As heat and wildfires rage in Los Angeles, attention turns to data centers: is artificial intelligence wasting our water resources?
As wildfires scorch wide areas of Los Angeles, residents and visitors are wondering if they can take simple actions—alongside donations and volunteer efforts—to help prevent the worsening destruction. Tips have circulated on social media urging families to conserve water by reducing their use of ChatGPT, as AI servers require large amounts of water to cool data centers and prevent them from overheating.
Artificial intelligence cannot produce content without relying on massive data centers packed with thousands of high-performance chips capable of processing queries in a short time. According to Daniel Kearney, chief technology officer at Firmus Technologies, a company focused on sustainable operational solutions for AI firms, the chips and computers inside these centers generate a tremendous amount of heat as a result of operating continuously around the clock.
Read also: UAE employs artificial intelligence to enhance cloud seeding
AI historian H.B. Newquist, author of “The Brain Makers: Genius, Ego, and Greed in the Quest for Machines That Think,” says that “running complex AI applications like ChatGPT requires immense computational power that generates massive heat, necessitating the use of water-intensive cooling systems to prevent servers from reaching dangerous temperatures.”
But how much water do we use in this process? According to Kearney, many cooling systems rely on water to absorb and dissipate heat via cooling towers or evaporation methods; some large centers may consume millions of gallons annually. As demand for AI services surges, especially in hot and dry climates like Southern California, water depletion can exacerbate drought impacts and hinder firefighting efforts.
Some believe that refraining from using ChatGPT in Los Angeles could help—even in a small way—reduce water consumption and redirect it toward putting out raging fires. While others might dismiss the usefulness of this individual action, asserting that its impact is limited, Kearney notes that the matter is cumulative: “Every time we postpone unnecessary questions on AI platforms, we save a portion of the water and energy required to run these servers. And if millions of people cooperate in such behavior, we may achieve a noticeable difference.”
Read also: What is open source?
AI companies face mounting challenges regarding the locations of data center construction, as many of these firms now favor colder regions with adequate water resources to reduce reliance on heavy cooling systems. Cooling alternatives that do not rely on water are also being developed—such as immersion cooling, where hardware is submerged in specially designed liquids to dissipate heat.
Currently, as artificial intelligence continues to grow at an unprecedented pace, tech companies may seek to innovate eco-friendly solutions so that the increased use of smart platforms does not lead to the depletion of natural resources. Until these new cooling technologies are ready on a wide scale, personal moderation in using AI platforms during climate crises—such as the Los Angeles fires—remains one possible way to contribute to water conservation.
