Myth check: “Every ChatGPT query uses a bottle of water”
The ‘bottle’ comes from older estimates that covered 10–50 answers or assumed far more energy. Today's reports: about 0.3 mL per prompt.
The claim in full “Each ChatGPT prompt or response uses about a 500 mL bottle of water.”
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.)
What the data shows
The claim: every question you ask ChatGPT uses up a 500 mL bottle of water to keep data centres cool.
What’s true. AI does use water, both to cool data centres and at the power plants that supply them. The bottle figure came from real research, done when no company published its own numbers.
Where it went wrong. The 2023 paper “Making AI Less ‘Thirsty’” estimated one 500 mL bottle for roughly 10–50 medium-length answers from GPT-3: 10–50 mL each, not a bottle each. It assumed about 4 Wh per request and counted power-plant water too. In 2024, the Washington Post and UC Riverside estimated 519 mL for one 100-word email written with GPT-4. Secondary coverage says this assumed about 0.14 kWh per email, roughly 400–600× the energy per query that OpenAI and Google later reported.
What current data says. Google reports 0.26 mL of on-site water per median Gemini prompt, and OpenAI’s figure works out to about 0.32 mL. Adding power-plant water gives roughly 1.35 mL. Mistral’s full life-cycle figure, which includes building the servers, is 45 mL per 400-token answer: about a tenth of a bottle. For scale, a 10-minute shower uses about 95 litres, the on-site water of roughly 360,000 median prompts.
What it means
A short chatbot question uses a few drops of water at the data centre, and about a millilitre counting power plants. Long documents, reasoning and agents use more. For water, where a data centre is built matters far more than any one prompt.
Caveats
- Long-document, reasoning and agent requests can use 10–100× more than a short prompt.
- Company figures aren’t independently audited.
- The early estimates were reasonable when no company disclosed data, and they rightly stressed that location and power-plant water matter.