Skip to content

TechnologyReading time 6 min.

What the PUE value measures – and what it does not say

The PUE value compares the energy demand of a data centre to what the IT actually consumes. It thus answers a narrow question precisely – and many obvious questions not at all.

Managed by the Operations department

A power meter with two current clamps on a light background; one clamp encloses a thick black power cable running across the image.

At a glance

  • The PUE divides the total energy of a data centre facility by the energy of its IT systems and is only meaningful as an annual value with known measurement points.
  • Climate and utilisation shift the value without anything changing in the technology: a half-occupied facility usually has a higher PUE than a fully occupied one.
  • The PUE says nothing about the efficiency of the servers, the origin of the power and the water consumption; there are separate metrics for this, such as the WUE.

What the value compares

PUE stands for Power Usage Effectiveness. The EN 50600-4-2 standard defines the metric as the ratio of two amounts of energy: the total energy consumed by a data centre facility, and the part of it that reaches the IT systems – servers, storage systems and network equipment.

PUE = total facility energy ÷ IT equipment energy

Both values are recorded in kilowatt hours over the same period. A value of 1.00 would be the limit where every kilowatt hour reaches the IT. It is unattainable: cooling, conversion losses in transformers and in the uninterruptible power supply (UPS), lighting and building management systems always require a portion of the energy.

Anything beyond 1.00 is overhead. With a PUE of 1.20, for every kilowatt hour of IT energy there are 0.2 kWh for cooling, losses and lighting; at 1.10 it is 0.1 kWh. The step from one value to the other halves the overhead; the total consumption, however, only drops by 8.3% for the same IT load.

Where it is measured and over what period

A PUE is only as reliable as its measurement points. The total energy includes everything supplied to the facility; usually this is predominantly electricity, recorded at the feed-in, where it arrives for IT, cooling and the building together. For IT energy, several points are possible: the output of the UPS, the output of the power distribution or the power strip in the rack.

The further the measurement point for IT energy is from the devices, the more distribution losses are attributed to the IT. This makes the value look better without the facility operating more efficiently. Two values are therefore only comparable if it is known where the IT energy was measured in each case.

The period is equally important. EN 50600-4-2 calculates with energy amounts, not with instantaneous power, and a value is only meaningful over a full year in which summer and winter balance each other out. A snapshot from a cold night shows a value that the same facility does not achieve on an annual average.

Finally, it matters whether a value is measured or calculated: a design value from planning describes what a facility is supposed to achieve, an operating value describes what it achieved in the past year.

Why climate and utilisation shift the value

Cooling usually accounts for the largest variable part of the overhead, and its effort depends on the climate. As long as the outside air is cool enough, it dissipates the heat via dry coolers without a chiller starting up; this free cooling primarily requires energy for pumps and fans. The more hours a year it operates alone, the lower the annual value. With the same technology, a location in a cooler climate therefore performs better.

Utilisation is often overlooked. A part of the overhead occurs regardless of the IT load: the no-load losses of the transformers, the base load of the cooling, lighting and building services. If the facility runs at half load, this fixed portion is distributed over less IT energy, and the PUE rises, even though nothing has changed in the technology.

Example calculation: the same facility at full and half load
Energy per yearFull loadHalf load
IT energy10,000 MWh5,000 MWh
Fixed overhead800 MWh800 MWh
Load-dependent overhead (12% of IT energy)1,200 MWh600 MWh
Total energy12,000 MWh6,400 MWh
PUE1.201.28

The load-dependent part halves with the load, the 800 MWh of the fixed portion remain. A data centre that is just being occupied therefore usually has a higher PUE than the same building at full occupancy.

This results in a peculiarity of the metric: making your servers more efficient can worsen the PUE. If fewer, but better utilised systems do the same work, the IT energy drops, the fixed overhead remains, and the ratio increases. The total consumption has nevertheless fallen – and that is what matters.

