UN report: AI data centres to double power and water use by 2030

The United Nations University Institute for Water, Environment and Health warns that AI‑driven data centres will consume 945 terawatt‑hours of electricity annually by 2030, roughly doubling the 448 TWh used in 2023. Water demand is projected to reach 9.3 trillion litres and the land footprint to exceed 14,500 km², while CO₂ emissions could climb to about 400 million tonnes. AI is expected to account for 40 % of total data‑centre power, with most consumption coming from continuous inference workloads rather than occasional training runs.

The report likens the projected electricity demand to the residential consumption of 1.3 billion people in Sub‑Saharan Africa and the water use to the annual needs of the same population. Emissions would be comparable to the United Kingdom’s total output in 2025. "When billions of people interact with AI each day, even a few milliwatt‑hours of energy per query add up to gigawatt‑hours of demand," noted lead investigator Professor Kaveh Madani. He added that AI is "also physical infrastructure: data centres, electricity generation, cooling systems, transmission networks, chips, minerals, land and water."

Experts in New Zealand highlighted secondary impacts, estimating AI could consume 3 % of global electricity, require land the size of the Hawke’s Bay region and produce e‑waste equivalent to 250 Eiffel Towers each year.