BRITAIN’S ECONOMY is far weaker for being outside the EU. Its booming AI sector may be a notable exception. So far this year British technology startups have raised over $14.5bn in venture capital, more than all other major European markets combined. Britain has 33 AI unicorns (startups valued at $1bn-plus), more than France, Germany and the Netherlands put together (though France has the biggest in Mistral, which makes continental Europe’s most advanced AI models). They include Wayve, whose driverless cars will be coming to London’s streets through Uber, a ride-hailing app. “It’s definitely a Brexit dividend,” says Keegan McBride of the Tony Blair Institute, a think-tank.
One benefit is that Britain is not beholden to the EU’s stringent AI Act, which critics say holds firms and their customers to burdensome standards. (Unsplash)
One benefit is that Britain is not beholden to the EU’s stringent AI Act, which critics say holds firms and their customers to burdensome standards. The British government has favoured innovation over regulation. Yet while Britain feels freer, in reality any ambitious AI firm will want to sell to the bloc, so must comply with the act anyway, says Alexandru Voica of Synthesia, one of Britain’s biggest generative-AI firms. In some cases, Brexit has made things “harder not easier”. Mr Voica points to steep visa fees and not being able to hire people fast enough as one reason the firm is now opening offices in mainland Europe. On June 9th the British government launched a global-talent scheme to help reduce such frictions.
Great ideas, even when stripped of Brussels’s red tape, still do not always turn into action. “AI growth zones”, designated areas where the government aims to build data centres more rapidly by fast-tracking planning approvals and prioritising grid connections, remain little more than labels. “The UK has a huge strategic advantage [over the EU],” says Pascal Levensohn, a venture capitalist helping to set up an innovation hub in Cambridge. “It hasn’t taken advantage of it.”
If the European comparison flatters Britain, then the American one humbles it. This month the British government celebrated £6bn ($8.1bn) pledged for investment in AI infrastructure. That is pocket change compared with the $700bn the four big American hyperscalers will probably spend on capital investment in 2026 alone.
In April OpenAI, the American firm powering ChatGPT, announced that it would be pausing Stargate UK, an infrastructure project, in part because industrial energy costs are four times higher than in Texas. The recent order by the Trump administration to prevent any non-American from accessing Anthropic’s latest models (notably Claude Fable 5) shows how access to frontier AI is at the pleasure of the American government. “Europe 2031”, an essay by a group of AI experts imagining how the continent will fare if it doesn’t take the technology more seriously, argues that Brexit’s biggest advantage may be that the British can “broker bilateral AI deals with Washington more easily”.
Maik up your mind
Britain still has immense opportunities in AI. A lot of the fundamental science behind today’s models was done by British boffins (one, Geoffrey Hinton, is known as AI’s godfather). Many of the co-founders and top executives at leading labs are British, from Jack Clark at Anthropic to Mustafa Suleyman at Microsoft. Wayve is part of a clutch of promising AI firms massing around DeepMind, Google’s AI company led by Sir Demis Hassabis, and based in London’s King’s Cross. The world will be far less informed about the risks of frontier AI if AISI, the British government’s pioneering AI Security Institute, continues to be locked out of Anthropic’s latest models.
Nor does Britain need to be entirely dependent on America. Jensen Huang, the boss of Nvidia, a chipmaker, talks of the AI stack as a layer cake made up of energy, chips, infrastructure, models and applications. The greatest economic benefits, he says, lie in the top tier—where Britain is best-placed to succeed. Although in the long run access to frontier AI is imperative, most applications do not need the latest models to be useful. Data from Artificial Analysis, a benchmarking firm, suggest that DeepSeek V4 Pro, a Chinese open-source model, offers roughly three-quarters the performance of Fable 5, but at less than a sixtieth of the cost.
To truly prosper in the world of AI, however, Britain must think bigger. It will need to fix the planning system and upgrade the grid in order to build more AI infrastructure. The government may need to stump up more than the £1.1bn it recently pledged for AI hardware. At the application layer, making the most of AI will require solving other governance challenges, such as joining up data in the health system.
Britain will have to do all this in the midst of a growing public backlash. A recent survey by researchers at King’s College London found that 57% of the British public thinks AI will lead to widespread unemployment and 22% that it will lead to civil unrest. In speeches, ministers shift awkwardly between AI’s risks and its opportunities. A better response has been to set up an AI Economics Institute, modelled on AISI, the first institute of its kind to track how AI may affect productivity, the labour market and the wider economy.
In time, the government will have to grapple with social reforms as a result of AI. For now, there is other work to do. “The technological revolution is already here,” says Mr McBride. Britain has an opportunity to be a player, but it risks becoming a spectator.
