The constraint at the center of Europe's AI deliberations is data. The continent faces difficult choices on artificial intelligence, and the sharpest case taking shape in that debate is that data sovereignty is the single domain where Europe can still assert genuine control, and where the growth potential is greatest.
Where the leverage sits in the stack
Every stage of a production AI system depends on the quality, scale, and legal provenance of the underlying data. A jurisdiction that controls the data layer retains real leverage even when it does not own the frontier models running above it. This is the structural position that Europe's existing approach to data governance has, at least partially, preserved.
The "difficult choices" framing signals genuine tradeoffs. Europe can pursue tight regulatory control over data flows: that posture yields sovereignty but creates friction for the infrastructure investment that AI development requires at scale. Loosening that grip to attract larger AI buildouts risks surrendering the one layer of the stack where the continent's position is still defensible.
The growth argument runs on input economics. High-quality, domain-specific European data, properly governed and kept within European jurisdiction, is an upstream asset that cannot easily be replicated elsewhere. If Europe can govern that input without choking off access to it, the argument is that it can participate meaningfully in AI value chains even without leading on frontier model development.
The open question is whether asserting regulatory sovereignty over data can actually be converted into an industrial advantage. That requires infrastructure and a governance framework that developers are willing to build on. The distance between the constraint Europe controls and the growth the current debate promises is where the difficult choices actually live.