How much water does AI use?
Ask a cloud AI 20 to 50 questions and it boils away about half a litre of clean water. Not as a figure of speech. A data centre keeps its computers cool by evaporating fresh water, and that water leaves as steam. It does not come back. Teaching GPT‑3 to talk used about 700,000 litres.
GreenCube answers on your own laptop, which is cooled by a fan. So a chat with GreenCube boils away none of it. That is the whole reason this page exists.
One honest warning before you read on: these numbers vary a lot depending on who is counting, where the building is, and the time of year. Anyone handing you one exact figure is simplifying.
Why does AI use water at all?
Servers get hot. Many data centres use evaporative cooling: clean fresh water is evaporated to carry heat away. That water is not returned to the river. It leaves as steam.
There's a second, bigger pile of water most headlines skip: the water used somewhere else, to make the electricity the building runs on. Whether a study counts that or not can change the answer by ten times or more.
The numbers
| What | Water | Source |
|---|---|---|
| ~20–50 ChatGPT queries | ≈ 500 ml | UC Riverside |
| Training GPT‑3 (~2 weeks) | ≈ 700,000 L | UC Riverside |
| Google's US data centres (2021) | ≈ 12.7 billion L | UC Riverside |
For scale: the 700,000 litres used to train GPT‑3 is roughly the water needed to manufacture 370 BMW cars, or about 320 Teslas.
An honest caveat. Not all researchers agree. Some analyses put a single ChatGPT conversation far lower, in the range of 10–25 millilitres, because they count only direct on-site evaporation, use newer efficiency figures, or model different data centres. The half-litre figure is the most widely cited; it is not the only one. The direction is clear and the order of magnitude is real, but treat any exact number with suspicion.
The bigger story is electricity
The water follows the power. These buildings already use about 1.5% of all the world's electricity, and the International Energy Agency expects that to roughly double by 2030, mostly because of AI. More power makes more heat, and more heat needs more cooling.
What actually reduces it
Honest answer: mostly things you do not get a say in: putting the buildings somewhere cold, cooling them with water that gets reused instead of boiled away, cleaner electricity, and smaller AI models that need less power in the first place.
The one thing you do control is where the model runs. A model running on your own laptop evaporates no data-centre cooling water. It draws a modest amount of electricity from a device you're already using. That is not a claim that local AI saves the planet. Your electricity has its own footprint. But it does take your questions out of the cooling-tower equation entirely.
AI that runs on your computer
No warehouse. Nothing boiling off water to stay cool. No server keeping your chats. GreenCube runs the AI on your own computer: chat, read your documents, understand pictures, all offline.
Get GreenCube, €8.99Sources
- UC Riverside, AI programs consume large volumes of scarce water (Shaolei Ren et al., "Making AI Less Thirsty")
- International Energy Agency, Energy demand from AI