Summary

• Italy’s Domyn is leading EUROPA, the consortium selected by the European Commission to build an open source frontier AI model.
• The model is intended to cover all 24 official EU languages and use European infrastructure.
• The project will test whether Europe can turn public compute, research capacity, and open deployment into systems that organisations can use in practice.

Domyn, the Italian artificial intelligence company formerly known as iGenius, has moved into the centre of Europe’s attempt to build a frontier AI model on European terms.

The Milan-based company is leading EUROPA, the consortium selected by the European Commission as the winner of its Frontier AI Grand Challenge. The project will develop an open source AI model covering all 24 official EU languages, with the Commission presenting the work as evidence that Europe can build advanced AI systems using its own infrastructure, research base, and industrial capacity.

The project is ambitious by European standards. The Commission’s challenge invited proposals for a model with more than 400 billion parameters, a scale associated with the world’s most advanced AI systems. The model is intended to be openly available and designed to perform at the frontier of global AI capability, while making advanced AI more accessible to businesses, researchers, and public institutions across Europe’s linguistic diversity.

The Commission’s selection of the EUROPA consortium gives the project a policy role as well as a technical one. Europe wants model capacity that is not entirely dependent on US or Chinese platforms, particularly where organisations need stronger control over data, hosting, transparency, auditability, and deployment.

Domyn’s positioning fits that need. The company has focused on regulated sectors such as finance, government, and industry, where AI adoption is constrained by governance rather than enthusiasm. Those customers often want frontier capabilities, but they also need to know where data is processed, how models are evaluated, whether outputs can be audited, and how systems can be controlled inside existing compliance structures.

An open source European frontier model could help with some of those constraints. It could allow local deployment, independent evaluation, specialist adaptation, and stronger inspection of model behaviour than closed systems usually permit. Covering all 24 official EU languages would also address a persistent weakness in AI adoption: many high-performing systems remain strongest in English, leaving European public services, smaller markets, and multilingual businesses with uneven capability.

Openness will not be enough on its own. Enterprise use depends on documentation, support, tooling, security, integration, evaluation, performance, and predictable operating cost. A model may be open, but customers still need teams that can deploy it safely, tune it responsibly, monitor it continuously, and connect it to business processes without introducing unmanaged risk.

The compute challenge is just as important. Frontier model development is capital intensive, energy intensive, and operationally demanding. US model companies have built close relationships with hyperscalers and attracted enormous private investment. European public compute, including EuroHPC infrastructure, can give projects such as EUROPA a base, but long-term competitiveness will require sustained funding, reliable capacity, and routes into commercial deployment.

The market risk is fragmentation. Europe has model companies, national AI initiatives, AI factories, open source communities, research institutions, and sector specialists, but customers do not buy ambition. They buy systems that work, integrate, comply, and remain supported. EUROPA’s success will depend on whether it becomes a usable foundation for public services, regulated industries, research organisations, and software vendors, rather than another evidence point in a sovereignty debate.

Its potential value is substantial. A credible European open source frontier model could improve bargaining power with global AI suppliers, strengthen public-sector AI options, support multilingual services, and give regulated companies more control over deployment. Europe’s AI position will not be fixed by one project, but Domyn’s consortium gives the region a practical test of whether sovereignty can be built through working infrastructure and deployable technology.