Power Usage Effectiveness (PUE)
Not all of a data centre’s electricity reaches its servers. A large slice is spent keeping them cool, converting and backing up their power, and lighting the halls around them — energy that does no computing at all. How large that slice is has become the industry’s headline efficiency number.
That number is Power Usage Effectiveness. PUE is the ratio of the total energy a data centre draws to the energy its IT equipment actually uses — a measure of how much overhead sits on top of the computing.
Power Usage Effectiveness (PUE) is a data centre’s total energy divided by the energy used by its IT equipment. A PUE of 1.0 would mean every watt reaches the servers; a PUE of 1.5 means half as much energy again goes to cooling, power conversion, and other overhead. Defined by ISO/IEC 30134-2, it is dimensionless and always at least 1.0 — best-in-class facilities reach about 1.1, while the global average has sat near 1.5 for years. It measures overhead efficiency only, not how clean the electricity is.
Definition — The Data Centre Overhead Ratio
Power Usage Effectiveness (PUE) is the ratio of the total energy consumed by a data centre to the energy consumed by its IT equipment — the servers, storage, and networking that do the actual computing. Everything else the facility uses — cooling and chillers, uninterruptible power supplies and their conversion losses, lighting, and building services — is overhead, and PUE expresses how large that overhead is relative to the useful load.
It is the most widely used efficiency metric in the data centre industry, introduced by the Green Grid consortium and standardised internationally as ISO/IEC 30134-2. Its appeal is its simplicity: a single dimensionless number, always at least 1.0, that captures in one figure how much of a facility’s power is “wasted” on everything that is not computing. A PUE approaching 1.0 is the ideal — a data centre where almost all the electricity reaches the IT.
PUE matters for carbon because data centres are large and growing electricity consumers, and overhead energy carries the same grid emissions as the computing it supports. Cutting PUE means the same computing is delivered for less total electricity — and less electricity means fewer emissions, for a given grid. It is also embedded, quietly, in higher-level models: the energy a Sustainable Web Design estimate assigns to the data-centre segment already includes a PUE overhead.
Definition at a glance
| What it is | Total data-centre energy ÷ IT equipment energy |
|---|---|
| Range | Dimensionless, always ≥ 1.0 (1.0 = ideal) |
| Overhead counted | Cooling, power conversion/UPS, lighting, building services |
| Defined by | ISO/IEC 30134-2 (originally the Green Grid) |
| Global average | ≈ 1.5 (Uptime Institute); hyperscale ≈ 1.1 |
| Does not measure | IT efficiency, grid carbon, or water use |
The Formula and How to Read It
The definition is a single ratio:
PUE = Total Facility Energy ÷ IT Equipment Energy
A PUE of 1.5 reads as: for every 1.0 kWh delivered to the IT equipment, the facility draws a total of 1.5 kWh — the extra 0.5 kWh being overhead. Equivalently, only two-thirds of the total energy (1.0 of 1.5) does any computing. A PUE of 2.0 means the overhead equals the IT load: half the energy computes, half is spent supporting it. And a PUE of 1.0 — never quite reachable — would mean zero overhead, every watt reaching the servers.
Because it is a ratio of energies measured over the same period, PUE is best assessed over a full year, capturing seasonal swings in cooling demand. A figure quoted at a favourable moment — a cool night, a lightly loaded hall — can flatter a facility that performs far worse annually.
Typical PUE Values
PUE spans a wide range in practice, from near-ideal hyperscale facilities to inefficient legacy server rooms. Representative values:
| Facility | Typical PUE | What it means |
|---|---|---|
| Theoretical ideal | 1.0 | All energy reaches the IT; zero overhead |
| Best-in-class hyperscale | ≈ 1.1 | ~10% overhead — advanced cooling, high load |
| Modern efficient facility | 1.2–1.4 | Well-run, purpose-built |
| Global average | ≈ 1.5–1.6 | ~50% overhead (Uptime Institute) |
| Older / small / mixed-use | 2.0+ | Overhead as large as the IT load |
The striking fact is how little the industry average has moved: despite a decade of efficiency drives, the global figure has hovered near 1.5, because most of the world’s computing still runs in older and smaller facilities, not the headline-grabbing hyperscale sites that reach 1.1. The gap between the average and the best is where most of the opportunity lies.
What PUE Does Not Tell You
PUE is genuinely useful, but it is narrow — and mistaking it for a measure of “green” is the most common error. Three things it deliberately ignores:
- IT efficiency. PUE measures overhead relative to IT load — not whether the IT load itself is useful. A hall of idle servers doing nothing can post an excellent PUE; adding overhead-free but wasteful computing can even improve it. PUE says nothing about work done per watt.
- Grid carbon. PUE is an energy ratio, not a carbon one. A facility at PUE 1.1 on a coal-heavy grid can emit far more per unit of computing than one at PUE 1.6 on hydro. Carbon depends on the electricity’s emission factor, which PUE does not see — that is what carbon usage effectiveness adds.
