The £500m plan to stop Britain being America’s ‘AI vassal state’

The White House has also unilaterally paused a “tech prosperity deal” with Britain designed to foster the sharing of information on AI.

If the transatlantic relationship were to break down further, and the White House took more drastic measures, Britain could be left hopelessly exposed.

The UK has no global AI giant it can turn to. DeepMind, the closest thing, was bought by Google in 2014 for around £400m, and many promising other start-ups go hunting for American cash.

“AI is now strategic infrastructure,” says Josh Burch, the co-founder of defence-tech firm Gallos Technologies and a former national intelligence official. “As nation states weaponise technology dependencies, control of platforms, data and supply chains now determines who holds leverage. It has become a primary arena of geopolitical competition.”

“It’s an astonishing level of over-reliance on American tech,” Sir Nick Clegg, the former deputy prime minister, said recently. “It’s a level of dependence which is not compatible with the kind of agency and basic sovereignty that a country like ours should aspire to.”

Ministers are now trying to correct that. Last month, the Government launched a £500m Sovereign AI fund, described as being designed to “secure long-term strategic advantage for the UK”. Earlier this month, it appointed Suzanne Ashman, a venture capital investor and Sir Tony Blair’s daughter-in-law, to run it.

James Wise, the fund’s chairman, says the fund’s goal is to ensure that more AI companies that might be critical to the technology’s future remain in Britain. “If any one of those [investments] is right, at scale, then it fundamentally upends the global AI stack as it is today,” he says.

The plan is not for Britain to develop an AI autarky, shut off from overseas companies.

The UK’s energy costs are too high for any AI lab to entertain developing models on British infrastructure, for example.

But ensuring more stay in the UK could give Britain greater leverage in an increasingly transactional world, as well as boost relations between the Government and tech industry.

The fund is small compared to the astronomical sums being raised by American venture capital investors. Nor do the investments it expects to make – up to £20m or access to state AI computing resources – give it much economic power.

Its investment in Ineffable Intelligence represented a tiny portion of the $1.1bn (£818m) the company raised, and the fund does not force companies to commit to investing in Britain.

“I respect the energy and the intent, but I do not think the model will work at scale,” says Matt Miller, a former partner at venture giant Sequoia who now runs his own fund, Evantic.

“When government capital is injected into a market that already has plenty of capital, it props up companies that could not raise from anyone else, it locks up talent in businesses that should be pivoting or winding down, and it creates what is effectively state welfare for AI.”

Wise says the Sovereign AI fund is “not about how much capital the Government gives you”. “The Government shouldn’t be subsidising at scale billions and billions of pounds. What’s more important is that the things people feel it’s been hard to work with government on are made easier.”

Ministers are ramping up their plans for tech sovereignty. Next month, Kendall is set to outline an “AI hardware plan” designed to take advantage of British expertise in semiconductors.

If Mensch’s two-year timeline is correct, there is not much time to waste.