Digital sovereignty took centre stage at the All-In 2026 conference in Montreal, where Canada’s Artificial Intelligence Minister Evan Solomon warned that failing to build data centres and advance AI at home would leave the country dependent on foreign providers.
Digital sovereignty and data-centre build-out
Solomon told attendees that “If we don’t build it here, we’ve got to buy it from someone else. And if we don’t innovate here, we’ve got to rent it from someone else. And if we don’t make the rules here, we’ve got to follow someone else’s.” He added that sovereignty does not mean isolation, describing it instead as “partnership with shared values.” The remarks were made at the All In conference in Montreal on Wednesday.
Germany’s digital minister Karsten Wildberger echoed the need for shared standards and oversight to ensure safe AI development while remaining competitive, saying “Nowhere does sovereignty matter more than AI. Few technologies have had such power to change our lives so quickly.” Wildberger urged middle powers such as Canada and Germany to turn concern into concrete action on AI production rather than remain on the sidelines.
The comments came amid renewed public anxiety about the risks of advanced AI after an employee at Anthropic quit earlier this month, warning that those systems could spiral out of control and destroy humanity. Concerns over digital sovereignty were also stoked by a recent mapping incident after “U.S. President Donald” last month dubbed Lake Ontario instead “Lake America,” a change that appeared on some online maps in Canada.
Allied partnerships, industry moves and safety funding
At the conference, Canadian research institute Mila and the German Research Center for Artificial Intelligence announced an agreement to strengthen AI ties between the two countries, focusing on safe AI systems and applications in industries such as energy. Canadian AI company Cohere and German startup Aleph Alpha formalized a previously announced merger intended to help them better compete with U.S. AI firms.
Cohere co-founder and CEO Aidan Gomez said the acquisition aims in part to attract companies and governments uneasy about relying on U.S. tech giants for digital services: “There need to be multiple democracies contributing to this technology. The world is going to diversify for resilience.” On international oversight, Gomez said it “is going to be tough — but the dialogue needs to be opened,” and called a multinational body preferable to isolated national regimes.
Solomon acknowledged in a later interview that Canadian businesses and governments remain heavily reliant on American tech, including cloud services from Amazon and Google, and that U.S. authorities have the right to access data stored by those companies. He said Canada’s trade relationship with the United States remains dominant and that his goal is not to build against any country but “to build for Canada,” pointing to Bell Canada’s announcement that it plans to quadruple the size of a data-centre project in Saskatchewan — a move described as a $50-billion investment and the largest capital investment in the province’s history.
Solomon also highlighted federal efforts to introduce privacy legislation intended to give citizens more control over their data and to establish a new regulator with powers to fine corporations. An AI literacy course for all teachers and high school students is set to launch next week.
Earlier in the day, Solomon and Wildberger announced $300 million in funding for LawZero, a nonprofit AI safety research organization founded by Yoshua Bengio and incubated at Mila.
The conference showcased both commercial and safety-focused AI work. Exhibits and demonstrations ranged from Mila’s open-source tools to detect AI image signatures, to industry displays by Microsoft and Nvidia. Dell presented a military-grade mobile data centre housed in a converted shipping container, while Aston Martin displayed its Formula One car with Cohere as an AI partner.
Michael Pelosi, country manager for Dell Canada, warned about the risks of relying on a single regional model: “If that model is coming from one market or one region and someone has the ability to turn it off, you’re in trouble,” noting the U.S. administration’s export ban on an Anthropic model through much of the summer.
The conference brought together government officials, researchers and companies to debate how democracies can both contribute to and regulate powerful AI technologies while reducing dependence on any single foreign provider.
