Artificial intelligence will only moderately increase global electricity demand by 2050, according to a new report.
Generally it’s been said AI will need so much power it will drain grids without more capacity but a report by Energy Intelligence says the estimates are wrong.
They looked at the future impact of AI across the energy sector, including electricity use, policy and the low-carbon transition.
While AI rollout will drive short-term power demand, particularly from data centres, the study suggests fears of a long-term explosion in electricity use are overstated.
Data centre energy demand is expected to rise from 460TWh in 2022 to 1,000TWh by 2026 – a jump of around 117%.
However, Energy Intelligence anticipates this growth will be offset over time by efficiency gains in processors, cooling technologies and optimised training algorithms.
“Our research indicates that forecasts predicting an explosion in energy demand due to AI are likely to be overstated,” said Michael Collins, Director of Energy Transition Research at Energy Intelligence. “While we may see increases in the short term, efficiencies in AI hardware and software and the benefits the technology will bring to the power sector, will likely be significant mitigation.”
The report also highlights how energy supply will influence the development of AI just as much as AI shapes demand.
It points to growing geopolitical considerations, with the US, EU and China positioning their energy systems to meet AI’s power needs. Labour have been looking at fusion to help power predicted AI demands.
Crucially, the report concludes that AI will act as a net accelerator of the low-carbon transition.
It is expected to reduce costs, improve efficiency in power generation and storage and ease the integration of renewables, helping decarbonise energy grids at pace.
Copyright © 2025 Energy Live News LtdELN