AI sovereignty in Europe: the market has already answered
When private capital offers roughly 11x the public stake, the question of whether Europe needs sovereign AI is already settled. The real discriminator is data-sensitivity mapping, not model performance.
The European Commission's InvestAI plan mobilises up to €200B by around 2030, including a €20B facility for as many as five AI gigafactories. The call for interest drew 76 expressions of interest, over €230B in proposed private investment, across 60 sites and 16 member states — with each gigafactory targeting more than 100,000 AI chips versus roughly 25,000 for a standard AI factory.
Read as an economist: when private capital proposes on the order of eleven times the public stake, the debate over whether sovereign AI is worth building is over. The market answered. What remains is an allocation and architecture question, not a question of principle.
For a regulated operator the discriminator is no longer raw model performance — those converge. It is the mapping of data sensitivity against use case, which then dictates the target architecture: US API where it is fine, EU sovereign cloud where residency matters, on-prem where it must never leave.
Data localisation and reversibility are moving into procurement criteria. The firms that map this early choose their architecture; the ones that don't have it chosen for them by whoever they signed with.
Analysis by
Cardan-AI Intelligence
Our research and analysis unit, dedicated to applied AI for business, industry and regulatory compliance.
Let's talk about your next competitive edge
A 30-minute conversation to identify your most profitable AI use cases.