欧洲人工智能现状——看不见的巨人

March 2026 “Europe is at a crossroad in the global AI race,but we are too slow. We have world-class talent,yet without fast, major investment in our owncompute and open-source models, we riskdependence on others. The countries andcompanies that move now will define the nextdecade. Europe is changing, but things are not moving fastenough. Let's build as if Europeʼs future depends onit—because it does.” Fabricio BloisiCEO State of AI in Europe: Executive Summary 1. The AI window is open. It won't stay open forever.AI is reshaping the global economy at a pace the world has never seen. Thecountries and companies that move now will define the next decade. Those that hesitate will spend it catching up. Europe is at thatinflection point - right now. 2. Europe has everything it needs to win. Except urgency and courage.World-class talent. A massive, AI-hungry user base. Arecord $21.8B invested in 2025, up 58% in a single year. The foundations are solid. But foundations don't win races - speed andcourage does. And right now, Europe is not moving fast enough. 3. We're building great companies and then handing them to the US.Europe matches the US at the starting line. But at breakoutstage, US investors outspend Europe 3x. At late stage, it's 9x. Over half of the capital scaling Europe's best AI companies is foreign.We are incubating the future for others to own. That has to stop. 4. The next wave of AI is vertical, applied, and European - if we claim it.General-purpose GenAI? The US won that round. Butthe next frontier - World Models - is wide open. Also 75% of European AI investment already targets high-value applications in health,energy, defence, and fintech. Europe's edge is clear: dominate in the applications layer - health, energy, defence, fintech - anduniquely positioned to lead in World Models. Two massive prizes. One window. The moment is now. 5. Europe doesn't have an innovation problem or a talent problem. It has a capital problem and a friction problem - and bothare fixable.European pension funds sit on €13 trillion, raising their ventures allocation to 3% unlocks €100 billion overnight.Streamline the regulatory stack, scale up EU-Inc into a real tech highway, and Europe's AI pipeline doesn't incrementally improve — itexplodes. State of AI in Europe: Key Insights Adoption & Application Talent & Research Geography & Ecosystems Investment & Startups ●European AI VC funding reached anall-time high of $21.8B in 2025, up58% from 2024 and nearly 10x thelevel in 2016.●AI now accounts for 30%+ of allEuropean VC funding - double theshare of three years ago.●Europe has a comparable number ofAI startups to the US but capturesonly 11% of global AI VC since 2023,vs. 77% for the US.●The funding gap is concentrated atscale: Europe invests 3x less atbreakout stage and 9x less at latestage ($12B vs. $141B in 2025).●Over half of late-stage capital forEuropean AI startups comes fromforeign (predominantly US) investors,raising strategic sovereigntyconcerns. ●London is Europe's undisputed AIcapital: ~1,700 VC-backed AI startupsworth $125B, raising $7.1B in 2025alone - vastly ahead of all other hubs.●Munich and Zurich lead in deep techand AI research; Paris is an emergingforce, particularly in AI modeldevelopment.●London, Munich, Paris, Zurich,Stockholm, Amsterdam, Berlin,Copenhagen, Oxford, and Cambridgeform Europe's core AI hub network -but none yet rivals the scale of BayArea, New York, or Boston.●Defence, Security & Resilience (DSR)is the fastest-growing AI sub-sector:$8.7B invested in 2025, up 55% YoYand 4x higher than five years ago.●The global AI landscape is bifurcatingaround the US and China, withEurope at a strategic crossroads - thesolution, as the data shows, lies inmobilising growth capital at scale. ●Europe is on par with the US infrontier AI talent - researchers activein industry and academia - with~133,000 vs. ~556,000 LLM usersglobally.●However, 53% of European AI talentsits in traditional economy roles vs.40% in the US, and only 33% indigital-native tech companies (vs.46% in the US).●US AI dominance is built on ~50%imported talent, largely European -but this talent flow is showing earlysigns of reversal.●European institutions are absentfrom the top-cited AI publications,and Europe accounts for just 3% ofnew global AI patents (vs. 70% for theUS and 14% for China).●Europe shows genuine competitivestrength in robotics, AImanufacturing, and deep techapplications such as autonomousdriving and AI drug discovery. ●AI adoption in most Europeancountries exceeds US levels, makingEurope the world leader in AIapplication uptake.●Europe records 2x the LLM usage ofthe US, though the dominant modelsare US- and China-built - effectivelyenriching foreign AI ecosystems.●Over 75% of European AI fundingtargets vertical, specialistapplications - in health, energy,fintech, security & defence, and legal.●Europe is competitively positioned inapplied AI sectors: AI for Energy,Security & Defence, and Healthattract str