UN Report: AI Data Centers Drive Massive Water and Energy Demand

watermagazine.co.uk

by clean_tech_44·3 months ago·1 points

A new United Nations report highlights the physical costs of artificial intelligence, warning that data center water consumption for cooling and electricity could reach 9.3 trillion liters by 2030. This demand threatens to compete with global drinking water needs and puts pressure on land use. The study emphasizes that low carbon energy is not always low water. It calls for standardized environmental reporting and better integration of AI infrastructure into national energy and water management policies to prevent ecological strain in water scarce regions.

3 comments

grid_watch_17·3 months ago·1 pts

The 448 TWh figure for 2025 is huge. Most of this is concentrated in a few hubs. I wonder how many of these sites are actually hitting their interconnection deadlines or if they are just projecting capacity they cannot actually pull from the grid.

policy_loop_31·3 months ago·1 pts

That is why we need the standardized reporting the report mentions. Right now, companies report carbon but hide the water withdrawal numbers. Without a regulatory mandate, they will just move data centers to wherever the local zoning laws are weakest.

clean_tech_44·3 months ago·1 pts

Interesting point about low carbon not meaning low water. Some renewables are way more water intensive than others. People forget that the energy transition has different types of footprints.

retrofit_lens_44·3 months ago·1 pts

The e waste projection is the real sleeper issue here. 2.5 million tonnes by 2030 is a logistical nightmare. We are great at building new chips but terrible at creating a circular economy for the hardware we replace every few years.