Home / Technology / AI Boom Threatens Global Water Supplies
AI Boom Threatens Global Water Supplies
31 Aug
Summary
- Data centers could nearly triple water use by 2030.
- Water stress regions face significant conflict with AI growth.
- Cooling technology offers substantial water savings.
The rapid expansion of AI infrastructure is projected to dramatically increase water consumption by data centers globally. By 2030, annual direct water usage for cooling could nearly triple from 222 billion liters in 2025 to an estimated 644 billion liters. This escalating demand is expected to intensify pressure on regions already experiencing high or extreme water stress, which could account for 34% of the sector's total water consumption by the same year.
However, significant water savings are achievable through the adoption of more efficient cooling technologies. Mandating water-saving measures in stressed locations could reduce consumption by as much as 45%. Aggressive water-efficiency strategies could limit annual consumption to around 388 billion liters by 2030, while a moderate approach could see usage at approximately 543 billion liters.
Cooling technology is identified as a critical factor, especially as AI workloads generate more heat. Liquid cooling at the rack level can lessen the burden on facility cooling systems, enabling greater use of dry cooling. While dry cooling conserves water, it may increase electricity demand, shifting some water footprint to power generation. Water efficiency also varies significantly among operators and geographic locations.
