AI's Climate Cost Grows as Data Centers Use More Energy and Water

Artificial intelligence, once celebrated for its potential to accelerate climate solutions, now faces a quieter but growing climate cost as it becomes an everyday utility. From smartphones to search engines, AI’s energy demand hinges on data centers that process, store and deliver information. Experts warn that while efficiencies are being pursued, the overall impact may persist or even grow unless energy sources and usage patterns change, and unless developers and climate advocates coordinate more closely.

The climate cost of AI rests largely with data centers. When AI is used, queries are sent to centralized servers, and those centers must be cooled with substantial water, often drawing power from fossil-fuel grids in the near term. In practical terms, evaluating energy footprints shows wide variation: a simple AI prompt can consume many times more energy than a non-AI alternative, while a short video or a longer digital task can dwarf an equivalent AI activity. The scale of this energy use matters because it translates into greenhouse gas emissions and water consumption, with local grid reliability and stress on communities living near dense data-center facilities as...