TORONTO — Canada’s new AI strategy aims to put the technology to work to grow the economy, secure the country’s place in the world, and help people trust and use it to their benefit.

In return, Ottawa is setting lofty targets. It projects the strategy, titled “AI for All,” will drive a three per cent increase in Canada’s gross domestic product, worth almost an extra $200 billion cumulatively by 2031, through increased labour productivity and the creation and sale of new AI products and services. The plan, which Prime Minister Mark Carney announced in Toronto on Thursday, also forecasts that adoption of the technology will generate 250,000 new jobs by 2031.

“The question isn’t whether AI will transform our lives. It will,” Carney said at a press conference in Toronto Thursday. “The question is will it improve the lives of all Canadians or benefit only a few.”

As The Logic first reported Wednesday, Ottawa is committing hundreds of millions of dollars to help AI firms access financing and compute, and to help companies in other sectors adopt the technology. The federal government is also promising to buy from major Canadian tech firms at home, and promote them abroad. Much of the money will flow through long-standing federal programs and offices.

Amid significant public distrust of AI, the strategy also includes measures to improve people’s understanding of how it works, and to train early- and mid-career workers in how to employ it in their jobs. And to address concerns about negative personal impacts, the Liberals are promising a small boost to AI safety research, and new protections and privacy rules, though the document doesn’t provide a lot of detail. 

Growing AI firms 

The strategy proposes to help AI startups scale up and compete globally through a mix of funding and contracts. 

As The Logic reported Wednesday, the federal government has committed $500 million for a new Canadian Tech Growth Fund, which will make investments in companies, and may take equity stakes in them. That marks a change in approach for Ottawa, which has typically provided loan-like financing or grants. 

Arvind Gupta, a University of Toronto computer science professor and member of the task force advising the government on the AI strategy, said the model could work if Ottawa invests alongside experienced venture capital firms rather than trying to pick winners itself, noting that the approach has worked well in Israel.

On top of the new growth fund, the strategy references the $1.75 billion Ottawa announced in the 2025 budget to draw in more private investment for innovative companies. That commitment includes $1 billion in funding for the government’s flagship venture capital fund, called the Venture and Growth Capital Catalyst Initiative, plus $750 million for “early growth-stage” funding.

Investment industry groups have spent recent months making the case for how that $750 million should be spent, with some saying it should be used at the earliest stages of growth and others at a later stage to help companies scale.

AI Minister Evan Solomon had promised that the strategy would help startups raise more capital at the early and growth stages. Successive federal governments have also tried to increase the amount of money available to startups through venture capital programs to indirectly or directly seed funds that back tech firms.

The finance department will also look at ways to encourage investors to “reinvest gains earned from successful tech companies into Canadian startups.” Build Canada, a tech advocacy group, has called for such a rollover tax break, citing a similar U.S. incentive, while the Conservative party campaigned in the 2025 election on deferring taxes on capital gains reinvested in Canada. Some members of the task force Solomon appointed last September to propose ideas for the strategy had recommended that Ottawa also increase tax breaks for the founders of AI startups. 

“Too many promising Canadian companies grow elsewhere,” the strategy acknowledges, citing the pull of the U.S. technology market, the world’s largest. 

The strategy carves out a special place for Canadian makers of so-called foundation models, which power generative tools like ChatGPT and Claude. The document cites Toronto-based AI firm Cohere and Montreal-based non-profit LawZero as examples. It proposes to treat foundation models as “strategic assets, with the potential to scale globally while remaining anchored on Canadian soil.” Ottawa plans to back research into such systems if they’re built with safety in mind, and to help export them to countries “actively seeking trusted alternatives.” 

Cohere co-founder Nick Frosst said the strategy shows ambition, and that he was glad to see the principles on which it is based, which include fostering trust, providing opportunity and safeguarding sovereignty. “It’s obviously a long way to go in between a new policy and how that affects the country and the industry, but it’s broadly great to see the steps that are being taken,” he told reporters at the announcement.

