Artificial intelligence has become a universal engine of transformation — a force multiplier for technological progress and a catalyst for new business models across every sector. Today’s technological frontier is shaped by Frontier AI Labs: commercial companies that combine cutting-edge research with rigorous deployment of foundation models, driven by clear go-to-market focus and operational excellence. The leaders are almost exclusively in the United States and China. Without training its own models, Europe risks deepening its strategic dependence on these technologies.

Europe has world-class AI talent – but almost no home-grown frontier labs. With Next Frontier AI, SPRIND is launching a five-to-seven-year initiative to change that – not by entering the current LLM race, but by leapfrogging to the next frontier. This was announced by the Federal Agency for Breakthrough Innovation today at EurIPS in Copenhagen, a European conference officially endorsed by NeurIPS, the most prestigious AI conference globally.

The goal: build at least three European Frontier AI Labs that explore new model classes, modalities, agentic systems and more efficient training regimes, develop their own models and run real products at scale. This is bigger than a single program – it’s a long-term initiative that combines funding, company building and support for large follow-on rounds.