The Canadian founder telling CEOs to rethink AI resilience

Cohere’s valuation shot up from $7 billion a year ago to a reported $20 billion after the Canadian builder of AI models for enterprises and governments announced a planned combination with Germany’s Aleph Alpha this April. That still makes it a minnow compared to the lavishly funded giants of Silicon Valley or leading Chinese model makers like DeepSeek and Moonshot. But Cohere Co-Founder and CEO Aidan Gomez believes he can compete on the global stage without outspending them.

The reason: Business leaders are waking up to the risk of becoming caught in the geopolitical crossfire between the US and China, as some were when Washington imposed export controls on Anthropic’s Mythos AI model in June. Cohere offers an alternative that, he says, can make companies’ technology supply chains more resilient while giving them “AI sovereignty,” or full control over their AI infrastructure.

Executives now worry as much about control as they do about the cost or power of the models they deploy, Gomez says, so competing in AI is not just about “building the biggest model or spending the most money.” Once a customer deploys Cohere’s software, “We can’t see in. We can’t switch it off. We can’t intervene in any way,” he says.

Skeptics see sovereign AI as a pipe dream, arguing that it will be prohibitively expensive to duplicate US and Chinese technology. Gomez says companies cannot ignore the idea.

“We’re entering into a period of great geopolitical change and uncertainty, and we’re seeing sovereignty be threatened all over the world, so it’s no longer a potential risk that you need to plan for. It is a present-day risk that is unfolding in front of us,” he says.

Sovereignty should not mean that everybody builds a copy of everything, he clarifies, as that would be “extraordinarily inefficient.” But he argues that CEOs should diversify their AI suppliers to avoid “single points of failure” or reliance on technology from authoritarian countries.

One recent analysis estimated that up to 40% of AI spending could be influenced by sovereignty requirements by 2030. Cohere this week released an IDC study of executives in four countries, which shows that most believe that AI sovereignty is important, but few can define it or have a firm plan for achieving it. Here’s how Gomez describes the concept, why he thinks CEOs should care about it, and why he sees it as a matter of democratic resilience.

This interview has been edited for clarity and brevity.

Andrew Edgecliffe-Johnson: How do you define Cohere’s place in the AI ecosystem?

Aidan Gomez: Our objective is to strengthen the sovereignty of nations that are looking to adopt AI. We focus on critical industries, like telecoms, financial services, healthcare, the grid, and water treatment, and we really want to strengthen democratic leadership in AI on the global stage. It shouldn’t just be the US doing this incredibly transformational technology; we need to ensure that more democracies are building these capabilities and providing a robust and resilient supply chain around the world.

Are you finding it easier or harder to get the attention of big global companies, because of your positioning outside the China-US duopoly in AI?

It’s much easier. There’s a very intense push [by companies] globally right now to diversify supply chains and boost resilience, and to take control over their sovereignty in technology, and we’re the ones that do that. We believe you can’t rent your sovereignty from someone. You need to own it. This year, we have seen a very steep rise in outreach.

So what analysis should CEOs be doing right now of the AI tools they’re using?

I think, for democratically aligned nations, relying on tools from authoritarian states is a core risk, and should not be done. It is simply not worth that risk. But beyond that, I think that they should be considering how concentrated their supply chains are, and planning for the risks of losing access or having services disrupted. They should be actively engineering systems that are resilient and robust for those sorts of events.

Do you think this technology embeds existing corporate power or challenges it?

It’s very supportive of creative destruction. It’s much more of an equalizing force than it is a force that entrenches power. You can see that with studies that show when you give an employee base access to AI agents, the average moves up substantially, but the lower end of the performance curve benefits the most. They move up way further than the people at the top. So the difference between your best and worst employee is much tighter as a result.

Is that one of the things that gives you hope that a Canadian company can compete?

I think so. Many things give me hope. I think Canada is an extraordinary place to build. It really embodies the values of progress, and I think we’re trustworthy, like a country that’s a very loyal partner, and we’re kind and friendly and all those good stereotypes about Canadians. So there are a lot of reasons to believe that Canada has a lot to contribute to the world, especially in this era of tumult.

Your starting point is about values-aligned nations as well as democratic leadership. Is this a specific AI issue, that companies need to think about their democratic responsibilities because of the power of this technology? Or do you think companies more broadly have a role to play in democracies thriving?

It’s not unique to AI, but there are certain industries where this is extremely important to care about, like AI, defense, and energy. Those of us in those industries owe a lot of attention and effort towards supporting the values that we want to see lead around the world. If you care about the world continuing to get better for people, you’re going to have to contribute to it. You can’t pretend that you don’t exist in this broader system, and that you don’t impact it. So I really do believe there’s a duty to protect progress. I really don’t believe in this idea that companies only exist to make money, and that’s all you can afford to care about.