Il Canada punta sulla sovranità dell’IA con 2 miliardi un fondo sovrano e una strategia ancora da definire

On April 28, 2026, the federal government unveiled the six pillars of Canada's long-awaited national AI strategy — a document that had been promised by the end of 2025, delayed twice, and arrived finally as a companion piece to the Spring Economic Update tabled by Finance Minister François-Philippe Champagne. The pillars were welcomed by some in the tech community and criticized by others as arriving too late and too light on specifics. Both reactions, examined carefully, are correct.

The stakes are real. Canada is not a small player in global artificial intelligence. The country is home to some of the field's foundational researchers — Geoffrey Hinton, Yoshua Bengio, and Richard Sutton all developed core components of modern deep learning at Canadian universities. The Mila institute in Montreal, the Vector Institute in Toronto, and the Alberta Machine Intelligence Institute form a research corridor that countries with vastly larger economies would be grateful to have. Canada was, by reasonable measure, the birthplace of the modern AI era.

The question in 2026 is whether it can translate that intellectual inheritance into economic leadership — and whether it is moving fast enough to do so before the window closes.

What the Federal Government Has Actually Committed

The headline number is $2 billion, committed in Budget 2024 to build and provide access to computing capabilities and technological infrastructure for Canadian AI researchers, start-ups and scale-ups. This investment operates through two mechanisms: an AI Compute Access Fund, which provides subsidised access to computing power for organisations that cannot afford to build their own infrastructure, and the Canadian AI Sovereign Compute Strategy, which is funding the construction of large-scale, Canadian-owned high-performance computing systems.

The rationale is straightforward and broadly accepted across the political spectrum: the country that controls its own AI compute infrastructure controls its own data, its own research agenda, and its own capacity to develop AI applications that serve national interests rather than the interests of American or Chinese technology companies. The American Cloud Act — which allows US courts to compel disclosure of data stored on US company servers, even when those servers are located outside the United States — makes the sovereignty argument not merely theoretical. It is a live legal vulnerability that $925 million earmarked over five years for public digital infrastructure is designed to address.

The Spring Economic Update added another instrument: the Canada Strong Fund, a proposed sovereign wealth fund launching with an initial $25 billion contribution from the federal government, intended to invest in Canadian projects and companies alongside the private sector. Opposition Leader Pierre Poilievre noted, accurately, that sovereign wealth funds are typically operated by countries running surpluses rather than deficits — Canada's projected deficit for 2025-2026 is $66.9 billion, $11.5 billion lower than projected but still substantial. The tension between fiscal prudence and strategic investment is not one the government has fully resolved.

The Six Pillars: What They Say and What They Don't

The six pillars of Canada's AI strategy, as unveiled this week, cover: scaling Canadian AI companies through growth capital and government procurement; building sovereign compute infrastructure; establishing governance and safety frameworks; developing AI talent; driving AI adoption across the economy; and positioning Canada as a global AI partner in democratic AI development.

Grace Lee Reynolds, CEO of the MaRS Discovery District in Toronto — Canada's largest urban innovation hub — called the pillars a signal that the government sees AI as an economic opportunity to lead rather than simply a technology to regulate. "The commitment to scaling Canadian AI companies, through growth capital and government as an anchor customer, is exactly what the startups and scaleups we work with need to grow here and stay here," she said in a statement.

The more sceptical response came from Patrick Searle, CEO of the Council of Canadian Innovators, who said the Spring Economic Update "does little to show that the government is taking the digital economy seriously or using it to strengthen Canada's major traditional economic strategies."

The gap between those two assessments reflects a genuine ambiguity in the strategy as published. The pillars are directionally correct. The specific mechanisms — the procurement rules, the capital deployment timelines, the regulatory framework for AI governance — remain to be detailed in the full strategy document, which has not yet been released.

Cohere and the Homegrown Unicorn Question

One of the most watched questions in Canadian tech in 2026 is when Cohere — the Toronto-based AI company co-founded by University of Toronto graduates and valued at US $6.8 billion after a US $500 million investment in late 2025 — will go public. CEO Aidan Gomez has said "soon" when asked about an IPO timeline, without committing to specifics.

Cohere's significance extends beyond its valuation. It represents Canada's most credible homegrown alternative to the American AI giants — a large language model company built on Canadian talent, backed by Canadian and international capital, with the potential to provide Canadian organisations with AI infrastructure that does not route their data through servers subject to American legal jurisdiction. The cash flow from an IPO, if it materialises, would fund the expansion of data centres and give the Canadian AI ecosystem something it currently lacks: a scaled, publicly traded anchor company that demonstrates the pathway from research excellence to commercial leadership.

Manitoba's announcement this week that it will become the first jurisdiction in Canada to ban youth from AI chatbots — including ChatGPT and Claude — adds a regulatory dimension to the picture. With British Columbia signalling support for similar measures and the federal government saying it is "very seriously" considering age restrictions on AI access, the governance framework for AI in Canada is moving in real time, sometimes faster than the strategy documents that are supposed to define it.

The Talent Problem No Announcement Has Solved

Underlying all of these investments is a challenge that $2 billion in compute infrastructure cannot directly address: Canada is losing AI talent faster than it is producing it. Tech job postings on Indeed Canada were down 19 per cent in early 2026 compared to 2020 levels. More than 150,000 tech employees were laid off globally in 2024 across 551 companies, and Canada was not insulated — there were cuts at Cisco, at Oracle's Kitchener office, and at blockchain firm Consensys.

The structural challenge is that AI is replacing the junior and mid-level technical roles that used to be the entry point into the industry, while the senior roles that remain require expertise that takes years to develop. The pipeline of talent that feeds those senior roles depends on entry-level positions that are contracting. Over 200 major Canadian companies now employ a chief AI officer, according to a recent Amazon Web Services survey. The people qualified to fill those roles are in high demand globally, and Canada's ability to retain them depends on offering not just research infrastructure but the commercial ecosystem — the funded companies, the competitive salaries, the career trajectories — that makes staying in Canada the rational choice.

The six pillars acknowledge this. The mechanisms to address it remain, as of this week, aspirational.