Canada is betting billions on sovereign AI data centers, but the promise of complete control over national AI infrastructure is running into a hard reality: sovereignty in a connected world is far more complicated than simply building data centers on Canadian soil. The federal government has committed $925.6 million over five years to support large-scale sovereign public AI infrastructure, with Telus becoming the first successful applicant to develop three new facilities in British Columbia. Yet technology experts and privacy advocates are raising uncomfortable questions about whether Canada can truly achieve the independence it's seeking when the infrastructure still depends on foreign hardware, foreign customers, and digital networks that don't respect national borders.

The concept of "sovereign AI" has become a rallying cry across governments worldwide, but Canada's approach reveals the fundamental tension at the heart of the strategy. Minister of Artificial Intelligence and Digital Innovation Evan Solomon has emphasized that new data centers will have sovereignty requirements, meaning they must be owned and controlled by Canadians and subject to Canadian law. However, the fine print tells a different story. Telus's new facilities will house more than 60,000 graphics processing units (GPUs) from Nvidia, a U.S. company, once fully operational. The data centers will serve not just Canadian governments and industry, but international customers as well. And the federal government has explicitly stated it won't restrict who can use these facilities.

What Does "Sovereign AI" Actually Mean in Practice?

The term "sovereign AI" has become increasingly popular among policymakers, but it means different things depending on who's using it. At its core, sovereignty in AI refers to a nation's ability to control its own data, infrastructure, and the AI systems that operate on that data, without dependence on foreign governments or corporations. However, the practical definition is far messier. Some experts argue that true sovereignty requires owning every component of the technology stack, from chips to software. Others suggest that sovereignty is more about governance and control than ownership, meaning a country can use foreign technology as long as it maintains operational control and legal authority over how that technology is used.

The MIT Technology Review Insights report found that enterprises "deeply committed" to controlling their data, infrastructure, models, and governance are delivering five times the return on investment (ROI) from generative and agentic AI initiatives compared to their peers. The study surveyed more than 2,000 senior executives across 13 countries and ran more than 15,000 simulations across 500 variables. The correlation between sovereignty commitment and AI success outcomes was 0.93, making it the strongest single driver of AI success identified in the research.

But here's the catch: that same research shows that 95% of organizations plan to establish their own AI and data platforms within the next three years, suggesting that sovereignty is becoming table stakes for competitive advantage. Yet the report also emphasizes that sovereignty doesn't mean isolation. Instead, hybrid environments are emerging as the dominant operating model, balancing innovation with sovereignty and regulatory control.

How to Build Sovereign AI Infrastructure: Key Considerations for Governments

• Physical Location Plus Operational Control: Data centers must be physically located in the country and controlled by domestic companies, but this alone doesn't guarantee sovereignty if the hardware, software, and governance frameworks are foreign-owned or foreign-controlled.
• Data Localization and Network Isolation: Keeping data within national borders requires more than just storing it locally; it requires isolating the network to prevent data from traveling across borders through digital pathways that don't respect national boundaries.
• Governance and Regulatory Authority: Establishing clear legal frameworks that give the government authority over how AI systems operate, who can access them, and how data is protected is essential, but requires coordination across multiple agencies and stakeholders.
• Talent Development and Skills Building: Governments need to invest in developing domestic expertise in AI safety, data governance, sovereign cloud architecture, and AI policy, areas where critical talent shortages currently exist.
• Hybrid Models and Selective Sovereignty: Rather than attempting complete isolation, governments are increasingly adopting hybrid approaches that maintain stronger control over sensitive national data and regulated workloads while continuing to leverage global technology ecosystems for innovation.

Sharon Polsky, president of the Privacy and Access Council of Canada, emphasized the importance of multiple layers of control. "Having the data in Canada is one thing. Having the companies that build and own the facilities be Canadian is important. Having them operate within Canada is also critical," she stated. Polsky also raised questions about whether private sector companies involved in data centers and telecommunications should be restricted from foreign ownership, suggesting that true sovereignty may require regulatory measures that go beyond current proposals.

Rudi Carolsfeld, co-founder of Green Edge Computing, a Victoria-based startup specializing in small-scale data centers, pointed out a technical reality that complicates the sovereignty narrative. "The national borders are not always guaranteed to be respected by digital traffic," he noted. "You'd have to isolate the network to keep it". This observation highlights a fundamental challenge: in a world where data flows across borders at the speed of light, maintaining complete control over information is technically difficult and economically costly.

