Drawing on insights from industry, policy, and Amazon Web Services research, Sasha Rubel, Head Public Policy for Generative AI, EMEA, analyses the gap between experimentation and transformation risks leaving Europe behind in the global AI race. As Europe reaches a critical inflection point, she outlines why progress is stalling and what must change.
Every artificial intelligence (AI) story in Europe seems to begin the same way: a brilliant idea, a promising pilot, a headline about a breakthrough.
Fewer reach a second chapter. The company that scales, hires hundreds of people, and becomes a category leader from a European base is still the exception, not the rule.
The data is striking. Recent research from Amazon Web Services (AWS), Unlocking Europe’s AI Potential 2026, is based on interviews with 17,000 businesses and 17.000 citizens across Europe and found that 54 percent of European businesses now use AI – up from 33 percent just two years ago.
That is genuine progress. But only 22 percent of adopters have reached advanced use – the stage where AI stops being bolted onto old processes and starts becoming the process itself.
More than half – 58 percent – remain stuck at the most basic level: tools layered onto workflows that haven’t fundamentally changed.
That gap between adoption and transformation is the real story of Europe’s AI moment. It isn’t a story about ambition, which Europe has in abundance, but about what happens after the pilot succeeds: scaling across borders, hiring the right people, and raising the capital to grow. That is where momentum stalls.
Europe was an early mover on AI governance. Clear, understandable, and consistent rules are an advantage when applied consistently.
Today, 68 percent of businesses don’t understand their obligations under the EU AI Act, and the legislation doesn’t operate in isolation.
It sits inside a dense web of EU-wide national data rules, sectoral regulations, and enforcement practices that vary across all 27 EU countries. A single rulebook becomes, in practice, 27 separate compliance exercises.
For a founder expanding, that is not a minor inconvenience. It shows up as 41 percent of overall IT budgets spent on compliance, legal costs, delayed launches and, sometimes, a decision to expand elsewhere first.
Nearly four in 10 European start-ups say they would consider relocating to grow faster in an environment with more predictable regulation and an enabling environment with access to funding and ease of scaling; amongst the highest-growth companies, that figure climbs past half.
When compliance complexity, rather than product quality, decides where a company grows, Europe is competing with one hand tied behind its back.
The second challenge is less visible but just as damaging: a shortage of people who know how to put AI to work.
75 percent of European businesses say their AI capabilities need improvement, with open digital roles currently taking close to seven months to fill.
But the deepest shortage isn’t in machine learning PHDs – it’s in the leadership layer: managers and executives who can redesign a workflow around what AI now makes possible, rather than simply adding AI on top of what already exists.
Without that fluency, adoption stalls in the innovation team, admired in a slide deck but never let loose on the factory floor, hospital ward, or customer service desk.
This shortage will only grow more acute as agentic AI systems capable of planning and executing entire workflows independently move from pilot to mainstream.
The third challenge is money that arrives too late. Europe has gotten better at funding early-stage AI ventures, but scale-up capital, along with larger patient checks that carry a company from proven product to global category leader, remains thin.
That gap hits deep tech founders hardest since their businesses require years of infrastructure investment before revenue can catch up.
The consequence isn’t difficult to see. Founders quietly headquarter their next funding round or next 100 jobs outside of Europe.
Each time that happens, Europe loses more than one company – it loses the tax base, supply chain jobs, and next generation of engineers who might have trained there.
None of this is a case against European AI – it’s the opposite.
We’re already seeing pioneering start-ups in Europe: Dutch start-up myTomorrows, active in 135 countries, is helping over 17,700 patients bridge the gap between medical innovation and access, empowering families with even the rarest conditions to discover and access pre-approval treatments globally.
ETERNO is tackling a different kind of time pressure: doctors often have just seven minutes to understand a patient’s full story, and its AI assistant, LENI, helps them listen better and decide faster.
In France, meanwhile, Quandela is proof that quantum computing doesn’t have to look futuristic to change the future. The company is building photonic quantum machines that run at room temperature and connect through existing fibre networks.
These are the companies who are solving problems that matter, built on European science and talent.
The question is whether Europe has access to the conditions required to house 100 more companies like these on a global scale, or if it will instead continue to produce brilliant first chapters that get finished elsewhere. Indeed, we need to make Europe the best place to scale, not just start.
Europe doesn’t lack ideas, researchers, or ambition. What it lacks is coherence – one predictable rulebook instead of 27 interpretations; an economy-wide investment in AI fluency instead of scattered pockets of it; capital markets willing to back scale with the same appetite they’ve shown during seed rounds.
None of this requires Europe to invent something new: it has already built the world’s most ambitious AI rulebook.
Rather, it requires the seriousness Europe has already shown in writing the rulebook redirected towards making the ecosystem around it work at full speed and scale.
