AI datacentres plan on-site power stations on huge scale despite climate risks, research reveals

by | Nov 11, 2026 | Top Stories

AI datacentres plan on-site power stations on huge scale despite climate risks, research reveals

Research by BloombergNEF reveals that artificial intelligence datacentre operators are planning extensive on-site fossil fuel power generation capacity to address delays in connecting to national electrical grids. The study found proposals for 181 gigawatts of gas-fired generating capacity across facilities planned in the United States, United Kingdom, and mainland Europe. The US alone accounts for approximately 90 percent of these planned installations.

The projected emissions from this scale of on-site gas infrastructure would total roughly 345 million metric tonnes of carbon dioxide equivalent annually in the US, comparable to the yearly greenhouse gas output from approximately 72.8 million fossil fuel-powered vehicles. The 181 gigawatt capacity represents nearly four times the combined peak electricity demand of Great Britain.

Developers characterize these on-site power stations as temporary solutions while awaiting grid connections, but analysts caution that economic incentives may encourage continued operation once facilities become operational, creating long-term dependency on fossil fuels. Mark Daly, head of datacentres and emerging technology at BloombergNEF, noted that electricity grid connection speed and the composition of new generation sources will be critical factors determining climate outcomes.

The UK datacentre sector submitted more than 100 inquiries during the previous fiscal period regarding gas connections due to grid access bottlenecks, according to Future Energy Networks, the UK gas industry representative organisation. Recent projects including facilities in Buckinghamshire and Bedfordshire have initiated plans for on-site gas-fired power stations. The UK government’s chancellor approved temporary on-site fossil fuel generation for datacentres through a datacentre covenant framework.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI