In short
Power usage effectiveness (PUE) is a data centre's total energy divided by the energy used by its computing equipment: closer to 1.0 means less lost to cooling and power conversion. Water usage effectiveness (WUE) is the litres of water used on site per kilowatt-hour of computing: lower is better. Neither says how much computing is done, how clean the power is, or whether the work was needed.
Steps
- Check whether the figure is for one site or a company-wide average.
- Check the period covered, ideally a full year, because cooling needs change with the seasons.
- Read PUE and WUE together, because a design that saves water can use more electricity.
- Pair them with grid carbon intensity to see the carbon picture.
- Ask how the figures were measured and whether they are reported to a regulator.
PUE and WUE measure a data centre's overheads, not the efficiency of the AI running inside it. They are useful for comparing sites, but only alongside other figures.
What is PUE?
PUE is total facility energy divided by IT equipment energy. A PUE of 1.0 would mean every kilowatt-hour reaches the computers; anything above that goes on cooling, lighting and power conversion. The Uptime Institute's 2025 global survey found that average PUE levels showed little change for the sixth consecutive year.
What is WUE?
WUE is on-site water use, in litres, divided by IT energy in kilowatt-hours. It usually covers cooling water used at the site, not the water used by the power stations supplying it.
Why read them together?
Cooling involves a trade-off. Evaporative cooling uses water but can save electricity; air or closed-loop cooling can save water but may use more power. A very low WUE with a high PUE, or the reverse, may simply reflect that choice, and the right balance depends on local water stress and the grid.
Common mistakes
- Comparing a best site with a fleet average.
- Treating low PUE as green. A site with excellent PUE on a high-carbon grid can still have a large footprint.
- Forgetting the work itself. PUE can't tell you whether an oversized model or idle GPUs are wasting the energy that does reach the computers.
In the EU, operators of data centres with at least 500 kW of IT power demand must report energy and water use under Delegated Regulation 2024/1364. See hosting and data centres.
Sources
- Uptime Institute global data center survey 2025, Uptime Institute (2025)
- Commission Delegated Regulation (EU) 2024/1364 on the first phase of the establishment of a common Union rating scheme for data centres, European Commission (2024)
- Data center metering and resource guide, Lawrence Berkeley National Laboratory (2017)
Related
How do I choose a sustainable data centre or host for AI?
What to ask an AI host or data centre: grid and hourly clean power, PUE, water use, hardware lifetimes, location, and per-account energy reporting.
How do I estimate the water footprint of AI?
Estimate AI water use from energy and a data centre's water usage effectiveness, add water used in generating the power, and say where it was used.
How do I reduce the water footprint of AI?
Cut AI's water use by using less compute, choosing hosts and regions with low water stress and good WUE, and timing heavy work where it helps.
Green AI hosting and data centres
How to choose where AI runs: grid carbon intensity, PUE and WUE, renewable energy claims, water use and siting, with questions to ask hosts.