Europe Is Defining Its Place in the AI Race

For years, the dominant narrative about AI in Europe has been one of catching up: fewer frontier labs, less compute, slower capital. What that narrative often fails to notice is what's actually happening beneath the noise, and how much of it is already in motion.

On June 19, the European Commission selected the Domyn-led Europa consortium to build Europe's frontier AI model, under the Frontier AI Grand Challenge, one of the EU's flagship AI initiatives. On the heels of that decision, Domyn CEO Uljan Sharka traveled to Brussels for a series of engagements to discuss where Europe's AI ambitions currently stand: one with Commissioner Ekaterina Zaharieva, and one with Executive Vice-President Henna Virkkunen. Taken together, the Commission's decision and the two conversations paint a clearer picture: Europe is steadily defining its position in AI, on its own terms.
‍

On producing AI and serving it

One of the key themes of the discussion with Commissioner Zaharieva was the distinction between the compute needed to build frontier AI models and the compute needed to serve them at scale. That distinction is often misunderstood, with debates about AI competitiveness tending to conflate the two.

Europe is demonstrably equipped to meet the challenge of producing a frontier model. However, serving that model — inference at the scale of hundreds of millions of people — is a different, larger challenge that actually demands massive new infrastructure. Here, Sharka made the case for an option the US and China don't need to consider in the same way: a decentralized approach, distributing the load across the EU27 rather than concentrating it in a handful of centralized sites, building directly on the foundation laid by EuroHPC and companies like Domyn.

The conversation also turned to research and funding, two areas where Sharka credited Commissioner Zaharieva directly. On research, he pointed to RAISE — The Resource for Artificial Intelligence Science in Europe — as an example of how AI can accelerate scientific discovery itself, from materials science to medicine, an area where European leadership already exists. On funding, the EU's first Scale-up Fund stands to fill a critical gap, drawing a line in the proverbial sand: Europe is prepared to back companies once they've outgrown the start-up stage, rather than watching them raise their growth rounds — and often relocate — elsewhere.
‍

On owning the frontier

With Executive Vice-President Virkkunen, the discussion shifted from how Europe builds AI to why it matters. The mandate behind the Domyn-led consortium is consequential: the Commission called for a frontier AI system exceeding 400 billion parameters, built open-source and supporting all 24 official EU languages, designed to be accessible to European businesses, researchers, public institutions and developers rather than locked behind a handful of providers. As part of the award, the consortium will have access to up to 2.5% of the EuroHPC supercomputing capacity for a year — one of the largest single allocations of AI compute the EU has ever granted.

"This project demonstrates that Europe can develop, control and continuously advance its own frontier AI capabilities," Sharka said. In his discussion with Vice-President Virkkunen, the stakes were framed in similar terms: Europe can lead in advanced AI on its own terms, building a model that competes with the best while staying true to its values — openness, trust, and strategic autonomy built in from the start, not bolted on afterward.

Sovereign infrastructure and ownership were central to the rest of the discussion. For Sharka and the Executive Vice-President, the ability to build, and not merely license, is what will let European businesses, governments and citizens actually innovate on top of AI rather than around it. The two also discussed regulation — not as a drag on competitiveness, but as an advantage. The AI Act, the High-risk guidelines, the Sovereignty Package and, most recently, the Cybersecurity and AI Action Plan add up to a regulatory architecture built around trust, openness and ownership.
‍

What these two conversations add up to

Taken together, they sketch a specific European strategy: a compute plan, a frontier model win, a research agenda already paying off, funding designed to keep growth at home, and a regulatory framework worth defending.

That is what consolidation looks like: not a single, decisive moment but a set of underlying pieces and the conversations that connect them. As they fall into place, they add up to something larger than any one part on its own. Steadily, they're mapping out a unique position for European AI to compete in, with its own values and on its own terms.