How AI is Guzzling Up the Water Supply

Tech companies are using a lot more electricity these days, likely due to the ballooning energy costs of AI. But there’s another way data centers, and by extension AI algorithms, tax the planet. They use a lot of water. Overall, Microsoft’s water consumption went up 34 percent in 2022 over the prior year, Google’s 22 percent. In 2023 Microsoft’s entire operations, including all data centers, consumed enough water to fill 15.6 billion water bottles.

You may have heard a statistic that having a conversation with ChatGPT is like pouring out a bottle of water. In this video, we investigate the data behind that comparison and found that it is likely an underestimate. The process of building, training and using AI consumes even more water than the water bottle itself

Proof reached out to the tech companies mentioned in this video for comment; none provided any comment.

Ingredients:

Hypothesis: Tech companies’ water use is skyrocketing due to data center expansion and they’re losing ground on their attempt to reduce their water footprint.

Sample Size: We reviewed individual corporate sustainability reports for 7 major tech companies (Alphabet, Amazon, Meta, Netflix, Microsoft, IBM, and Intel) and data from CDP disclosures, a not-for-profit charity that centralizes corporate environmental disclosures as well as a dozen non-tech Fortune 25 companies like Walmart, CVS, Ford, and GM. We also reviewed white papers and other supplemental documentation published by the companies about their mitigation strategies and academic research on the subject of data center water use.

Technique: We compared total water usage and water consumption per year by company going back as long as the company publicly reported such data. We also used data on water consumption and data center water efficiency compiled by UC Riverside researchers

Key Findings: Microsoft water consumption went up 34 percent in 2022, Google’s 22 percent. In 2023 Microsoft’s entire operations, including all data centers, consumed enough water to fill 15.6 billion water bottles.

Limitations: There are no corporate requirements to report water usage, so some like Amazon do not report water usage at all. Those that do report something often use different units and have their own definitions of water usage and consumption. There is no generally accepted method for separating AI-related resource usage within data centers from other computational requirements.

Why we think news needs an ingredients label
   • What's in your news?  

Sources:
Making AI Less “Thirsty:” Uncovering and Addressing the Secret Water Footprint of AI Models
https://arxiv.org/abs/2304.03271

Corporate sustainability disclosures
Google
https://sustainability.google/reports...
Amazon
https://sustainability.aboutamazon.co...
Meta
https://sustainability.fb.com/2023-su...
Microsoft
https://blogs.microsoft.com/on-the-is...

Related videos:
How AI is Hogging the Electric Grid
   • How AI is Hogging the Electric Grid  

How AI is Killing Climate Goals
   • How AI is Killing Climate Goals  

https://www.proofnews.org/Proofnews.org
  / proof_news  
  / proof__news  

Join us in making trustworthy, verifiable information the new baseline: https://www.proofnews.org/donate