SOURCE CHECKED 2026-10-16 / iea-2025-v1
Water doesn't disappear. So why does consumption matter?
“The water comes back” and “the local water supply is under pressure” can both be true. They answer different questions.
Start with place and time
Imagine taking a litre from a local supply. Returning it to that source soon afterwards is different from evaporating it into the atmosphere. The water still exists, but that does not make it available to the same users at the same time.
This is why a picture of Earth losing its water would be misleading. Our moving water is an illustration of an annual volume, not a measurement of planetary depletion.
Three questions to ask about a water claim
- Is it withdrawal or consumption? Water taken from a source and water not returned to that source are different quantities. Do not add or compare them as if they were identical.
- Which boundary? On-site cooling, electricity supply and semiconductor manufacturing describe different parts of the footprint. A global supply-chain total is not one facility's direct water use.
- Where and when? A global annual number does not establish the impact of a particular source, watershed, season or development. Local evidence may change the conclusion.
A useful reply to a sweeping claim
“Which year and boundary does that number use? Is it consumption or withdrawal, and is there evidence for the local water source?”
A source-stamped receipt makes the annual scale easier to explain. It does not replace those local questions or prove that every data centre causes the same harm.
Make a scoped volume receipt ↗ · Check its source and assumptions ↗
Further reading
USGS: thermoelectric water use and cooling consumption ↗
IEA: data-centre water use, including indirect supply ↗