A text prompt uses a few drops of cooling water, about 45 mL over its whole life cycle
Google and OpenAI report about 0.3 mL per typical prompt. Adding power-plant water gives 1.35 mL; Mistral's whole-life figure is 45 mL.
Water: one AI prompt vs everyday things
Millilitres of water. Log scale: each gridline is 10× the previous. Blue = AI.
What this measures: How much water one AI text prompt uses, next to everyday activities and products.
Scopes differ, so compare like with like. The AI prompt, toilet and shower rows are direct use. Coffee, beef and jeans are supply-chain or life-cycle footprints (coffee and beef include rainwater taken up by crops and pasture); the fair AI comparison for those is the life-cycle row (Mistral, 45 mL). Long-document, reasoning and agentic requests use far more than a median text prompt.
Show data table and sources
| Item | Value | Note |
|---|---|---|
| One AI text prompt (Gemini median, on-site) | 0.26 mL | data-centre cooling |
| Same prompt incl. power-plant water | 1.35 mL | our calculation, US average |
| One AI response, full life cycle (Mistral Large 2) | 45 mL | incl. hardware manufacturing |
| 2024 “bottle per email” estimate (GPT-4) | 519 mL | assumed ~140 Wh per email |
| One toilet flush (US standard) | 6.06 L | direct use |
| One 10-minute shower | 94.6 L | direct use |
| One cup of coffee | 140 L | supply-chain footprint |
| One quarter-pound beef patty | 1.75 m³ | supply-chain footprint |
| One pair of jeans, cradle to grave | 3.78 m³ | life-cycle footprint |
- How much energy does Google's AI use? We did the math (Google Cloud)
- Our contribution to a global environmental standard for AI (Mistral AI)
- 2024 United States Data Center Energy Usage Report (Lawrence Berkeley National Laboratory)
- AI water usage, droughts, ChatGPT, Microsoft (reports the 519 mL estimate) (Fortune)
- Residential toilets (US EPA WaterSense)
- Showerheads (US EPA WaterSense)
- The water footprint of coffee and tea consumption in the Netherlands (University of Twente (Chapagain & Hoekstra))
- The green, blue and grey water footprint of farm animals and animal products (Mekonnen & Hoekstra)
- Levi's 501 life cycle assessment results (Levi Strauss & Co.)
How much water an AI prompt uses depends on what you count. There are three levels, and none comes close to a bottle.
At the data centre. Google reports 0.26 mL of cooling water per median Gemini text prompt, about five drops (0.12 mL counting only the chips doing the work). Sam Altman’s figure for ChatGPT, 0.000085 gallons per query, is about 0.32 mL, but he didn’t say what it covers. Microsoft (2026) models 0–0.067 mL of cooling water for a typical query to a frontier model (one of the most capable models).
Adding power plants. Power stations also use water to make the electricity. Using LBNL’s US average of 4.52 litres per kWh, a 0.24 Wh prompt adds about 1.1 mL, for roughly 1.35 mL in total (our calculation).
The whole life cycle. Mistral’s peer-reviewed life-cycle analysis also counts building the servers. It gives 45 mL per 400-token response (tokens are word pieces) from its Mistral Large 2 model.
So the honest range for a text prompt runs from a few drops on site to a couple of spoonfuls over the full life cycle.
What it means
For a short chatbot question, the water used is small however you count it. When you see a figure, check whether it is on-site cooling only or also counts power plants and building the servers, because that alone moves it from drops to spoonfuls. For water, where data centres are built matters more than any single prompt.
Caveats
- These are fleet averages. The same prompt in a hot, dry place can use several times more cooling water, and local water stress is what matters.
- Mistral Large 2 is a different model and setup from Gemini or ChatGPT.
- Heavier requests (long documents, reasoning, agents, video) use proportionally more.
- No company figure here is independently audited.