- Water. Some of the cheapest ways to cut PUE — evaporative and water-based cooling — raise water consumption. A low PUE can hide a heavy water footprint, which is why water usage effectiveness is tracked alongside it.
The link to real emissions runs through energy: the data-centre energy in a Sustainable Web Design estimate — around 0.055 kWh per GB [GreenCalculus digital.web.swd.v4.operational.data_centre_kwh_per_gb · SWD MODEL · v2026.203] of data served — is higher than the servers alone would draw precisely because it carries a PUE overhead. But turning that energy into CO₂e still needs the grid factor. PUE gets you from IT energy to total energy; it does not get you to carbon on its own.
The Metric Family: DCiE, CUE, WUE
PUE anchors a family of data-centre efficiency metrics, each closing one of the gaps above:
| Metric | What it measures |
|---|---|
| PUE | Total energy ÷ IT energy (overhead efficiency) |
| DCiE | Data Center infrastructure Efficiency — the reciprocal, 1 ÷ PUE, as a percentage |
| CUE | Carbon Usage Effectiveness — kg CO₂e per unit of IT energy (adds the grid) |
| WUE | Water Usage Effectiveness — litres of water per unit of IT energy |
| ERE | Energy Reuse Effectiveness — credits waste heat reused off-site |
All are standardised in the ISO/IEC 30134 series. Read together they give the full picture PUE alone cannot: how efficient the overhead is (PUE), how clean the energy is (CUE), and what it costs in water (WUE).
Common Confusions
- Treating a low PUE as “green”. PUE measures energy overhead, not emissions. A low PUE on a dirty grid can be worse for the climate than a higher PUE on clean power — carbon needs CUE and the grid factor.
- Thinking PUE reflects IT efficiency. It measures overhead relative to the IT load, not whether that load is doing useful work. Idle servers can post a great PUE.
- Comparing spot figures. PUE varies with season and load. Only an annualised figure, measured consistently, is comparable between facilities.
- Ignoring the water trade-off. Water-based cooling can cut PUE while raising water use — check WUE alongside it.
- Reading PUE below 1.0. It is mathematically impossible: total energy always includes the IT energy, so PUE is always at least 1.0.
Frequently Asked Questions
Power Usage Effectiveness (PUE) is the ratio of the total energy a data centre consumes to the energy used by its IT equipment alone — the servers, storage, and networking. Everything else the facility draws, such as cooling, power conversion and UPS losses, and lighting, is overhead, and PUE captures how large that overhead is relative to the computing. It is the data centre industry’s standard efficiency metric, introduced by the Green Grid and standardised as ISO/IEC 30134-2. PUE is dimensionless and always at least 1.0: a value of 1.0 would mean every watt reaches the IT, while the global average sits near 1.5.
PUE is total facility energy divided by IT equipment energy, measured over the same period — ideally a full year to capture seasonal cooling swings. If a data centre draws 1,500 kWh in total and its IT equipment uses 1,000 kWh, its PUE is 1.5. Read that way, a PUE of 1.5 means that for every unit of energy reaching the servers, half a unit more is spent on overhead; only two-thirds of the total energy does any computing. A PUE of 2.0 means overhead equals the IT load. Because it is a ratio of energies, PUE is dimensionless and cannot fall below 1.0, since the total always includes the IT energy.
The theoretical ideal is 1.0, where all energy reaches the IT and nothing is lost to overhead — unreachable in practice. Best-in-class hyperscale data centres achieve around 1.1, a roughly 10% overhead, through advanced cooling and consistently high load. Modern purpose-built facilities typically fall between 1.2 and 1.4. The global average, reported by the Uptime Institute, has sat near 1.5–1.6 for years, because most computing still runs in older and smaller facilities rather than the most efficient sites. Anything above 2.0 indicates that as much energy is spent on overhead as on the computing itself, typical of older or mixed-use server rooms.
No — this is the most common misreading. PUE is an energy-overhead ratio, not a carbon metric. A data centre with an excellent PUE of 1.1 running on a coal-heavy grid can emit far more per unit of computing than one at 1.6 running on hydro or wind, because carbon depends on the electricity’s emission factor, which PUE does not capture. Turning energy into emissions needs the grid intensity — the job of Carbon Usage Effectiveness (CUE), which multiplies IT energy by the grid’s carbon factor. A low PUE means less total energy per unit of computing, which is good, but the climate impact of that energy still depends entirely on how it was generated.
PUE measures energy efficiency — total facility energy divided by IT energy — and stops at kilowatt-hours. Carbon Usage Effectiveness (CUE) goes one step further and measures emissions: the total CO₂e from a data centre’s energy per unit of IT energy, which means it incorporates the carbon intensity of the electricity supply. Two facilities with the same PUE can have very different CUE if one runs on renewable power and the other on fossil generation. Both are defined in the ISO/IEC 30134 series, along with Water Usage Effectiveness (WUE); together they describe a data centre’s efficiency in energy, carbon, and water respectively.
Need someone who does this? 6 carbon accounting & inventory providers in our directory →