What the PUE does not capture

The PUE says nothing about the IT itself, the origin of the power, the water and the waste heat. Four gaps matter when assessing a location:

  • Efficiency of the IT. How much computing work a server performs per kilowatt hour is not in the PUE; an inventory of old, poorly utilised systems can achieve a very good value. Higher supply air temperatures also reduce the cooling effort, but can cause the fans in the servers to spin faster – and their energy counts as IT energy.
  • Origin of the power. A PUE of 1.20 remains the same whether the power comes from coal or hydropower. There are separate metrics for this: the share of renewable energy and the emissions per kilowatt hour of IT energy.
  • Water. Evaporative cooling reduces the power demand for cooling and consumes water in return. A location can thus improve its PUE while its water consumption rises. The metric for this is the WUE (Water Usage Effectiveness): litres of water per kilowatt hour of IT energy.
  • Waste heat. If the heat from the servers is fed into a heating network, this is not credited in the PUE. Whether and how much waste heat is used is shown by a separate metric, the Energy Reuse Factor (ERF).

For ENTRONYX, this list separates two responsibilities. We operate servers, storage and network ourselves, their efficiency is in our hands: we do not replace hardware according to depreciation periods, but when the performance per watt of a new generation justifies the replacement, including the manufacturing emissions of the new device. We contractually source our electricity from renewable sources, proven by guarantees of origin and power purchase agreements. This is a contractual parameter and not a physical specification: the electricity from the local grid flows at the location.

Water consumption and waste heat are part of the building technology of the locations. Water consumption is therefore part of what we check before selecting a location and in our existing portfolio, and the Sustainability page shows it as WUE. Where an annual average is not yet available for a location, it is omitted there rather than being included as an estimated figure. Feeding waste heat into a district heating network is not yet measured at the locations where the platform runs.

How to compare values from different locations

PUE values are only comparable if they were generated in the same way. Check every value stated by a provider or a tender in this order:

  1. Clarify the period: an annual value from operations is comparable, a monthly or instantaneous value is not.
  2. Clarify the measurement points of the IT energy: if they are at different distances from the devices, part of the difference between two values is a matter of measurement.
  3. Clarify whether the value is measured or calculated: a design value does not belong next to an operational value.
  4. Clarify utilisation and climate: a fully occupied location in the north and a half-occupied one in the south differ without one being built worse.
  5. Put WUE and electricity procurement next to it: only together do the three figures show the energy demand, origin of the electricity and water consumption of a location.

We did not build the data centres in which the ENTRONYX platform runs; we select them based on energy efficiency, legal framework and connectivity. A location is only considered if the cooling is based on indirect free cooling, adiabatically supported: outside air cools a closed water circuit via dry coolers, and mechanical cooling only runs when the outside temperature is no longer sufficient for this. The other requirements, from closed cold aisle containment to power supply, are listed on the Data centres page.

Overhead consumption per kWh of IT energy = (PUE − 1) × 1,000 Wh

PUE per location where the platform runs
LocationCountryPUEOverhead consumption per kWh of IT energy
Frankfurt am MainGermany1.14140 Wh
DüsseldorfGermany1.12, provisional120 Wh
HelsinkiFinland1.0880 Wh

The values for Frankfurt am Main and Helsinki are annual averages from operations; for Düsseldorf the value is provisional because an annual evaluation is not yet available. This overview does not show the measurement points of the IT energy.

We select based on energy efficiency; the best value in our portfolio is 1.08 in Helsinki, the highest is 1.14 in Frankfurt am Main. For every 1,000 kWh of IT energy, the difference in overhead consumption is therefore 60 kWh. Latency, electricity procurement, connectivity and certificates per location are listed under All locations.

Rule of thumb

A PUE value is only a comparable figure with the period, measurement points and utilisation. If one of the three is missing, you are comparing numbers, not locations.

Published on 8 November 2026. Prices and product values in this post are taken from the price list and reflect their current status.

Further reading

Read more

Questions for the department

Technical questions about a post are answered by the person responsible for the department — in a conversation, not a presentation. For quotes and framework agreements, sales will lead the initial conversation.