Aryana Noroozi
Thank you for reading this post, don’t forget to subscribe!
Artificial intelligence (AI) is often discussed in the context of its innovation and efficiency. But behind every chatbot and AI model is a physical and environmental footprint; this looks like rows of energy-hungry servers housed in data centers that draw on the same power, water and air that California communities rely on.
Never miss a BVN beat. Stay up to speed on the latest BVN news.
A new report from the nonpartisan research group Next 10, An Assessment of California Data Centers’ Environmental and Public Health Impacts, places context and data at the forefront of that footprint.Â
The report examines the electricity use of California data centers from 2019 to 2023, finding that it nearly doubled. It also found that water consumption almost doubled, and the public health costs tied to air pollution from data centers more than tripled. The study also projects what those trends could look like by 2028 as AI usage expands – especially in places with heavy data center build-out like Santa Clara County and the Los Angeles region.
The report doesn’t advocate against data centers or AI outright, but it does raise urgent questions about where these facilities are sited, who bears the health and environmental burdens, and what tools the state has to push operators toward cleaner energy, smarter water use and less-polluting backup power.
Black Voice News spoke with Shaolei Ren, Associate Professor at the University of California, Riverside, and Stephanie Leonard, Research Director for Next 10, about their findings and how AI is reshaping California’s energy and water landscape, and what they want residents and policymakers to take away from their work.
What follows is a Q&A, edited for length and clarity.
BVN: What does the new report on data centers in California analyze and why is it important?
Join the Black Voice News Publisher’s Circle
Join the Black Voice News Publisher’s Circle and help sustain the vital work of local journalism. Your support powers impactful stories and community-driven initiatives that drive real change. Together, we can amplify voices, inspire solutions, and build a more equitable future.
SL: The report looks at the impacts on California from data centers looking at energy or electricity use, water use, carbon emissions, and then public health impacts. … [The authors] looked at the actual impacts from 2019 to 2023 and then forecasted them out to 2028 and they have a low growth and high growth scenario.
BVN: How significant is the potential impact of data centers on California’s resources and what challenges exist around transparency?
SL: “One thing that stood out to me was just how much electricity use they could be consuming by 2028 in the high end … enough to power 2.4 million average American households, which is [an] insane amount of electricity. … But also … the lack of transparency from these companies on how much data is actually out there about how much water and energy they are consuming.”
BVN: Where did you find the data for these findings, and how reliable is it?
SR: We didn’t actually get any data from the data center operators themselves… we mainly rely on two sources, which we think are the most credible ones in the field. So one is from the Lawrence Berkeley National Lab and then we also look at the Electric Power Research Institute report… Based on those, we estimate California’s energy use for data centers.”
BVN: What is one practical step data centers can take to lessen health and environmental impacts?
SR: “I think modernizing the backup [generators] … is something achievable, especially when we are building new data centers. This not only has … health impact reduction … but also allows data centers to use their backup generator for demand response without harming the communities.
BVN: For readers not familiar with AI, can you explain why AI-powered data centers use so much energy?
SR: If you have an AI model … whenever you enter input, they will take your input, translate it into numbers … and they will multiply it with their model parameter, many, many billions of parameters doing these billions of calculations… It’s just like you’re pressing your calculators, but billions of times automatically by the computer. Suppose you do it billions of times, it’s going to be draining your battery almost immediately, and those computers are a lot more power hungry than your simple calculator.