The UK government's multibillion-pound tech deal with the United States, signed during Donald Trump's visit, has sparked debate over the environmental impact of new datacentres needed for artificial intelligence. Nvidia boss Jensen Huang advised UK ministers to burn more gas to meet energy demands, conflicting with Labour's net-zero goals.
The agreement welcomes vast energy-hungry datacentres built by US companies, driven by generative AI's growing need for data. Labour hopes these will boost economic growth and public sector productivity, but critics warn of insufficient discussion about downsides, including reliance on a few powerful American firms.
Former deputy prime minister Nick Clegg described the deal as “just another version of the United Kingdom holding on to Uncle Sam’s coat-tails.” A 2023 MIT paper called for a more deliberate approach, warning that generative AI's “unfettered growth” incurs significant environmental costs.
The International Energy Agency expects global datacentre electricity consumption to more than double by 2030, reaching 945 terawatt hours—equivalent to Japan's current demand. Another study found larger AI models use up to 4,600 times more energy than simpler ones, with rapid adoption potentially causing a 24-fold increase in electricity use.
Alongside the tech deal, the UK agreed on nuclear cooperation, including small modular reactors, but these may take years to build. Datacentres also require water for cooling, and the Environment Agency has warned that AI's rapid expansion makes future water demand forecasting impossible.
Research for the Department for Energy Security and Net Zero offers a reassuring comparison of AI energy use to analogue alternatives, but it remains unclear whether AI reduces overall energy demand. The technology's adoption coincides with rising electricity needs from electric vehicles and heat pumps, leaving unresolved questions about reconciling competing demands.
