How much water and power do data centres actually use?
It depends on scale, cooling technology and local climate, so a single figure is misleading. What matters for policy is a disclosure requirement: facilities reporting projected and actual electricity use, power usage effectiveness, water usage effectiveness and backup-fuel use, before and after commissioning.
Why a single number is the wrong question
Two facilities of the same compute capacity can differ several-fold in water use depending on whether they use evaporative, air, or closed-loop liquid cooling, and whether they're sited in a humid or arid location. Publishing a national average figure invites developers to argue their project is "typical" rather than disclosing their actual design. The more useful policy tool is a standard set of metrics every significant facility reports.
| Metric | What it measures | Why it matters |
|---|---|---|
| PUE | Power Usage Effectiveness — total facility power ÷ IT equipment power | Shows how much energy overhead cooling and infrastructure add |
| WUE | Water Usage Effectiveness — litres per kWh of IT energy | Makes cooling-technology water demand comparable across sites |
| Peak & projected demand | Grid draw at full build-out, not just at commissioning | Lets grid operators plan transmission capacity ahead of need |
| Backup-fuel use | Diesel or gas generator hours and volumes | Flags air-quality and emissions exposure for host communities |
Requiring these four figures — rather than a single aggregate estimate — gives regulators, counties and communities something they can actually act on: whether a specific facility, in a specific catchment and grid node, is one the area can sustainably host.