When a single American hyperscaler can drop $100 billion on AI infrastructure in a single year, what’s a middle power supposed to do? Canada’s answer, delivered at Web Summit Vancouver this week by the country’s first-ever Minister of Artificial Intelligence and Digital Innovation isn’t surrender. Nor is it a moonshot attempting to match U.S. spend dollar-for-dollar.

Instead, it’s a softer declaration of digital independence, with an acknowledgment that independence doesn’t necessarily mean isolation.

"Sovereignty is not solitude," Minister Evan Solomon said in a press conference Tuesday morning at the Vancouver Convention Centre. "We are going to do business with the United States. They are our biggest customer."

That said, Solomon announced 44 new projects across British Columbia receiving up to $66 million through Canada’s Compute Access Fund, a $300-million program launched in July that the minister says is "wildly oversubscribed." The goal is Canadian data centers serving Canadian companies and Canadian citizens, along with any other global customers that might come along. The fund subsidizes 50 cents on the dollar for compute access, rising to 67 cents on the dollar when that compute is Canadian. It’s a thumb on the scale, Solomon says, which is modest by hyperscaler standards but explicitly designed to build a domestic compute customer base for the data centers Canada is now starting to approve.

Recipients include SenseNet, the Vancouver-based wildfire-detection company founded by a PhD student named Hamed Noori. Noori watched the 2018 BC wildfires choke the city with smoke and decided AI should solve it. SenseNet now operates in 11 countries with more than 30 patents. Also funded: Spare, which uses AI to optimize paratransit and public transportation routes in roughly 250 cities globally, including BC Transit and Translink, Vancouver’s public transportation network.

Part of the overall project is building new data centers in Canada at a time when data centers are getting more controversial. Mega projects like Kevin O’Leary’s Utah-based 9 gigawatt data center face public scrutiny over their power and water use, and the tax breaks they demand from local governments. O’Leary’s project, for instance, will get an energy use tax reduction from the standard 6% to 0.5%. Also, the local authority will rebate 80% of the property tax back to O’Leary Digital.

Canada has different ideas around how to build and manage data centers.

The day before the BC funding announcement, Solomon was at Canadian telco Telus’s headquarters to announce an MOU for three new AI-focused data center sites in British Columbia.

The pitch: sustainability-by-design with closed-loop water systems drawing waste rainwater from BC Place stadium plus heat reuse, essentially cooling chips by warming homes. On its face, this is a data center that fits within existing infrastructure and benefits local communities rather than taking from them, with Solomon promising that power use would not increase electricity rates for ordinary people.

It’s also a very real test of whether Canadians will tolerate AI infrastructure in their backyards.

"We should be very open to real citizens’ concerns," Solomon said. "Technology moves at the speed of innovation and citizens move at the speed of trust. And you got to move at the speed of trust."

That’s an acknowledgment that the U.S. data-center backlash which has resulted in local pushback is coming for Canada too unless the government gets ahead of it.

A big question remains, though.

If Canada hopes to achieve some level of digital, AI and cloud sovereignty from the American AI and cloud companies that currently dominate pretty much the entire world outside of China, how is that possible on hundreds of millions of dollars when compared to the hundreds of billions that the tech giants are deploying?

I asked Solomon that question, and he had a four-part answer: talent, energy, capital and partnerships to develop options.

On the talent side, Solomon said that “Waterloo is graduating more engineers than Stanford” and that Canada’s last budget put $1.7 billion into a talent attraction program targeting 1,000 of the world’s top AI researchers and their labs. That’s the largest such program in the G7, Solomon said.

On the energy side – critical for power-hungry AI chips – Solomon said that Canada has “the largest clean energy grid, lots of it unused, but the [largest] potential in the world.”

Hydro Ottawa, Solomon noted, has already deployed AI to manage its own grid more efficiently, which he said was key to more efficient management of the power that Canada is currently generating. "The best new energy is the energy that we’re not already using," he said, "and that is better than building more." Translation: the best new megawatt is the megawatt you don’t have to generate … because you’re saving it with a more efficient grid, or more effective use.

Capital, of course, is a challenge.

Pressed on how Canada competes when individual U.S. firms outspend the entire fund 300-to-1, the minister answered: “Access to capital is a challenge in Canada. We're working on that. You're 100% right to identify that. Companies need access to capital from seed, Series A, Series B, right up the chain.”

One of the core ways to compete is building partnerships to develop more options. Canada’s recent sovereign technology alliance with Germany is building "an alternative to either the hyperscalers or the hegemons," Solomon said. The point isn't to beat OpenAI. It's to make sure Canadian companies, Canadian governments, and Canadian citizens have a real choice that doesn't put their data under foreign jurisdiction.

"You need optionality," Solomon said. “And we are creating options, so people have choice.”

Of course, compute for inference is only one side of the AI coin. That’s the side that generates the answers we seek from Claude or ChatGPT or Gemini. But the foundational AI models that enable inference compute to generate the answers we want are another matter yet, and most of those are American or Chinese. Solomon says that Canada can compete here as well.

"There’s only four countries in the world that have a foundation model: the U.S., China, France with Mistral, and Canada with Cohere," he said.

That’s a fairly aggressive read of the foundation-model landscape. There are credible arguments for adding the UAE, the UK, and others depending on how you draw the line, but the political logic is unmistakable. Canada has invested over a quarter of a billion dollars in Cohere, and the federal government just bought 1,400 Cohere licenses for its own internal use.

"The government plays a role on the demand side as well to put our thumb on the scale to say we will champion our champions," Solomon said.

Just yesterday Cohere’s Chief AI Officer Joelle Pineau joined Solomon at Web Summit’s opening night. The company hit $240 million in annual recurring revenue in 2025, is backed by Nvidia, AMD, and Salesforce and is widely expected to pursue an IPO. Whether Cohere can scale into genuine global competition with OpenAI and Anthropic is a very open question. Cohere recently announced an acquisition of German AI firm Aleph Alpha that values the combined entity at roughly $20 billion, which might be a good start.

Add it all up and it’s a complex multipart strategy with many moving pieces mixed in with all the other challenges that Canada faces geopolitically with the United States and trade. But ultimately, Solomon said, it’s deeply pragmatic.

“We’re not team pom-poms — like, yay, AI’s going to save everything. We’re not team pitchfork — don’t build data centers, stop,” he said. "We're team pragmatic. That's the Carney government."

Per capita foreign direct investment into Canada has run at roughly 2x the U.S. rate over the past year, Solomon said. And while the country is investing in sovereign AI compute, it’s also moving forward on AI chatbot regulation for minors, modernized privacy laws, deepfake legislation, and a forthcoming AI and copyright advisory board.

Whether the strategy works will depend on whether the $300 million investment – and billions in other initiatives – can compound into something that grows over time.