The Liberal government is also pledging to use its buying power to help homegrown firms, meeting its own need for new technology tools while acting as “a strategic anchor customer” that companies can tout when selling abroad. The new Digital Transformation Office, first announced in last November’s budget, is designed to speed up procurement. Ottawa has already made some small-scale purchases from firms like Cohere and Ada Support as part of pilot programs or tests. 

Previous federal attempts to buy more from startups and innovative companies have not succeeded. Ottawa gutted the last such program in February 2024 after departments consistently failed to meet their spending obligations. 

Regulating AI

The strategy also gives some thought to putting guardrails around AI, though it’s not immediately clear how some of the new safety measures will be enforced.

The prime minister acknowledged that Canada ranks near the bottom of countries in AI training, literacy and trust, and the threats the technology poses to privacy and safety.

The government plans to watermark AI-generated content, for example, and pitched a “Canada Trusted AI Certification” program to help people identify trustworthy AI products.

The strategy also includes a vague promise to protect critical systems from cybersecurity threats posed by advanced AI systems by working with frontier AI companies and international partners. 

Earlier this spring, Solomon told The Logic that Canada’s institutions can withstand the threat posed by Anthropic’s latest AI model, Mythos, which has proven to be particularly adept at cracking cyberdefences. He cited the capabilities of Canada’s AI Safety Institute as one of the reasons he felt so confident. 

The government has pledged $50 million to expand its AI Safety Institute to track emerging AI risks, do technical research and evaluate AI models. 

Solomon also turned to the institute when it emerged that the teenage shooter who killed eight people in Tumbler Ridge, B.C., in February after exchanging disturbing messages with OpenAI’s ChatGPT that the company failed to flag to police. 

Solomon asked the institute to examine OpenAI’s safety protocols and report back. The tragedy in Tumbler Ridge put a spotlight on the safety risks associated with AI chatbots. The government is now considering restricting children’s access to chatbots as part of its long-awaited online safety legislation. 

The strategy also promises to table a new law to update Canada’s long-outdated privacy legislation to safeguard consumers’ data. The previous Liberal government made a similar attempt to bring Canada’s privacy laws into the modern age, but the bill never made it through Parliament. 

The strategy also emphasizes the need to build public understanding of AI, and get more people using the technology. It proposes free, entry-level online training, and programs run by libraries and community organizations. It also promises to give all post-secondary students “access to trusted AI agents.” 

Ottawa has already renewed funding for AI skills education for kids via the CanCode initiative—one of several innovation programs set up by the Liberal government under former prime minister Justin Trudeau that was set to end—with an additional $30 million for the next two years as part of April’s spring economic update.

Driving AI use

The strategy sets a big target to diffuse AI across the economy, aiming to increase business adoption of the technology from 12 per cent today to 60 per cent by 2034. 

To help get there, the federal government is adding $500 million to a currently $200-million AI program run by its seven regional development agencies that helps firms pay for work to adopt and commercialize AI. The initiative offers loan-like funding of between $125,000 to $5 million, depending on the province and the project.

The strategy also incorporates BDC’s LIFT program, a $500-million initiative announced in April to help over 1,000 small- and medium-sized businesses adopt AI. The initiative pairs firms with industry advisors to figure out where to use the technology, and provides financing to install it. A BDC report published Wednesday found that just 30 per cent of Canadian SMEs currently use generative tools, with non-adopters leaving some $150 billion in GDP on the table. 

As a complement to the financing programs, Ottawa aims to develop a new literacy and adoption assessment tool, which small firms can use to figure out whether they’re ready to use AI, and for what. 

It’s also expanding the AI Compute Access Fund, which helps firms pay for the processing power they buy from cloud service providers, with an extra $700 million. The oversubscribed program has so far awarded $66 million of its original $300 million budget to 44 companies, including venture-backed tech firms like Basetwo, EarthDaily and Vendasta. 

