Can Canada’s Emerging AI Blueprint Successfully Challenge Global Tech Superpowers?
The biggest deal in the Canadian artificial intelligence sector started over coffee last fall in Frankfurt, when Cohere Inc. chief executive officer Aidan Gomez met informally with a small team from German company Aleph Alpha, including Ilhan Scheer, now its co-CEO. Both AI companies were considered darlings by their respective governments, homegrown competitors in a field dominated by U.S. tech giants. At the coffee shop, the group realized they also had a similar philosophy about AI development. “His vision was very appealing,” Mr. Scheer recalled, referring to Mr. Gomez. “For us, sovereignty was important, and safety.” They left that meeting thinking about how to work together, he recalled.
In April, Toronto-based Cohere and Heidelberg-based Aleph Alpha announced a deal to combine operations, and this week, they said they had signed the agreement. (The deal still requires regulatory approval.) “I’m so proud to be able to bring Canadian-German values to this technology,” Mr. Gomez said Wednesday on stage at Montreal’s All In conference, the country’s largest AI expo.
The goal is to create a bigger, more powerful AI company to provide large language models and services to governments and businesses, acting as an alternative to OpenAI, Anthropic and Google. Mr. Gomez was particularly fired up this past week, branding some of his competitors as a “cartel” over their support for pacing AI development over safety concerns.
The deal is also the most visible sign yet of how players in Canada and like-minded allies are striving for influence over the most consequential technology of our age. AI carries global repercussions, but development is concentrated among a small group of powerful companies in the U.S. amid heightened geopolitical tensions under President Donald Trump. China is not far behind in AI development, either. That reality could also explain why both the Canadian and German governments are in talks to provide funding to Cohere in its next round, as The Globe and Mail previously reported.
Canada is pursuing a few options, in fact, to ensure that AI is not the sole domain of the U.S. and China. If there was one theme at All In, a two-day extravaganza of AI boosterism, nerdy technical discussions and the occasional robot, it was that Canada needs options. “Every country needs the capacity to make choices about the technologies shaping its future,” federal AI Minister Evan Solomon said on stage on Wednesday. To that end, Canada is forging deeper ties with Germany and other countries, while putting money into AI safety initiatives that could lead to a new way of developing the technology.
But it won’t be easy. U.S. AI companies are enormous (Anthropic was last valued at US$965-billion compared to the US$20-billion valuation Cohere is seeking) and the amount of funds required to compete are equally enormous. That puts a lot of pressure on Cohere to succeed. It cannot subsist on the largesse of governments or survive long-term as a plan B for the corporate world. “It has to be more than a backup,” said Olivier Blais, co-founder of Moov AI, a consulting company in Montreal.
While the deal has been presented as a merger, it is more akin to an acquisition of Aleph Alpha. “There is a financial reality to it,” Mr. Scheer said, who was attending All In. “Cohere is bigger.” (The company will bear the Cohere name and be dual headquartered, with Mr. Gomez as CEO.)
Aleph Alpha knows how hard it is to keep up in AI. The company was founded in 2019 to build LLMs, but struggled to gain traction and sidelined model training in 2024 to instead help companies and the public sector implement AI. “We can’t compete because we don’t have billions, so let’s do something else,” Mr. Scheer said, characterizing the thinking at the time. He joined in 2025 as chief growth officer but as co-CEO, he has pushed the company back into model training, only specialized ones tailored for clients.
He acknowledged that trying to keep up with the AI giants is akin to staying in the slipstream behind a race car, but felt geopolitical events have created an opportunity. “The U.S. relationship with these companies is causing a shadow on their services and products in Europe,” he said. “With the trust established in us and with the enterprise-ready AI we’re delivering, that’s where we can scale and beat them.”
Some believe the tide is turning for companies such as Cohere. Mr. Blais at Moov AI (which is owned by global communications agency Publicis Groupe) is a fan of Cohere’s models. “There are several very good frontier-level systems” in the market, Mr. Blais said. “You don’t need to get the latest and greatest for everything.” He is also a fan of the Paris-based company Mistral, which builds closed and open-weight AI models and helps companies integrate AI.
