Jennifer Holly: Britain must prioritise AI growth to remain competitive and secure its AI sovereignty

Jennifer Holly is a penultimate year Classics and English student at Magdalen College, Oxford.

Britain is home to an abundance of innovation that deserves to be a government priority. A stellar example is Google-owned AI company DeepMind, which was founded in London and, resisting the pull of sunny Silicon Valley, is still based in the less sunny King’s Cross. The company, founded in 2010, was the first startup to define its aim as Artificial General Intelligence (AGI) and is now behind most of Google’s AI efforts.

It was an aspiration fit for the most ambitious entrepreneurs of Silicon Valley, and not what we might expect from academics occupying covert offices within sight of the old-world railway stations of King’s Cross St Pancras. The company, however, has become one of Britain’s greatest assets. One of its founders, Sir Demis Hassabis, was jointly awarded the 2024 Nobel Prize in Chemistry for AlphaFold, which uses AI to make breakthroughs with protein structures. He was knighted in the same year.

Contrary to DeepMind’s success, Britain has cultivated an unfortunate reputation that makes it a surprise that Demis Hassabis decided to keep DeepMind in London, even after it was bought by Google for ~$500 million in 2014. The company’s proximity to the railway tracks that propelled Britain to prosperity should be an inspiration for the entire nation to win the AI revolution. But instead, Britain is failing to present a compelling case to keep more of its innovators, by stifling venture capital and domestic AI compute capacity.

Not only do these failures risk the flight of innovators, but if Britain cannot securely power its own innovation in domestic data centres, the nation endangers its AI sovereignty.

I can’t ignore that many are, justifiably, sceptical about AI. It spews streams of slop, and often spits out answers that make you doubt its capability to replace serious jobs. But it isn’t going anywhere, as it’s still improving exponentially. Global investment in data centres for AI is projected to reach $6.7 trillion by 2030. It would be too risky a gamble for Britain to pitch itself behind AI’s curve.

For its improvement, AI requires mammoth funding. Undoubtedly, Google DeepMind couldn’t achieve its growth without an American tech giant propping it up with billions of dollars. Britain resents this. Google DeepMind, in a recent House of Lords report, was only mentioned as a mere example of a UK company tending to “gravitate” towards its source of finance, meaning “we lose some of the benefits for the UK economy”. This ignores the benefits the company still provides the nation, including high-paying jobs for 2,000 of the best innovators. Without American big tech footing the bill, these innovators wouldn’t be here, and wouldn’t have sufficient access to data centres powering cutting-edge research.

It’s unrealistic for Britain to aim to become a cash cow rivalling the US, but it’s unproductive to rest on wilted laurels. The nation must learn to play to its strengths.

Freedom from the labyrinth of EU regulations, including the characteristically bureaucratic AI Act, gives the UK leeway to communicate a more accelerationist outlook on AI that will attract innovation and investment. It’s already doing this. Instead of a regulatory prison, Britain is encasing AI in a regulatory sandbox – the “AI Growth Lab”, and has pledged to cultivate four “AI Growth Zones” to build Britain’s AI strength in North and South Wales, the North East, and Oxfordshire.

But the government needs to put its policy where its mouth is. Amenable regulation provides the scaffolding, but capital gets the construction underway.

The UK’s risk aversity bleeds into its approach to venture capital – in the US, pension funds provide 72 per cent of venture capital, compared to only 10 per cent in the UK. The same House of Lords report commends Mansion House reforms, recognising the need to release sluggish capital. But Britain’s pension capital might spiral down the cultural doom loop – again, manoeuvred towards “safe” assets like real estate. But overall, the report is happily imperative. If capital doesn’t naturally flow to innovation, it threatens stripping tax relief from funds that err too far on the side of caution.

If capital gets the construction going, infrastructure ensures its durability. AI investment moves at a break-neck pace – if Britain now fails to ensure investors that it can power their data centres, its home-grown innovation will be powered by the US, France, or nations further afield.

New data centre projects face 700 GW worth of delays to get connected to electricity grids (to put that into perspective, Britain’s yearly peak demand for electricity hovers around 45 GW). The government has pledged priority to data centres as Nationally Significant Infrastructure Projects (NSIP), but they still must wade through a regulatory swamp and face the highest industrial energy prices amongst advanced economies.

A missed opportunity for AI growth can quickly turn into a national security risk. As AI serves as the bedrock for more industries, including defence, healthcare, and energy, the outsourcing of compute power becomes increasingly dangerous. Reliance on foreign AI infrastructure exposes Britain to the other nations’ whims, relinquishes control of domestic intelligence, and leaves key industries at the mercy of cyberattacks. Even if the government cannot stretch to make economic growth a priority, it must stretch to securing AI sovereignty.

Britain is not lacking in innovation and expertise. Even the civil service boasts British expertise in AI – the AI Security Institute (AISI) has partnered with OpenAI and other AI leaders, garnering attention for its skill in rooting out loopholes that prompt AI to forgo its safeguards.

A loss of domestic AI capabilities would amount to an embarrassment at best, and a sustained national security risk at worst. Britain has cultivated a good attitude towards AI growth, but talk is cheap – this era-defining industry will not stop for the National Grid or HMRC.

To allow Britain’s brain to compete on the world stage, the government must listen to AI leaders and take their ambitions seriously. Hassabis’s goal for AI is “radical abundance”, where all the resources humans need to live well will be in constant supply. Economics struggles to envision what that will entail, or whether it’s possible – but Hassabis wants it for Britain. Google DeepMind launched an ambitious and focused new partnership with the government in December, and alongside commitments to education and responsibility, a new automated research lab will open to focus on materials science – exactly where AI needs to be directed for this “radical abundance” to have a chance.

Many won’t buy into this vision – I’m sceptical of it – but it is imperative to recognise that AI, despite the slop it admittedly produces, will power our industries. The government must create an environment fit to meet the technology’s demands, or risk falling into dangerous irrelevance.