California lawmakers pass data center water transparency bills as OpenAI CEO's claims debated
OpenAI CEO Sam Altman claimed 38,000 ChatGPT queries use the same water as one almond, but experts state this is impossible to verify due to a lack of public data from the secretive industry. California lawmakers have passed two bills, now awaiting Governor Newsom's signature, aimed at increasing transparency regarding data center water usage and requiring disclosure for permits. This legislative push comes amidst growing concerns over water consumption and local opposition to data center developments across the state.
OpenAI CEO Sam Altman recently claimed that 38,000 ChatGPT queries use as much water as producing a single almond in California, asserting that concerns over AI's water consumption do not withstand scrutiny. He equated the water usage of a large data center to that of an office building. However, experts, including Shaolei Ren, an electrical and computer engineering professor at UC Riverside, and Michael Kiparsky, director of the Wheeler Water Institute at the UC Berkeley Center for Law, Energy, & the Environment, have challenged Altman's statements, citing the impossibility of verification due to the severe lack of public data regarding data center water use.
The debate coincides with California Governor Gavin Newsom's consideration of two bills, authored by Assemblymember Diane Papan, a San Mateo Democrat, which have been passed by state lawmakers. These bills aim to enhance transparency by requiring data center operators to report their water sources and usage when seeking business permits and preventing local governments from approving new data centers unless developers disclose their water plans and fund any necessary water infrastructure upgrades. Governor Newsom had previously vetoed a similar bill, expressing reluctance to impose rigid reporting requirements without fully understanding the economic impact.
Despite opposition from the tech industry and business groups, the legislative momentum is growing amid widespread data center disputes and local government bans across California. Researchers point out that focusing solely on annual water consumption overlooks peak demand periods, which significantly strain community water systems. Studies indicate a substantial increase in data center water use, with projections of up to 150 billion gallons for hyperscale facilities between 2025 and 2030. Experts suggest that adopting less water-intensive cooling systems, despite potential increases in energy use, could reduce water consumption by as much as 50%.