Do you know what…?
Do you know how much electricity AI’s data centres could consume by 2030?
Artificial intelligence is often discussed as though it existed purely in the cloud, yet its rapid expansion is increasingly the physical limits of the energy system. The International Energy Agency estimates that electricity consumption by data centres could roughly double from 485 TWh in 2025 to about 950 TWh in 2030, close to 3% of global electricity demand. AI-focused facilities are expected to grow even faster, making grid capacity, transformers and reliable power supply potential . The paradox is striking: software may feel weightless, but the infrastructure behind it is decidedly power-hungry. As demand scales up, operators will have to contend with grid constraints while efficiency improvements try to keep consumption in check.
The wider picture is more nuanced than headlines about an impending energy crunch might suggest. The IEA expects data centres to account for less than 10% of global electricity-demand growth between 2024 and 2030, meaning industry, electrification, electric vehicles and air conditioning remain larger drivers overall. Nevertheless, data centres are unusually concentrated geographically, so their local impact can be far more pronounced than their global share implies. In my view, that distinction is the crux of the debate: the challenge is not simply how much electricity AI uses, but whether grids can deliver it in the right places, at the right time, without creating avoidable . In other words, .