A flawed approach to renewable energy is undermining Britain’s AI drive - Chris Hocknell

Yet we shouldn’t be fooled by Labour’s heady yet hollow announcement that it would “rip up the rules’ to allow for easy planning, building and funding of new Small Modular Reactors in the UK.

History tells us that nuclear reactors built in the UK have a habit of being wildly behind-schedule, over budget.

Hinckley Point C is five years later than promised and at least £5bn over schedule. Sizewell C’s costs have ballooned from £20bn to £40bn. It’s hard to believe that a regulatory taskforce, seemingly set up overnight, will be able to cut through the structural regulatory quagmire that British nuclear power faces.

In the meantime, Labour’s plan to become an ‘AI Superpower’ will rely on an increasingly precarious energy system. Government energy policies are undermining Britain's ability to power any heavy industry, including the AI industry. Between the Renewable Obligation certificates, Feed-In Tariffs, Strike Prices and grid balancing charges, Government renewable policies have led to energy prices that are more than double those found in America.

Electricity costs are the defining factor in global AI competition. AI workloads require almost complete power reliability. This makes them particularly ill-suited to more complex grid systems like the UK’s.

The contrast between US and UK approaches to AI infrastructure could not be starker. In the US, the ‘Stargate’ project states (or rather decrees) a $500bn commitment to building 20 massive AI data centres. These facilities will likely have dedicated power plants ensuring reliable, cost-effective operation.

Meanwhile, the government's latest promise of AI is leadership layered on top of its original fantasy of 95 per cent renewable power by 2030. These goals are incompatible, and the reasons are simple.

Let’s consider grid instability. National Grid's balancing costs have quintupled to £2.4bn annually, while industrial standing charges have risen sixfold since 2018.

We then need to combine this with infrastructure costs. The renewable transition requires £54bn in grid reinforcement to accommodate more renewable power, plus it needs to invest in some form of battery storage to provide reliability (expect tens of billions of pounds), with all costs being passed onto the consumer and industry in one form or another.

Then add energy market dysfunction into the mix. The UK's flawed electricity market structure also forces renewable generators to be compensated at gas-plant prices, creating a structural premium that makes UK industrial power 159 per cent more expensive than US rates.

While the US is building AI infrastructure at unprecedented scale, the UK's last major AI company, DeepMind, has quietly shifted its most compute-intensive operations to US data centres. There’s little point in conducting world-beating AI research if it gets commercialised abroad.

The market has already voted with its capital. In 2024, US data centre investments reached $158bn, while UK investments fell to £363m - the lowest level since 2010.

The UK government's simultaneous proclamations of AI leadership and the pursuit of the world's most expensive electricity would be amusing if they weren't so economically damaging.

Chris Hocknell is the director of sustainability consultancy Eight Versa.