In partnership with
France approved six new reactor builds in under 18 months. China commissioned ten reactors last year alone. The UAE's Barakah plant powers an entire national AI strategy. Meanwhile, the UK announced that 140 data centres want 50 GW of grid connections, more than the country's entire peak demand, and has no coherent plan to supply them with nuclear power.
The UK has the ambition. It has the legacy. It even has the sites. What it lacks is a regulatory and planning framework that connects nuclear generation to AI demand at the speed the market requires. Our analysis identifies five structural failures and the reforms that could unlock a £40bn ($51bn, €48bn) opportunity.
The problem nobody's discussing
The Nuclear Regulatory Taskforce published 47 recommendations in November 2025. The government accepted them "in principle." Six months later, implementation plans remain vague. Meanwhile, data centre developers queue for grid access that may never arrive.
Consider the mismatch. Ofgem's demand connections review reveals roughly 50 GW of data centre requests in the queue. Britain's peak electricity demand this year sits at 45 GW. The numbers expose a planning system built for incremental growth confronting exponential demand.
France operates 56 reactors generating 70% of national electricity. EDF is building six more. The French regulatory framework, overseen by the Autorité de Sûreté Nucléaire (ASN), enables fleet deployment through standardised design approvals. China's National Nuclear Safety Administration approved ten new reactors in 2025 alone. Both nations treat nuclear as industrial infrastructure. The UK treats it as a series of bespoke megaprojects.
Hinkley Point C illustrates the consequence. Costs have climbed to £35bn ($44bn, €42bn). Unit 1 will not generate power until 2030. A project conceived in 2008 will take over two decades to deliver its first electron. No hyperscaler can plan around that timeline.
Why the current framework fails
Fragmented regulation. The UK splits nuclear oversight across the Office for Nuclear Regulation (ONR), the Environment Agency, Natural Resources Wales, and local planning authorities. The Taskforce report described an "overly complex" system where "risk aversion" and "processes over outcomes" drive cost and delay. France consolidates authority under ASN. The difference is years of approvals.
No behind-the-meter pathway. In the United States, the Nuclear Regulatory Commission (NRC) is actively reviewing frameworks for nuclear co-location with data centres. Amazon, Microsoft, and Google have signed power purchase agreements with nuclear operators. The UK has no equivalent regulatory pathway for direct nuclear-to-data-centre supply.
Semi-urban site restrictions. Current population density criteria were written for 1960s reactor designs. The Taskforce recommended updating site assessment criteria to reflect modern Small Modular Reactor (SMR) safety characteristics. Until this happens, co-location near demand centres remains effectively blocked.
Engineering solutions that work elsewhere
France: fleet standardisation. EDF's 56 reactor fleet uses three standardised designs. Each new unit benefits from regulatory precedent. The Flamanville EPR, despite its own delays, created a template that Hinkley and Sizewell now follow. France's lesson: approve the design once, deploy it repeatedly.
UAE: integrated national strategy. The Emirates Nuclear Energy Corporation (ENEC) built four APR-1400 reactors at Barakah in under a decade. The secret was a single regulatory authority, a standardised Korean design, and political commitment that did not waver between election cycles. The UAE now explores nuclear-powered desalination and industrial zones.
China: parallel processing. The China National Nuclear Corporation (CNNC) builds reactors while regulatory review continues in parallel, compressing timelines from decades to years. China targets 150 GW of nuclear capacity by 2035. Our analysis shows this approach delivers reactors in five to seven years from approval to grid connection.
What is the UK's biggest barrier to nuclear-powered data centres?
Strategic Purpose: Validates which regulatory reform area resonates most with our audience, informing future policy content prioritisation.
The strategic disconnect
The UK government understands the opportunity. The AI Opportunities Action Plan designated Culham as the first AI Growth Zone. The Advanced Nuclear Framework targets private investment in next-generation reactors. Great British Energy, Nuclear (GBE-N) selected Rolls-Royce SMR as its preferred bidder, with construction at Wylfa starting in 2026.
Yet these programmes operate in silos. GBE-N plans reactors. The Department for Science, Innovation and Technology (DSIT) plans AI zones. Ofgem manages grid connections. Nobody owns the intersection. The result is a 50 GW queue that nobody can serve.
Regulatory evolution already underway
Culham offers a proof of concept. The former Joint European Torus (JET) fusion site has 575 MW of peak grid capacity with a dual-supply resilient connection expected by mid-2026. The UK Atomic Energy Authority (UKAEA) is actively seeking supercomputing and AI tenants. This is the right instinct, but it remains a single site, not a national strategy.
The Taskforce's proposed Commission on Nuclear Regulation, a single decision-making body replacing fragmented oversight, would be transformational. The Prime Minister's strategic steer called for "one-team regulation" and less "wasteful gold-plating." Words need to become statute.
The path forward
Create a nuclear-AI co-location licence. A dedicated regulatory pathway for behind-the-meter nuclear supply to data centres. This exists conceptually in the US. The UK should formalise it.
Implement the Taskforce recommendations by Q4 2026. Every quarter of delay costs the UK market share. France and the UAE are not waiting.
Designate five AI Growth Zones with nuclear supply. Culham, Wylfa, Sizewell, the former Cottam coal site, and Hartlepool. Each has existing grid infrastructure. Each could host co-located facilities within this decade.
Investment implications
The immediate priority is Culham. Its 575 MW capacity, existing nuclear site licence, and government backing make it the lowest-risk entry point for nuclear-AI co-location in the UK.
The strategic value lies in the Wylfa SMR programme. Rolls-Royce SMR's Generic Design Assessment is expected to conclude by December 2026. Early movers who secure power agreements now will benefit from first-of-a-kind capacity in the mid-2030s.
The temporal consideration is urgent. Data centre developers who cannot secure UK grid connections within 18 months will build in France, the Netherlands, or the Nordics instead. The UK's £40bn ($51bn, €48bn) nuclear-AI opportunity has a closing window.
Bottom line
The UK has every ingredient for nuclear-AI leadership: legacy sites, engineering talent, political will, and 50 GW of demonstrated demand. What it lacks is a regulatory system that connects supply to demand at the speed AI requires. France proves fleet standardisation works. The UAE proves integrated strategy works. China proves parallel processing works. The UK must stop treating each reactor as a constitutional event and start treating nuclear as the industrial infrastructure AI demands. The window to secure 2027 capacity is open now. It will not stay open.
Next week: 200 Nuclear-DC Sites Mapped by Crowdsourced Intelligence
Speak your prompts. Get better outputs.
The best AI outputs come from detailed prompts. But typing long, context-rich prompts is slow - so most people don't bother.
Wispr Flow turns your voice into clean, ready-to-paste text. Speak naturally into ChatGPT, Claude, Cursor, or any AI tool and get polished output without editing. Describe edge cases, explain context, walk through your thinking - all at the speed you talk.
Millions of people use Flow to give AI tools 10x more context in half the time. 89% of messages sent with zero edits.
Works system-wide on Mac, Windows, iPhone, and now Android (free and unlimited on Android during launch).
