Europe Is Fed Up and Wants Its Own AI

The framing of European AI ambition as a geopolitical response to Washington's policy volatility obscures a harder structural question: Europe isn't just competing with American labs, it's competing with Chinese open-weight models that are already undercutting those labs on price.

That cost pressure is exactly what our coverage of Lindy's migration from Claude to Deepseek (The Decoder, June 26) makes concrete. If a well-funded American startup finds Deepseek's economics compelling enough to abandon Anthropic entirely, European startups building on domestic models will face the same comparison the moment their governments' preferred models ship. Public funding can close a capability gap against US frontier labs, but it doesn't automatically produce the inference cost structure that open-weight Chinese models already offer. European policymakers appear to be optimizing against one competitor while a second, arguably more disruptive one is already in production workloads.

Watch whether any of the announced European AI funding commitments attach explicit cost-per-token targets or benchmark requirements to disbursement. If they don't, the programs are optimizing for political optics rather than market competitiveness.