Global AI water use in 2025 is not precisely known, but the best available estimates suggest that data centers and AI workloads together used hundreds of billions of liters, likely in the same ballpark as all bottled drinking water consumed worldwide in a year. Public sources frame this as a rapidly growing and poorly monitored footprint rather than a single, agreed‑upon number.

Why there’s no single number

  • Companies do not yet publish standardized, AI‑specific water figures, only partial or aggregated data center water use in sustainability reports.
  • Water use depends heavily on location, cooling technology, energy mix, and whether the model is being trained or just serving user queries.

What recent studies suggest

  • A 2025 academic analysis estimated that AI systems alone use around 312.5–764.6 billion liters of water per year , comparable to all bottled water consumed globally annually.
  • Advocacy groups projecting U.S. AI growth estimate hundreds of billions of gallons per year by the late 2020s , indicating that 2025 is already in the high hundreds‑of‑billions‑of‑liters range worldwide.

Data centers, not “the model on your screen”

  • Most AI‑related water is used indirectly to cool servers and generate the electricity powering those servers , not for the software itself.
  • Individual chat or image requests only account for a tiny fraction of total consumption, but at internet scale and with large training runs, the cumulative impact becomes significant.

Why this became a trending topic

  • As AI use exploded between 2023 and 2025, journalists and researchers began comparing AI’s water footprint to household and national water use, raising public concern.
  • Some widely shared statistics were later criticized as misleading or overstated, sparking forum debates about how to correctly compare “water per query” with everyday activities like streaming or eating a hamburger.

Bottom line and context

  • There is no exact, universally accepted figure for “how much water AI used in 2025,” but credible work points to several hundred billion liters globally, and rising fast.
  • Policymakers and researchers are now calling for transparency standards, location‑sensitive regulation, and more efficient cooling and hardware to keep AI’s water footprint in check.

Information gathered from public forums or data available on the internet and portrayed here.