Key Takeaways
- The UK government has revised its forecast for annual water consumption by AI data centres, lowering the 2035 projection from 1 trillion cubic metres to a range of 0.1–0.5 trillion cubic metres per year.
- This adjustment arrives amid growing criticism from MPs and environmental groups who question how the Labour administration’s ambition to become an “AI maker” squares with its clean‑power‑by‑2030 and net‑zero‑by‑2050 commitments.
- Campaigners, exemplified by Tim Bierley of Global Justice Now, warn that the expansion of AI infrastructure threatens to unleash massive electricity demand, potentially forcing new fossil‑fuel generation and undermining climate goals.
- The government counters that achieving its clean‑power target would push AI‑related emissions to the lower end of the forecast range, noting improvements in hardware efficiency and the decisive role of grid decarbonisation and AI adoption speed.
- Overall, the debate highlights a tension between technological ambition and environmental responsibility, urging clearer policy alignment, public consultation, and accelerated renewable‑energy integration.
Revised Water Consumption Forecast
The latest government assessment cuts the expected yearly water use of AI data centres dramatically. Earlier estimates had placed consumption at a staggering 1 trillion cubic metres by 2035. The revised figure now falls between 0.1 and 0.5 trillion cubic metres each year over the coming decade. This downward adjustment reflects updated modelling that incorporates anticipated gains in cooling‑system efficiency, water‑recycling technologies, and perhaps a more conservative outlook on AI workload growth. While the numbers remain substantial, the shift signals an acknowledgment that previous projections may have overstated the resource strain posed by expanding AI infrastructure.
Scrutiny Over AI Ambitions and Climate Targets
The revision has not escaped notice from legislators and green advocacy groups. MPs and environmental NGOs are increasingly vocal about the apparent contradiction between the Labour government’s pledge to be an “AI maker” and its legally binding climate objectives: achieving clean power by 2030 and reaching net‑zero emissions by 2050. Critics argue that pursuing large‑scale AI expansion without a clear, enforceable plan to power those data centres with renewable energy risks derailing the nation’s decarbonisation trajectory. The timing of the forecast release—described as a “quiet release”—has further fueled suspicions that the government is attempting to downplay the environmental implications of its AI strategy while still courting investment from major tech players.
Critique from Environmental Campaigners
Tim Bierley, campaign manager at the nonprofit Global Justice Now, delivered a sharp rebuke of the government’s stance. He characterised the revised forecast as an admission that AI data centres constitute a “climate catastrophe,” warning that their electricity appetite could compel the UK to fire up new fossil‑fuel plants at a moment when the country should be eliminating such sources. Bierley warned that the scale of planned AI growth would “drive a coach and horses through the U.K.’s climate goals and plans for the energy transition,” suggesting that the policy is being pursued primarily to curry favour with figures like former U.S. President Trump and Silicon Valley interests. He stressed that the expansion is unfolding without meaningful public consent and urged policymakers to treat the situation as a wake‑up call to reassess priorities.
Government’s Defense and Mitigation Claims
In response, officials have highlighted that meeting the clean‑power target would position AI‑related emissions toward the lower bound of the forecast spectrum. They point out that AI hardware has already become markedly more energy‑efficient, with further gains expected as chip designs advance and workloads are optimised. Nevertheless, the government acknowledges that the majority of emissions tied to AI data centres will be indirect—stemming from the electricity required to power and construct these facilities. Consequently, the ultimate climate impact hinges on two variables: the speed at which the UK decarbonises its energy grid and the pace at which AI adoption accelerates across sectors. Faster renewable integration and slower AI growth would mitigate emissions; the opposite trajectory could exacerbate them.
Implications for Energy Policy and Public Consent
The controversy underscores a broader policy dilemma: how to harness the economic and innovative promise of AI while safeguarding the nation’s climate commitments. Experts argue that any credible AI strategy must be coupled with concrete measures to secure renewable electricity supplies—such as long‑term power purchase agreements, grid‑scale storage investments, and incentives for on‑site generation at data‑centre campuses. Additionally, the episode highlights a democratic deficit; large‑scale infrastructure decisions with significant environmental ramifications are being made without robust public consultation or transparent impact assessments. Moving forward, policymakers may need to institute mandatory climate‑compatibility reviews for AI projects, establish clear timelines for grid decarbonisation that align with AI rollout plans, and create mechanisms for community oversight to ensure that technological advancement does not come at the expense of environmental stewardship or public trust.

