Gartner warns AI data‑center power demand could double by 2030

A Gartner report warns that the rapid growth of artificial‑intelligence computing is likely to hit an energy ceiling. It projects that global electricity consumption by AI data centres could rise from about 565 TWh today to more than 1,200 TWh by 2030, and that installed capacity may increase from roughly 132 GW to 290 GW. AI‑focused servers already account for 31 % of the sector’s power use and are expected to outpace conventional servers by 2027, with a 26 % year‑on‑year growth forecast for 2026. The analysis flags the need for up to $720 billion in upgrades to electricity distribution networks to avoid a bottleneck that could curb AI expansion. In response, industry leaders are emphasizing “frugal AI”‑inspired designs that improve token‑per‑watt efficiency and reduce overall energy footprints.

The findings underline energy security as the new battlefield for AI competitiveness, shifting focus from chip performance alone to the ability to power advanced hardware at scale.