Artificial intelligence industry in Canada

• 1. Introduction

  ›

  Canada's artificial intelligence industry begins, for many listeners, with a phrase spoken in Montreal. At the All In AI conference in September 2025, Minister of Artificial Intelligence and Digital Innovation Evan Solomon called building digital sovereignty the most pressing democratic issue of the time. He also introduced a task force of 26 people, charged with updating the country's AI strategy. The urgency makes more sense once you notice the odd position Canada occupies. Its universities helped produce ideas that reshaped the field, and its researchers have won some of science's highest honours. Yet turning that scientific strength into everyday business use has been harder than anyone hoped. So how did one country become a source of breakthroughs and a cautious customer at the same time? Who built the research centres that anchor the field, and what are their tools doing in hospitals, mines and Inuit communities? And what rules, money and machinery will decide whether Canada keeps its place?

  33302

• 2. The Hesitant Adopters

  ›

  By June 2024, only 12% of Canadian firms were using AI in their products or services. That was the lowest share among OECD countries. The trend has since bent upward. Between mid-2024 and mid-2025 the rate doubled, rising from 6.1% to 12.2%. Statistics Canada's September 2025 figures show a divided business community. About one-third of firms said they had no plans to adopt AI over the coming year, while 14.5% intended to start using it to make goods or deliver services. Firms that held back gave three main reasons: the technology seemed irrelevant to them, they lacked knowledge, or they worried about privacy. The consulting firm PwC estimates that Canada moves at roughly three-quarters of the American pace. Geoffrey Hinton, the British-Canadian computer scientist, put the risk bluntly in 2025. He warned that slow uptake among Canadian companies could cost the country the early lead it once enjoyed. The question is where that lead came from, and the answer starts in a Toronto lab.

  17303135

• 3. The Toronto Breakthrough

  ›

  In 2012, a neural network called AlexNet entered the ImageNet Large Scale Visual Recognition Challenge and sharply cut the error rate. It was built at the University of Toronto by Alex Krizhevsky, Ilya Sutskever and Geoffrey Hinton. AlexNet is a deep convolutional neural network, and it showed what deep learning could do when paired with GPU acceleration. Its result helped establish Canada's reputation for AI leadership and pushed the wider technology sector to take up deep learning. The design later found a second life in medicine. Adapted versions help analyze radiographs, mammograms and other scans, flagging abnormalities and supporting clinical diagnosis. Hinton's decades of work on these foundations earned him the Nobel Prize in physics in 2024. He was not the only Canadian whose career pointed in this direction, and one of the others had started building an institution decades earlier.

  26282320

• 4. Institutes and Experts

  ›

  Yoshua Bengio founded the group that would become Mila in 1993, long before deep learning was fashionable. He won the Turing Award in 2018 for his pioneering work in the area. Today Mila is a collaboration among four academic partners in Montreal. It forms one leg of the Pan-Canadian Artificial Intelligence Strategy, alongside Amii in Alberta and the Vector Institute in Toronto. Three institutes in three cities give the national effort its structure. Fakhreddine Karray represents a more applied strand of Canadian expertise. His work on operational AI has had a practical effect in intelligent transportation systems, virtual healthcare and driver safety. Research of that kind has to leave the laboratory to matter. Two Canadian tools show how that can look, one built to read crowds and the other to watch over hospital patients.

  23123

• 5. Prediction and Patient Care

  ›

  Polly, an AI system from the Ottawa start-up Advanced Symbolics Inc., took its name from its first assignment, which was to analyze the politics of Brexit. Work began in 2015. Polly was designed to analyze and anticipate how target audiences behave, so that organizations could shape their communication and advertising. In 2016 it drew wide attention for correctly forecasting both the Brexit referendum and the American presidential election won by Donald Trump. The company says clients now span healthcare, politics, entertainment and mental health research. Chartwatch works in a very different setting. This Canadian tool studies patient data to catch subtle signs that someone's condition is worsening, so that care teams can step in sooner. A 2024 study found it reduced unexpected hospital deaths by 26%. Hospitals are not the only places where Canadian AI handles life-and-death stakes, because another system now directs enormous trucks across an oil sands mine.

  152534

• 6. Machines in the Oil Sands

  ›

  A 2020 Ernst & Young report described how Canada's oil and gas industry was putting AI to work. It listed routine, repetitive and dangerous tasks that robotic process automation and machine learning could take over. The same report pointed to better production and processing, smoother transportation logistics, closer equipment monitoring and preventative maintenance. It also expected AI to automate up to 50% of job competencies in upstream oil and gas by 2040. Suncor Energy shows the idea in practice. The oilsands giant runs a large fleet of autonomous trucks and has brought AI into its dispatch system at the Mildred Lake mine. As of 2024, that system handles routine decisions such as assigning trucks to dump stations and sending them to refuel. The technology reaches well beyond heavy industry, though, and some of its most carefully argued uses concern language rather than machinery.

  2422

• 7. Language and Data Sovereignty

  ›

  Since the mid-2010s, Indigenous organizations have partnered with the National Research Council of Canada to build technology for language revitalization. In 2025 the work took a new step. Inuit researchers and technology partners launched an AI initiative to help revive and preserve Inuktitut. A CBC article that year recorded both the promise and the unease. AI can support revitalization, but Inuit leaders stress data sovereignty and who owns the information. They also insist that Indigenous leadership guide the work, so that development stays transparent, private and accountable. The concern echoes the phrase Evan Solomon used in Montreal, applied here to a single community's words and culture. Questions of control like these are also what formal rules and public money are meant to settle.

  39129

• 8. Rules, Money and Machines

  ›

  In November 2022, the government proposed the Artificial Intelligence and Data Act, known as AIDA, as part of the Digital Charter Implementation Act, Bill C-27. Voluntary measures followed, including a Code of Conduct for Generative AI in September 2023. Federal officials also backed advanced computing infrastructure and the Canadian Artificial Intelligence Safety Institute, called CAISI. A 2024 report from Innovation, Science and Economic Development Canada argued that public awareness, trust and AI literacy are essential for responsible adoption. Money is now attached to the ambition. The 2025 federal budget sets aside over $1 billion across five years for the AI and quantum computing ecosystem. Computing power gets its own line. The Canadian Sovereign AI Compute Strategy received up to C$2 billion in Budget 2024, and up to C$700 million of that is meant to draw private investment into new or expanded data centres. As of April 2026, Statista counts 287 data centres in Canada. Alberta wants many more, and its AI Data Centres Strategy targets C$100 billion worth of AI data centres under development by 2030. The Wonder Valley AI Data Centre Park near Grande Prairie is the major project in that plan. It was exempted from provincial environmental impact assessment in April 2026, yet it still needs permits proving it can be built and run safely.

  19273686