Corporate Canada has historically been slow to adopt new technologies, contributing to what the Bank of Canada has described as a productivity “emergency” for the country. AI proved no different in the early stages, with Canadian business rates of use trailing other advanced economies. 

Task force members had called for tax breaks for firms that adopt AI, with incentives to buy from Canadian developers. During the 2025 federal election, the Liberals promised a rebate for small and medium-sized businesses of 20 per cent of their AI adoption expenses, but the incentive was not included in the strategy released Thursday. 

Building AI infrastructure

Ottawa plans to use its buying power to try to increase the amount of compute available to Canadian firms and researchers.

The federal government will back data-centre projects of at least 100 megawatts, larger than the country’s largest existing facility not directly owned by a U.S. cloud giant. It’s betting that committing to buy processing power for its own tech needs will help the builders of new AI infrastructure attract private investment, adding more capacity for companies and universities in Canada.

November’s budget gave Solomon the mandate to negotiate such data-centre deals. Ottawa has received 160 proposals, and deals in the works could add 850 megawatts of capacity by 2030, and grow to as much as 2.3 gigawatts over time. 

Canadian companies, governments and researchers currently rely on “infrastructure they do not own, govern or meaningfully influence,” the strategy says. Organizations are sometimes storing data or using AI located in foreign countries. “These dependencies are a strategic exposure Canada cannot afford to leave in place,” according to the strategy.

The document acknowledges that so-called hyperscalers—the cloud arms of U.S. tech giants—are likely to sell much of the AI compute that Canadian organizations will use for the foreseeable future. The country will “continue to welcome that foreign investment where it delivers clear benefits,” the strategy says. Still, the strategy makes clear the need for a “sovereign alternative.”

The federal government is already in talks to back Telus to build three new compute facilities in British Columbia with a combined capacity of 150 megawatts. Several other firms, including telecom rival Bell as well as data-centre specialists Hypertec and eStruxture have touted their Canadian expansion plans as a way of meeting Canada’s need for sovereign AI infrastructure.

Several task force members had called for the federal government to help remove regulatory and paperwork barriers to data centres, which the strategy does not address.

Ottawa also aims to make more data available for innovators to build new products, including via $100 million in funding to expand Vital, a program based out of Toronto’s St. Michael’s Hospital that will let researchers work with anonymized health records. And the federal government will launch a new $100-million program with the Canadian Institute for Health Information that brings together research and clinical trial datasets. 

People and skills for AI

Under the strategy, the government will also expand a long-standing program designed to help universities recruit and retain top AI researchers. It aims to increase the number of AI chairs, which help AI institutes and schools pay for faculty and their labs, from 143 to nearly 200. The government will also provide $130 million for programming at the three national AI institutes to help commercialize research discoveries.

The chairs program—which is managed by the Canadian Institute for Advanced Research—formed the core of the original national AI strategy, announced with $125 million in funding in March 2017. Last July, CIFAR and the three national AI institutes proposed a $186-million renewal package to expand the initiative amid what they described as an “unprecedented global war for AI talent.”

The strategy aims to create opportunities for up to 90,000 students to learn AI skills in the workplace. It’ll offer those through existing federal programs for student work placements, Canada Summer Jobs and Skills for Success, as well as third-party organizations like Mitacs. Ottawa also plans to back the expansion of employer-led training programs for mid-career workers. 

The federal government claims it’s “pursuing a pro-worker approach to industrial AI,” in which the technology works alongside humans rather than instead of them. To give workers more of a say, Ottawa is proposing to set up six new sector-specific tables at which governments, employers, unions, post-secondary institutions and Indigenous groups can discuss skills gaps and the talent needed to fill them. 

Gupta said he’s generally encouraged by the strategy but that some of the measures were too vague to draw conclusions on. “The words all sound really good to me,” he said, “but I’d like to see the details.”

With files from Catherine McIntyre in Toronto