**
In a startling revelation, the UK government has significantly revised its estimates on the carbon emissions generated by artificial intelligence datacentres, raising the figures by an astonishing factor of more than 100. This update, quietly released, suggests that these energy-intensive facilities could emit as much as 123 million tonnes of carbon dioxide (CO₂) over the next decade—equivalent to the annual emissions produced by approximately 2.7 million individuals. The implications for the country’s climate strategy are profound, particularly as the urgency to address global emissions grows.
Dramatic Revision of Emission Estimates
The updated figures were disclosed in a revision of the UK’s “compute roadmap,” a strategic initiative aimed at establishing a premier computing ecosystem to support AI innovation. Initially, the government’s estimate for annual emissions from AI datacentres was a mere 0.142 million tonnes of CO₂, a figure now recognised as grossly underestimated and subsequently deleted.
The Department for Science, Innovation and Technology (DSIT) now projects that the carbon footprint of the burgeoning AI infrastructure could range from 34 million to 123 million tonnes of CO₂. This alarming new estimate represents between 0.9% and 3.4% of the UK’s projected total emissions from 2025 to 2035. Crucially, the lower end of this spectrum hinges on advancements in efficiency regarding AI models and hardware, alongside a rapid transition to a decarbonised energy grid.
Energy Demands and Fossil Fuel Dependency
AI datacentres are notoriously power-hungry, requiring far more electricity than traditional datacentres that primarily store data. A significant portion of this electricity continues to be sourced from fossil fuels, exacerbating the climate crisis. Experts are increasingly concerned that the rapid expansion of AI capabilities may lead to a surge in energy consumption, potentially doubling the overall electricity needs of the UK.
This alarming trajectory comes at a time when there is a growing consensus on the need to curb emissions. Patrick Galey, head of investigations at the climate campaign group Global Witness, emphasised the precariousness of the current situation, stating, “We have a handful of years until our carbon budget is exhausted.” He cautioned against squandering the remaining capacity on technologies that may not benefit the broader population, particularly when a staggering 750 million people globally lack reliable access to electricity.
Government’s Dilemma and Accountability
The DSIT’s revision follows findings by the independent watchdog Foxglove and investigative journalism from Carbon Brief, which highlighted the significant underestimation of potential emissions from AI datacentres. Tim Squirrell, head of strategy at Foxglove, pointed out the contradiction between the government’s commitment to achieving net zero by 2050 and its enthusiastic support for the unchecked expansion of AI infrastructure, which could dramatically escalate the nation’s energy consumption.
Despite the gravity of the findings, the government has refrained from commenting publicly. This silence raises questions about accountability and the need for a more transparent approach to energy policy and climate commitments.
Why it Matters
The revised estimates underscore a critical juncture for the UK as it seeks to balance technological advancement with its environmental responsibilities. As the nation pursues ambitious AI initiatives, it must reckon with the substantial carbon footprint that accompanies such growth. Failure to address this issue could not only jeopardise the UK’s climate goals but also undermine public trust in government commitments to sustainability. The urgency for a comprehensive strategy that includes rigorous emissions assessments and a commitment to renewable energy sources has never been more pressing. The future of both the UK’s digital economy and its ecological integrity hangs in the balance.
