Technology transforms societies. They provide a litmus test of in whose interest a country is run.
Across many of the essentials of a dignified life – energy, water, housing, transport, care – markets have been set up to extract where they should invest and to price for profit where they should meet need, leaving households paying a “privatisation premium” on every bill and fare. This was the argument made in The Productive State,, its content the soul of Andy Burnham’s political economy.
Artificial intelligence now risks becoming the newest entry on that list. Whether readers like it or not, digital technologies – and artificial intelligence specifically – are increasingly embedded in modern work and life. The models on which these systems depend are an essential layer of the economy, and they are owned by a handful of American corporations. The recent shock of the US government placing export controls on Anthropic – which briefly pulled its most capable models from users worldwide before the controls were lifted – is a stark reminder of the geo-strategic dangers of dependency on US Big Tech. As far as that litmus test goes, there is a great deal of room for improvement.
In that context, recent reports that Burnham’s team is preparing a ‘reset’ of the current AI strategy are, on their face, welcome. The Financial Times describes a plan to move away from a model built around courting the American hyperscalers and towards one that makes a more significant effort to build genuinely sovereign alternatives, and puts British workers at its heart.
Yes, there are geo-strategic risks of interdependence.
There are also huge doubts about the public’s current inability to capture, control and share the wealth these technologies generate. In the US there is now a live debate about the federal government taking an ownership stake in frontier AI companies, with some, including senator Bernie Sanders, proposing a public share of up to 50 per cent to seed a sovereign wealth fund for all citizens. Britain’s lack of domestic capabilities means we can’t even begin to have this kind of conversation.
Starmer’s government understood these risks, launching its Sovereign AI Fund in April with £500m to invest in British AI startups. Like other governments, there is now a race to develop ‘sovereign’ AI capabilities to reduce dependency on both the Chinese and US tech stacks. Britain is no exception.
But the approach has been clouded by confusion over what ‘sovereignty’ means in practice. My organisation, The New Contract, reveals in Freedom of Information request responses published today that more than one in three of the companies backed through the Sovereign AI Fund are US-owned – and that the government attached no conditions to these investments, leaving the state helpless to prevent them relocating or being bought out by a Big Tech giant.
The political will has also been lacking to use one of the main weapons in our industrial-strategy arsenal: public procurement. Last week saw reports that a Burnham premiership might seek to strip Palantir technology out of the NHS. Leaving aside the company’s highly contested claims that its technology is ‘life-saving’, it shouldn’t be shocking that the state would support a market for domestic alternatives through its procurement decisions as part of a wider sovereignty strategy. One would suspect it isn’t out of ideological disagreement with Palantir co-founder Peter Thiel that Giorgia Meloni’s far-right Italian government is also distancing itself from the company.
Sadly, reliance on Big Tech has been the norm. Amazon and Microsoft alone control up to 80 per cent of the UK cloud market, and the CMA reports restrictive licensing overcharges Britain by at least £500m a year – yet ministers have kept signing deal after deal with the same US giants, increasingly for AI now too, ceding control to San Francisco and locking us into systems that are punishing to leave. Breaking free will be made harder still by Big Tech’s rapacious lobbying: under Starmer, their access to ministers has dwarfed that of civil society groups.
For an alternative approach we could look across the Channel. A Harvard Business School study found that France’s policy of favouring open source in government procurement was associated with a 9 to 18 per cent annual rise in tech start-ups. Paris now backs home-grown tools – from Visio, set to replace Microsoft Teams and Zoom for 200,000 civil servants by 2027, to its AI champion Mistral – and in 2025 it became the first government to endorse the UN’s Open Source Principles, committing to being “open by default”.
The prize here is a real public stake in the technology of the future – and the building blocks already exist. Britain leads the world in open-source software contributions per head, a £13bn ecosystem that current strategy barely notices. The government should expand public investment in compute, as the Ada Lovelace Institute has urged. There should be hard public-interest conditions on any private access to the data, energy and land we already own; and there should be public support for the platform co-operatives and other democratically-owned models.
Britain can’t go it alone – no mid-sized economy can stand entirely apart from Washington and Beijing in the AI race. But that is no reason to surrender to the fatalism of the status quo. The alternative is to join with other countries – the EU member states, but also Canada, Australia, New Zealand and the democracies of East Asia – and pool our collective weight.
We should collaborate on open models, shared standards and infrastructure so that Britain can shape the terms of the AI age rather than have its fate decided elsewhere. Let’s see if Burnham’s government is up to it.
Content from our partners
Related
