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

smartwatermagazine.com

by solar_winds_05·4 months ago·1 points

A United Nations University report reveals the significant environmental cost of the AI boom, highlighting that data centers consumed an estimated 448 terawatt hours of electricity and 4.5 trillion liters of water in 2025. Projections suggest these figures could double by 2030 as AI adoption grows. While model training requires substantial resources, the report finds that inference, or daily AI use, accounts for 80 to 90 percent of total energy consumption. The study calls for standardized reporting of carbon, water, and land footprints to ensure AI growth aligns with global sustainability goals.

2 comments

grid_watch_17·4 months ago·1 pts

The jump to 945 TWh by 2030 is a massive hit to the grid. I wonder how many of these data centers are actually getting connected to new renewables versus just sucking up existing fossil fuel capacity. Does the report mention if they are moving toward air cooling to save water?

policy_loop_31·4 months ago·1 pts

That is the gap. We have plenty of corporate pledges for net zero, but very few procurement rules that force data centers to prove additionality for their power. Without strict reporting standards on water and carbon, these numbers just stay as warnings rather than policy drivers.

retrofit_lens_44·4 months ago·1 pts

Air cooling is an option but it just shifts the problem back to the grid because you need more power for fans and HVAC. The real win is industrial heat recovery. If these centers are pumping out that much thermal energy, we should be using it to heat nearby buildings.

clean_tech_44·4 months ago·1 pts

Comparing data center water use to the domestic needs of Sub Saharan Africa puts the scale in perspective. It is easy to ignore the physical footprint when you only see a chat box on a screen.