Europe is falling behind the US on AI use at work, and the main reason is not technology, according to a new report.The report from Sifted, sponsored by London Business School, says Europe has an organisational gap rather than a tech gap. Firms struggle with skills, management, trust and the cost of fitting AI into how they work.
“Doing a fun thing with ChatGPT and running your whole business on AI are two very different things,” Celonis executive Rupal Karia told Sifted.
The numbers
About 43% of US workers used AI for their jobs in early 2026, against 32% across Europe, a St Louis Fed analysis found. US workers spent more than three times as many work hours using AI as those in Germany, France and Italy.
Only 20% of EU firms use AI at all, up from 13.5% a year earlier, Eurostat data shows. Denmark leads at 42%, while Romania sits at 5.2%.
Even where firms use AI, few have gone all in. The report cites McKinsey data showing 88% of organisations use AI somewhere, but just 7% have fully scaled it.
“There is no inherent roadblock to Europe leading in AI adoption,” said Nicos Savva, a professor at London Business School.
Stuck in pilot mode
Karia, who runs northern Europe and the Middle East for German software firm Celonis, said many companies had underestimated the human side of change. Adoption has been slower than expected, he added.
Executives cited unreliable outputs, cyber risk and running costs as reasons for caution. Some firms are now limiting AI use as they weigh cost against benefit, LBS deputy dean Gary Dushnitsky told Sifted.
Returns are also hard to find. The report cites research from marketing firm Whito showing 75% of firms using AI see productivity gains, but only 12% see more revenue.
Paul Maker, technology chief at UK firm Aiimi, put it bluntly.
“Everyone’s been given a hammer and now they’re desperately looking for nails,” he said.
Fear and slow change
Fear of job losses is also slowing take-up, said LBS professor Lynda Gratton.
“If you feel AI will destroy your work, then you won’t want to adopt it,” she said.
LBS dean Sergei Guriev warned that rushing could backfire if people stop learning to think for themselves. A quick start, he said, could undermine longer-term adoption.
Some of the delay is normal. The report cites Goldman Sachs economist Elsie Peng, who found the PC took about 15 years to show up in productivity data.
Where it is working
Adoption is fastest in law and banking. More than 85% of banks supervised by the European Central Bank now use AI, the report says.
Belgian startup Oper built an AI agent to check mortgage paperwork for underwriter Creafin. It is named after Herman Verdoodt, a Creafin credit technician, who said his team spends far more time fixing data than writing mortgages.
Factories are slower. Many cannot shut down for days to rewire their systems, said Dan Hobbs, chief executive of Irish safety firm Protex AI.
Europe’s own route
The report argues Europe should not try to copy the US on frontier models. More than 75% of advanced AI computing power sits in the US, against less than 4% in Europe, it says.
Instead, it points to startups that place engineers inside client firms, much like Palantir does. It also highlights AI roll-ups, where startups such as Dwelly buy traditional businesses and rebuild them around AI.
Some European AI firms are also choosing to stay independent. UK avatar startup Synthesia reportedly turned down a $3bn offer from Adobe last year, the report notes.
Dushnitsky said Europe’s pitch could rest on control and resilience rather than price.
“Perhaps the European value proposition should be built around resilience, confidentiality and privacy,” he said.
That echoes founders who want smarter AI rules, not fewer. The report concludes that good rules can speed adoption by giving firms the confidence to try.