Land and data-center sites, North Carolina
Guides

What is PUE (power usage effectiveness)?

PUE, or power usage effectiveness, is the total energy a data center uses divided by the energy its IT equipment uses. A lower number is more efficient, and the theoretical minimum is 1.

See data-center power requirements

What PUE means

PUE means the ratio of a data center’s total facility energy to the energy used by its IT equipment. IT equipment includes servers, storage, and network gear. Total facility energy adds everything else that keeps that equipment running, such as cooling, lighting, power conversion losses, and offices.

The metric is defined in an international standard, so operators calculate it the same way.

How PUE is calculated

PUE is calculated by dividing total facility energy by IT equipment energy over the same period.

PUE = total facility energy ÷ IT equipment energy

A PUE of 1 would mean every unit of energy reaches the IT equipment and none goes to overhead. That is the theoretical minimum, and no real facility reaches it. Real facilities report a number above 1.

A measurement over a full year gives a fairer figure than a single day. Cooling load changes with the seasons, and a winter reading flatters the result.

What raises or lowers PUE

Cooling design and climate raise or lower PUE more than any other factor, because cooling is usually the largest overhead.

  • Cooling design. Evaporative, closed-loop, air-cooled, and liquid designs each draw different amounts of electricity. Coolant run close to the chips can cut fan and chiller energy.
  • Climate. Cool, dry air lets a facility use outside air or dry coolers for more of the year. Hot, humid weather makes chillers work harder.
  • Power conversion. Transformers, backup power systems, and distribution equipment lose some energy as heat.
  • Load level. A facility running far below its design load often has a higher PUE, because fixed overhead is spread over less IT energy.

What PUE does not measure

PUE does not measure water use, carbon, or how much useful work the computers do. A facility can lower its PUE by switching to evaporative cooling, which uses more water and less electricity. That tradeoff is why water and energy are read together. See data center water usage.

PUE also says nothing about the servers. An old, inefficient server fleet can sit inside a building with a low PUE.

Why PUE matters for a site

PUE matters for a site because it sets how much power the utility must deliver beyond the IT load. A higher PUE means more total power for the same computing. See data center power requirements.

Climate and water decide which cooling designs a site supports, and cooling design drives PUE. A site with a mild climate, or with water for evaporative cooling, gives a buyer’s engineer more choices. A site without public water pushes the design toward dry cooling and more electricity.

The buyer’s mechanical engineer sets the design. The utility confirms whether it can serve the total load. See what we check for power and substations and what we check for water and sewer.

What the public record shows

The public record shows climate data, water and sewer service areas, and nearby transmission. It cannot show a future facility’s PUE, because that depends on a design that does not exist yet. Operators report PUE for existing facilities themselves, and the method behind a reported figure is not always public.

Questions

What is a good PUE?

Lower is better, and the theoretical minimum is 1. What counts as good depends on climate, cooling design, and the age of the facility, so compare facilities under similar conditions.

Does a lower PUE mean a greener data center?

Not by itself. PUE measures energy overhead only. It does not measure water use, the source of the electricity, or carbon.

Who calculates PUE?

The operator measures total facility energy and IT energy and calculates the ratio. Design-stage estimates come from the mechanical and electrical engineers.

Tell us what the project needs

Load, acreage, counties, and timeline. We reply with how we would run the search and where we would start.