Europe's AI Race Is Moving Beyond Who Builds The Biggest Models

To stay competitive and relevant, Europe’s AI industry is finding that building advanced AI models is only part of the battle. While new open-weight and sovereign models continue to emerge in Europe, even there the landscape is getting competitive. Recently, German search engine Ecosia is moving away from French AI company Mistral in favor of open-weight models delivered through European infrastructure startup Melious AI.

At the same time, Europe’s AI companies are attracting billions in new investment, expanding their foundation model offerings and developing businesses that compete across nearly every layer of the AI market.

For European businesses, the issue is no longer simply whether they can find an alternative to US-based hyperscale foundation model companies OpenAI, Anthropic or Google. Increasingly, they want control over where models run, how their data is handled, what they pay for inference and how easily they can switch between providers. The ability to operate AI on European infrastructure, even when the underlying model might come from somewhere else, is becoming a competitive selling point.

This raises an interesting question about Europe’s ambitions to establish itself as a major global AI power. Does Europe need to develop its own frontier models to compete with the United States and China? Or could the bigger opportunity lie in providing the infrastructure, specialized technology and applications that make AI commercially useful, regardless of who originally developed the models or where they are from?

Balancing AI Sovereignty, Open Models, Cost And Performance

The shift was not about abandoning open-weight AI, which Mistral itself has championed, but about gaining greater flexibility in model selection, inference costs and infrastructure control.

Focusing On Infrastructure Rather Than Building Models

Europe’s Race To Build Foundation Models And Infrastructure

Europe has other approaches to foundation model development.