AI Datacenters Plan 181GW of On-Site Gas Plants Amid Grid Delays
New research finds AI datacenter developers in the US and Europe are proposing massive on-site fossil fuel plants due to grid connection delays.
Developers of artificial intelligence datacenters in the United States, United Kingdom and mainland Europe are planning to build on-site natural gas power stations at a scale that dwarfs national energy benchmarks, according to new research published by The Guardian.
The proposed gas-fired capacity totals 181 gigawatts — nearly four times the entire peak electricity demand of Great Britain. Developers say the fossil fuel plants are necessary because delays in securing connections to national power grids threaten to stall construction and operations at a critical moment for the AI industry.
Read more Cyberattacks on Critical Infrastructure Raise Epidemic Fears →
The findings raise significant concerns among climate researchers and energy analysts. On-site gas generation bypasses the cleaner energy mix increasingly present on national grids, locking in carbon-intensive infrastructure for facilities expected to operate for decades. Critics argue the scale of proposed capacity underscores a tension between the technology sector's climate commitments and its accelerating energy appetite.
The research highlights a structural bottleneck: grid operators in many countries have years-long queues for new large-scale connections, leaving datacenter developers — under commercial pressure to bring facilities online quickly — turning to private generation as a workaround. The consequence, analysts note, is a potential surge in direct greenhouse gas emissions tied directly to AI infrastructure expansion.
Continue reading at Business | The Guardian
