Simplify the Rulebook The AI Act, GDPR, the Data Act, NIS2 - each with its own obligations and timelines - pile cumulative burden on the startups Europe most needs. The Digital Omnibus, tabled in late 2025, is a genuine opportunity to consolidate enforcement and rationalise the data framework. It should be fast-tracked and adopted by mid-2026.

Where Europe Can Win Europe has lost the generative AI race - only Mistral has produced notable foundation models. Pretending otherwise is not a strategy. But losing one race does not mean losing the war. Trust as a Moat US labs are going closed-source; China has embraced open-source for global reach. Europe's path is a third way: trusted, open-source AI built to European values - something like what "Made in Germany" once meant for manufactured goods. Own the Distribution The most urgent threat to European AI sovereignty is not the models Europeans use - it is the apps (and soon agents) that control distribution. In virtually every vertical, dominant platforms are American or Chinese. A continent that relies on foreign apps for its digital life has surrendered its AI future to the companies that own those apps. Vertical AI Over 75% of European AI investment already targets specialist applications. In AI for energy, Europe leads the world with 50% of global VC in that segment. Health, energy, defence, and financial services are precisely where European expertise creates durable advantage. The Next Wave: World Models Systems that understand the physical world - enabling robotics, autonomous systems, manufacturing and logistics - represent a frontier where no one has yet established dominance. Europe is at parity with the US in the talent required. This is where Europe must concentrate its ambition, before this window closes as the GenAI window did.