Mistral has established Montreal as its Canadian base, and has domestic customers in the financial services and manufacturing sectors. “The industries in Canada are really a match made in heaven,” said Marjorie Janiewicz, Mistral’s chief revenue officer, who also attended All In. Like Cohere, Mistral has positioned itself as a sovereign AI company, and users can freely modify its open-weight models, for example. Ms. Janiewicz said clients are increasingly asking about the risks of getting cut off from models. (Earlier this year, Anthropic temporarily disabled its Fable and Mythos models following a U.S. government directive.) “Companies are really starting to realize the importance of owning your own AI and not relying on renting that AI from somewhere else,” she said.
The big U.S. AI companies are just as interested in working with Canada, of course. George Osborne, the U.K.’s former chancellor of the exchequer who is now a managing director at OpenAI, struck a sanguine tone during a brief interview at All In, reiterating the company’s interest in potential data centre capacity in the country. “Canada looks very attractive,” he said. “One, it’s a government that’s got its act together and wants to engage with international investment. Second, the AI adoption plan is more coherent and advanced than most countries.”
But OpenAI is facing serious questions about its approach to safety after disclosing this summer that its agents had hacked into another company as part of a cybersecurity evaluation. Hugo Larochelle, the scientific director of Montreal-based AI research institute Mila, said he was troubled by the fact that OpenAI was not properly monitoring its own agents. “That removed a lot of my own trust in whether they can be shepherds of the technology,” he said.
“Why is it that we should continue trusting you?”
Indeed, the episode points to the need for alternative approaches to AI, according to some experts, which is part of the reason Canada and Germany are providing up to $300-million in funding to Montreal-based LawZero, which was also announced at All In. A non-profit founded by Yoshua Bengio, LawZero is developing an AI system that will not exhibit the harmful behaviours of today’s AI models.
“We were convinced that this is the path, that we undertake more efforts to develop from first principles AI systems that are safe,” said Karsten Wildberger, Germany’s Minister of Digital Transformation and Government Modernization.
While few would doubt the credentials of LawZero’s founder, who has made significant contributions to building modern AI, the non-profit is pursuing an experimental and uncertain path. Mr. Wildberger said the $150-million Germany is willing to put up is a small price to pay, however. “The price of acting too slowly would be a much bigger number,” he said, adding that there are milestones attached to the funding.
Elsewhere at All In, Alberta Technology and Innovation Minister Nate Glubish was pitching a different way for Canada to assert leverage. In 2024, he launched a data centre strategy to court big tech companies and developers to build in the province, touting Alberta’s natural gas supplies and the economic benefits for Albertans. But in an interview, he also highlighted that it could prove to be a source of geopolitical strength. “The best way we can ensure that Alberta and Canada have influence over the future of AI development is to ensure that the critical infrastructure that powers the next generation of AI technology includes locations in Canada and Alberta,” he said.
Having domestic infrastructure means access for Canadian AI researchers and companies, he said, but more than that, the U.S. wants to maintain AI supremacy over China. To do that, the industry needs data centres – and the electricity that Alberta can provide. “Jurisdictions that can offer an environment where that infrastructure can be built quickly, so they can maintain that technological lead, will have far more influence,” he said.
Canada is increasingly looking abroad, too. It formed a “sovereign technology alliance” with Germany in February to work more closely on AI, and the joint funding for LawZero is one sign that the collaboration is starting to pay off. What’s more, Mr. Wildberger said there is “huge” interest from other countries to participate in the alliance. “We have received so many questions and requests to collaborate,” he said. On Friday, he and Mr. Solomon held a meeting with counterparts from the U.K., France, Spain, Norway and Australia to discuss AI.
Mr. Wildberger declined to comment directly on whether the German government is in talks to contribute funding to Cohere. “If there is a good reason to leverage public money to develop certain things that are of national interest, that’s something to be discussed,” he said. “However, the vast majority has to be carried forward, obviously, by the private sector.”
Jaxson Khan, founder of Toronto-based consultancy Aperture AI, said the Canadian and European Union markets, not to mention those of other middle powers, amount to a substantial bloc with massive purchasing power. “Canada needs multiple options on the AI front and to reduce dependency where possible,” he said. “We can encourage people to use and adopt whatever platform works for them, but it is important that we invest in domestic, globally significant AI champions.”
For now, the geopolitical trends are working in favour of Cohere and Aleph Alpha. But the combined company will still have to compete on its own merits, which Mr. Scheer recognizes. “If the relationship with the U.S. would be different, I don’t know if we would have the same conversation,” he said.
It is hard to compete with the giants, no doubt. But if there was no will to see alternatives succeed, it would surely be even harder.
This article was first reported by The Globe and Mail
