to the deep learning breakthroughs that made modern AI possible, and the institutions they built, including major AI research institutes in Toronto, Montreal and Edmonton, remain internationally significant.

Montreal in particular developed an unusually dense academic AI community, and Edmonton became a centre for reinforcement learning research.

Canada was also the first country to publish a national AI strategy, focused on research capacity, talent and institutes rather than regulation or industrial policy.

The commercialisation gap:

The persistent Canadian complaint is that research excellence has not translated proportionately into Canadian companies. Major international AI labs recruited heavily from Canadian institutions, and while several opened Canadian offices, retaining research locally, the resulting products and profits are booked elsewhere.

Canadian AI startups exist and some have scaled significantly, particularly in enterprise language technology, but the pattern of promising companies being acquired by American buyers is well established.

The structural causes are familiar: shallower venture capital, smaller domestic market, proximity to a much larger economy that pays more, and limited public procurement that might anchor domestic firms.

Where AI is applied:

Financial services are the largest commercial adopter, with major banks running substantial AI operations for risk, fraud and customer service.

Natural resources matter: mining, forestry and energy apply machine learning to exploration, operations and predictive maintenance, and the oil sands industry is a significant user given the scale and cost of its operations.

Agriculture across the prairies uses precision techniques at large scale, and Canadian agricultural technology has genuine strength.

Healthcare research benefits from single-payer systems that generate population-level data, subject to provincial privacy rules that vary.

Infrastructure:

Canada has real advantages for data centres: cold climate reducing cooling costs, abundant hydroelectric generation in several provinces, notably Quebec, British Columbia and Manitoba, and political stability.

Quebec in particular has attracted data centre investment on the strength of cheap hydropower, though the province has at times restricted new energy-intensive connections as demand grew.

Constraints:

Talent retention is the defining issue. Salary differentials with the United States are large, and immigration to the US is straightforward for Canadian professionals.

Capital for scaling companies is limited relative to need.

Regulation has been slower to arrive than in Europe, with federal AI legislation debated at length without swift resolution, leaving businesses with less certainty than they would prefer.

The outlook:

Canada's research position remains genuinely strong and its institutes continue to produce influential work. Its infrastructure advantages, cold weather and clean power, are becoming more valuable as compute demand grows.

The unresolved question is the same one it has faced for a decade: whether the country can convert world-class research into Canadian-owned industry, or whether it continues to function as a highly effective research and development annexe for companies headquartered elsewhere.