AI consumes water primarily through data centers and cooling systems, as well as semiconductor manufacturing.
Data centers use cooling systems that consume billions of gallons of water annually, with Google's data centers using 5 billion gallons of fresh water for cooling in 2022.
Semiconductor manufacturing, essential for AI chips, can use up to 10 million gallons of ultrapure water per day.
To make AI more water-efficient, companies can improve cooling technologies, transition to renewable energy, promote transparency and accountability, and innovate in semiconductor manufacturing.