Urgent.News

What's breaking now, across thousands of outlets.

AI

Today I learned a single Google Gemini query uses about five drops of water, but that's not the whole truth

Google's per-prompt figure skips the water power plants use

Today I learned a single Google Gemini query uses about five drops of water, but that's not the whole truth

A technical paper from Google estimates that a single Gemini Apps text prompt uses approximately five drops of water for cooling at their data centers. This figure is based on their 2024 average and is not independently verified. Google compares this water use to the energy consumed by watching TV for less than nine seconds and claims that over a year, the energy and carbon footprint of a median prompt has decreased by 33 and 44 times, respectively.

However, the water figure does not account for off-site cooling water consumption, which could be significantly larger. Researchers have criticized Google's method for not including all water usage, such as that needed for generating electricity in data centers, which can be several times greater than direct cooling water use. Google's water consumption has increased by 34% to 10.9 billion gallons in 2025, despite their claims of efficient AI prompt usage.

Written by urgent.news from TechRadar's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Every outlet that reported this, and when →

Read the original at techradar.com →

More in AI

Introducing workload recovery and expanded infrastructure checks on Amazon SageMaker HyperPod

Large training jobs can stall or compute wrong results even when every node looks healthy. Amazon SageMaker HyperPod now adds workload recovery and expanded infrastructure checks that catch these…

  • MultiNodeTrainingValidation monitors small model training on provisioned instances
  • Hung job detection monitors Slurm jobs for stopped progress or failures
  • HyperPod Kubernetes Resource Monitoring Agent checks Kubernetes layer for issues

More from Wednesday 18 November →