Lack of water usage transparency at AI data centers raises concerns
A new report highlights the lack of transparency regarding water usage by generative AI data centers, which are increasingly locating in rural, drier areas like Gilroy, California. These facilities strain local water systems, and communities often lack resources to advocate for themselves. Despite a gubernatorial veto last year, legislators are pushing new bills to mandate public disclosure of data center water use.
A recent report by the think tank Next10 and Santa Clara University indicates a significant lack of transparency concerning water consumption by generative AI data centers. These new facilities require substantial amounts of electricity and water, but existing regulatory loopholes allow AI companies to avoid reporting their water usage, placing a burden on California's already strained water resources.
Iris Stewart-Frey, an environmental science professor and lead author of the report, emphasized the need for data centers to disclose water use to local water supply planners and communities. This transparency would enable informed decision-making in areas where water is a critical issue. The report notes that AI companies often choose drier, rural, and less affluent regions, which typically have more fragile water systems, for their data center locations. Gilroy, California, is cited as an example, being a historically agricultural community with a closed groundwater basin that is climatically hotter and drier than the rest of Santa Clara County, making it highly vulnerable.
Despite California's complex water system, the absence of accurate data from AI data centers complicates sustainable water distribution across the state. Last year, Governor Gavin Newsom vetoed a measure that would have mandated water use disclosure from data centers. However, state legislators are continuing to advocate for greater regulation, with two new bills currently introduced to make this data publicly available.