Skip to content
by Projet Celsius
Projet Celsius
Key figureWaterUpdated on 11 November 2026

0.26mL

Water for one question to Gemini

Median · written question · Google measurement, August 2025

A median text query to Gemini consumes 0.26 mL of water, about 5 drops, for data centre cooling, according to the measurement published by Google in August 2025.

On-site cooling water only: the water of electricity and manufacturing is not counted.

See the source
Six clear plastic water bottles, full and closed, standing on a table
Water bottles: a question is counted in millilitres. Photo ajay_suresh, Wikimedia Commons, CC BY 2.0

Definition and scope

Water consumed, that is evaporated, by the cooling of Google's data centres for a median text question to Gemini. About 5 drops.

What is included

  • Water evaporated by the cooling towers of Google's centres

What is excluded

  • The water used to produce electricity
  • The water used to manufacture chips and servers
  • Water withdrawn then returned to the environment

How the value is obtained

Google relates the water consumed by its sites to the energy of the question, using the average water efficiency of its fleet.

Limits and uncertainties

  • The narrowest scope of the published figures: 170 times less than Mistral AI's life cycle assessment.
  • Fleet average: a site in a dry area can consume far more.

The order of magnitude

All figures

0.26 mL on a water scale, with another reference figure and 3 values from the linked chart, brought to the same unit. Each tick multiplies by 10.

1 mL1 L1 m³1,000 m³1 million m³1 billion m³0.26 mLQuestion to Gemini0.32 mLOpenAI, ChatGPT10 to 50 mLUC Riverside, GPT-345 mLMistral AI, Le Chat560 billion litresData centres worldwide
1 mL1 L1 m³1,000 m³1 million m³1 billion m³560 billion litresData centres worldwide10 to 50 mLUC Riverside, GPT-345 mLMistral AI, Le Chat0.32 mLOpenAI, ChatGPT0.26 mLQuestion to Gemini
This figureAnother reference figure (link to its entry)A value from the chart "The water of a question, from 0.26 to 50 mL depending on scope"Logarithmic scale

The other published estimates

The chart and its sources

The chart "The water of a question, from 0.26 to 50 mL depending on scope" compares 4 estimates of the same quantity. This entry uses 0.26 mL (status: measurement published by the provider). The differences stem first from what each one counts.

  1. Google, GeminiCooling water evaporated on site0.26 mLSelected value
  2. OpenAI, ChatGPTUnknown: no published method0.32 mL1.2 times more
  3. UC Riverside, GPT-3Cooling and water used to produce electricity10 to 50 mL38 to 190 times more
  4. Mistral AI, Le ChatLife cycle: data centre, electricity, server manufacturing45 mL170 times more

In context

See the chart and its data

The water of a question, from 0.26 to 50 mL depending on scope

Water attributed to a written question to an AI, in millilitres (logarithmic scale)

2023 to 2025

A factor of about 190 separates the lowest, 0.26 mL (Google), from the highest, 50 mL (UC Riverside): it stems from scope and task, not from a calculation error.

0.11× 1010× 10100 mL× 10ScalelogarithmicmL (logarithmic scale)Google, Geminion-site cooling, measured0.26 mLOpenAI, ChatGPTdeclaration, no methodology0.32 mLUC Riverside, GPT-3cooling and electricity10 to 50 mLMistral AI, Le Chatfull life cycle45 mLa factor of 170with Google, by scope
Source: Google; OpenAI; UC Riverside; Mistral AI · projetcelsius.com/en/ai · CC BY 4.0AI Impactby Projet Celsius

Update history

The hub changelog
Answer sentence added at the top of the entry
First version of the entry

More on water

The whole Water topic →

Charts and maps

ChartWaterWater for one question, from 0.26 to 50 mL depending on scope× 170from Google, 0.26 mL, to Mistral AI, 45 mLChartWaterMicrosoft's water use, 2020-202550%of withdrawals in financial year 2025 in water-stressed areasMapWaterWater stress, country by country17.6%world average in 2023Key figureWaterWater consumed by the world's data centresWorld · per year, in 2023 · water consumed, direct and indirect · IEA

Analyses

Where does this figure come from?Water20 October 2026"An email written by an AI is a bottle of water"The comparison comes from a paywalled Washington Post article (September 2024), produced with researchers: we were unable to check its figure, so we do not use it. Their method is published, and providers have since measured the water used by one question.

On the Projet Celsius blog ↗

Blog article ↗WaterHow much water does a ChatGPT query use?

Across the hub

AI Impact DatabaseAll the figures, their methodology and their sources.AI Ranking10 assistants rated out of 100: performance, ethics and environment.AnalysesThe signed pieces: analyses, viewpoints, "Where does this figure come from?".Charts and mapsYou are hereThe reference visuals, with their data and sources.ToolsOpen calculations, your assumptions on the left, the result on the right.

A question about a figure?

An error, a source to add, a methodology question, a press request: write to us. Guillaume Pakula, Clément Reynaud, Sébastien Pierfederici and Alexandre Veilly will reply.

Write to us →

Projet Celsius, which runs this hub, also helps companies measure and reduce the footprint of their AI use.

Accounting for AI in your Bilan Carbone® ↗
Projet Celsius, environmental strategy consultancy

We support companies that account for AI in their Bilan Carbone®

Visiter Projet Celsius ↗
  • Institut Pasteur
  • SNCF Gares & Connexions
  • Radio France
  • Institut Curie
  • Institut Polytechnique de Paris
  • ESSEC Business School
  • Qualiopi, certified process