UK Departments at Odds over Energy Demands of AI Datacentres

Why It Matters

A divergent view on AI datacentre energy use jeopardizes the UK’s ability to meet net‑zero goals while pursuing an AI superpower agenda, creating regulatory uncertainty for investors and policymakers.

Key Takeaways

• DSIT projects 6 GW AI datacentre demand by 2030
• DESNZ forecasts only 528 MW growth for commercial services sector
• Discrepancy represents a ten‑fold gap in government energy estimates
• AI compute could emit up to 123 MtCO₂ over ten years
• Policy misalignment risks UK net‑zero targets and investment certainty

Pulse Analysis

AI datacentres are becoming the new energy frontier, with global estimates suggesting that advanced machine‑learning workloads could consume as much power as small nations. The United Kingdom, eager to position itself as an AI hub, introduced the UK Compute Roadmap in 2025, pledging at least 6 GW of AI‑capable capacity by 2030. This ambition aligns with broader trends where cloud providers and tech giants are building specialised facilities to accelerate large‑scale model training, a move that promises economic growth but also raises substantial electricity and carbon‑intensity questions.

The internal clash between DSIT and DESNZ reveals a deeper governance challenge. DSIT’s aggressive 6 GW target contrasts sharply with DESNZ’s modest 528 MW projection for the whole commercial services sector, a disparity that suggests either a data‑gap or divergent assumptions about AI adoption rates. The revised emissions range—34 to 123 MtCO₂ over a decade—places AI compute at up to 3.4% of the UK’s projected total emissions, a non‑trivial share that could strain the nation’s carbon‑budget 7 and complicate decarbonisation pathways. Without a unified modelling framework, policy makers risk double‑counting or overlooking critical load‑growth scenarios.

For industry players, the uncertainty translates into investment risk. Developers of AI infrastructure need clear signals on grid capacity, renewable‑energy procurement, and carbon‑pricing mechanisms. Coordinated planning—potentially through a joint task force linking DSIT, DESNZ, and the AI Energy Council—could harmonise forecasts, streamline approvals, and ensure that new datacentres are built alongside clean‑energy projects. Such alignment would safeguard the UK’s net‑zero credibility while still attracting the high‑value AI investments needed to stay competitive on the global stage.