Why Are Governments Suddenly Prioritizing Sovereign AI?

The shift toward sovereign AI isn't driven by ideology alone; it's driven by geopolitical anxiety and concrete security concerns. A recent case in Canada illustrated the stakes. A Canadian launched a lawsuit against the U.S. Department of Homeland Security, which allegedly sought "vast swaths of information" through Google about his personal life following social media posts critical of Donald Trump's administration. U.S. laws give the country's intelligence and law-enforcement services broad powers to access data stored on U.S. servers, even if that data belongs to Canadian citizens.

Across Asia-Pacific, governments are treating sovereign AI as a matter of national resilience. According to research commissioned by Dell Technologies and conducted by the International Data Corporation (IDC), sovereign AI surged from the seventh-highest to the second-highest government investment priority across the region within just one year. The study surveyed 360 government IT decision-makers across eight Asia-Pacific markets, including Singapore, India, Indonesia, Japan, and South Korea. Nearly half of Asia-Pacific government agencies are actively evaluating sovereign AI technologies, while more than a third are already running pilot projects and proof-of-concept deployments.

Over three-quarters of respondents said sovereign AI would strengthen resilience against geopolitical risks and supply-chain disruptions. National security and cyber resilience ranked as the top expected area of citizen impact from sovereign AI investments, followed by justice and public safety, financial administration, healthcare, and social services.

The urgency is intensified by the rise of agentic AI, autonomous systems that take actions and trigger workflows with minimal human oversight. More than half of enterprises already have AI agents in production making real-time decisions on operational data. With more than 1 billion active agents projected by 2029 executing 217 billion actions per day, the question of who controls these systems and the data they access has become critical.

"Sovereignty defines which agents can touch the data, in which region, under which policies, and how all of that is monitored and audited. Rather than being a brake on agentic AI, sovereignty sets the safe operating boundaries that allow organizations to scale with confidence," said Devin Pratt, research director at IDC.

Devin Pratt, Research Director at IDC

The Skills Gap That Could Derail Sovereign AI Ambitions

One of the most overlooked challenges in the sovereign AI push is the shortage of specialized talent. Nearly nine in ten government organizations surveyed in the Asia-Pacific region reported critical technology talent shortages, with the most difficult roles to fill including AI safety and alignment researchers, data governance specialists, sovereign cloud architects, and AI policy professionals. This talent gap could become a major bottleneck preventing governments from moving AI projects beyond pilot stages into trusted national-scale deployments.

Canada faces similar challenges. Building sovereign AI infrastructure requires not just capital investment, but expertise in areas like cryptography, data governance, and AI safety. The federal government's commitment of $925.6 million over five years is substantial, but it's unclear whether this funding includes sufficient investment in workforce development and talent attraction.

Telus's announcement of three new data centers in British Columbia represents a significant step forward. The Kamloops expansion and the Mount Pleasant facility in Vancouver will open later this year, while the downtown Vancouver facility will come online in 2029. The project will get more than 98% of its electricity from renewable sources, and waste heat from the two Vancouver facilities will be used to help heat homes, addressing both environmental and efficiency concerns.

"We can have something that's built and made in Canada, for Canadians, in a sovereign way," said Mirko Bibic, chief executive of Bell Canada.

Mirko Bibic, Chief Executive of Bell Canada

Yet even Bibic acknowledged the complexity of the challenge. "We're going to do business with the United States. They're our biggest customer and sovereignty is not solitude," he noted, echoing a sentiment shared by government officials and industry leaders alike.

Louis Têtu, executive chairman of Montreal-based Coveo, an AI platform with 800 employees in Canada, offered a pragmatic perspective on the sovereignty question. "Canadian data centers should use the best technology in the world, but be operated by Canadian firms that also control the network with Canadian governance. That allows us to essentially make sure that we are immune to geopolitics; to anybody else trying to turn us off. And it's a national interest imperative," he stated.

The sovereign AI movement reflects a broader recognition that technology infrastructure is no longer just a business issue; it's a national security issue. But as Canada and other nations invest billions in sovereign AI infrastructure, they're discovering that true sovereignty is less about building walls and more about building resilience, governance, and the expertise to maintain control over systems that are increasingly autonomous and interconnected. The real test will come not when the data centers are built, but when they're operating at scale and governments must navigate the inevitable tensions between sovereignty and innovation, between control and collaboration.

" }