.svg){kind=small}
PUE, or Power Usage Effectiveness, is a widely used data centre efficiency metric that compares the total energy used by a facility with the energy delivered to its IT equipment. It helps operators understand how much additional energy is being consumed by cooling, power distribution, lighting and other supporting infrastructure.
A PUE closer to 1.0 indicates that a greater proportion of the facility's energy is reaching IT equipment rather than being used by the supporting environment. However, PUE has important limitations and should not be treated as a complete measure of data centre sustainability or IT efficiency.
What does PUE stand for?
PUE stands for Power Usage Effectiveness.
The metric was introduced by The Green Grid and has become an industry-standard way of describing facility energy efficiency. It is also formalised within the ISO/IEC 30134 family of data centre key performance indicators.
How is PUE calculated?
The basic calculation is:
PUE = Total Data Centre Energy ÷ IT Equipment Energy
For example, if a data centre uses 1,500 kWh of total energy during a period and 1,000 kWh reaches IT equipment:
1,500 ÷ 1,000 = 1.5 PUE
This means that for every unit of energy used by the IT equipment, another 0.5 units are used by facility infrastructure.
What is included in total data centre energy?
Total facility energy can include electricity used by:
- servers, storage and networking
- cooling systems
- fans and pumps
- UPS losses
- transformer and distribution losses
- lighting
- security and monitoring systems
- other supporting facility equipment
The exact measurement boundary matters. Comparing PUE values is only useful when the measurement scope and method are understood.
What counts as IT equipment energy?
IT equipment typically includes the systems performing the digital workload, such as:
- servers
- storage
- network switches
- routers
- other IT processing equipment
The appropriate measurement point should follow the selected standard or reporting methodology.
What is the best possible PUE?
The theoretical lower limit is 1.0.
At a PUE of 1.0, all measured facility energy would be delivered to IT equipment with no energy used by cooling, distribution losses or other support systems.
Real data centres cannot eliminate every facility overhead, so practical PUE values are above 1.0.
What is a good PUE?
There is no single number that defines a “good” PUE for every facility.
Performance depends on factors such as:
- facility age
- climate
- IT load
- redundancy design
- cooling technology
- building type
- measurement boundaries
In its 2026 Global Data Center Survey, Uptime Institute reported an industry-wide annual average PUE of 1.52. It also noted that newer, larger facilities with modern power and cooling designs routinely report values around 1.3 or lower.
This benchmark is useful context, but comparing a small legacy enterprise facility directly with a new hyperscale site can be misleading.
Why does lower PUE usually mean better facility efficiency?
Imagine two facilities each deliver 1 MW to IT equipment.
Facility A: PUE 2.0
Total site power is approximately 2 MW, meaning around 1 MW is used outside the IT load.
Facility B: PUE 1.3
Total site power is approximately 1.3 MW, meaning around 0.3 MW is used outside the IT load.
Facility B uses much less supporting energy for the same IT power demand.
What affects PUE?
Cooling efficiency
Cooling can represent a substantial part of non-IT energy use. Efficient airflow, containment, chillers, pumps, economisation and control systems can reduce cooling overhead.
See Data Centre Cooling Explained.
UPS efficiency
UPS systems introduce electrical losses. Modern systems can operate efficiently when loaded and configured appropriately, while older or lightly loaded systems may be less efficient.
Transformers and power distribution
Every conversion and distribution stage can introduce losses.
IT load
PUE can worsen when a facility is lightly loaded because some supporting systems continue consuming energy even when IT demand is low.
Climate
Cooler outdoor conditions can provide more opportunities for economisation and reduce mechanical cooling demand.
Redundancy
Highly resilient infrastructure can require additional equipment operating at partial load, which may affect efficiency.
Why can PUE change throughout the year?
Cooling demand changes with weather and humidity.
A facility may therefore have:
- lower PUE during cooler months
- higher PUE during hot weather
- different performance as IT load changes
For meaningful comparison, annualised or regularly measured PUE is generally more useful than a single favourable reading taken on one day.
Why can PUE improve when IT load increases?
This can appear counterintuitive.
Some facility systems have fixed or semi-fixed energy consumption. If more IT equipment is added while facility overhead increases more slowly, a larger proportion of the total energy is attributed to IT and PUE can improve.
This does not necessarily mean the site is using less electricity overall.
A facility can therefore improve its PUE while total energy consumption rises.
Why PUE is not the same as sustainability
PUE measures facility infrastructure efficiency. It does not directly tell you:
- whether the electricity comes from renewable energy
- how much carbon is emitted
- how much water is consumed
- how efficiently servers are utilised
- how much useful computing work is produced
- what happens to retired equipment
A facility powered by high-carbon electricity could have an excellent PUE while still producing significant carbon emissions.
That is why sustainability programmes use additional metrics and management approaches.
What is CUE?
Carbon Usage Effectiveness, or CUE, relates data centre carbon emissions to IT energy consumption.
It gives a different perspective from PUE by considering carbon impact rather than only facility energy overhead.
What is WUE?
Water Usage Effectiveness, or WUE, measures water use in relation to IT energy consumption according to the relevant methodology.
This matters because some highly energy-efficient cooling systems can consume substantial water.
Improving one sustainability metric can therefore create trade-offs elsewhere.
What is Renewable Energy Factor?
Renewable Energy Factor, or REF, measures the proportion of data centre energy supplied from renewable sources within the defined framework.
Operators can use renewable energy procurement alongside efficiency improvements to reduce environmental impact.
Can PUE be below 1.0?
Under the standard PUE definition, a value below 1.0 would indicate a measurement or boundary problem because total data centre energy cannot logically be less than the energy consumed by the IT equipment contained within that total.
Can two data centres with the same PUE have very different efficiency?
Yes.
Two sites with PUE 1.4 could differ significantly in:
- server utilisation
- hardware age
- carbon intensity
- water consumption
- workload efficiency
- total energy consumption
PUE should therefore be combined with other operational and sustainability measures.
How can data centres improve PUE?
Improve airflow management
Prevent hot and cold air from mixing using containment, blanking panels and better rack layout.
Optimise temperature setpoints
Avoid unnecessarily low supply temperatures while staying within appropriate equipment guidelines.
Use efficient cooling systems
Modern chillers, pumps, fans and variable-speed controls can reduce energy consumption.
Use economisation
Suitable outdoor conditions can reduce reliance on mechanical refrigeration.
Improve UPS loading and efficiency
Right-sizing infrastructure and using efficient operating modes can reduce electrical losses.
Retire unnecessary infrastructure
Unused or lightly loaded cooling and power systems can create overhead.
Improve IT utilisation
Although IT utilisation does not directly reduce facility overhead, consolidating workloads can allow inefficient servers and infrastructure to be retired.
PUE and AI data centres
AI workloads complicate data centre efficiency discussions.
High-density accelerators can increase IT energy dramatically, while liquid cooling may reduce some cooling overhead. A facility's PUE can therefore look efficient even as total energy demand grows substantially.
Operators need to consider:
- total energy use
- useful computational output
- carbon
- water
- capacity constraints
- heat reuse
PUE remains useful, but it should not be the only measure used to judge an AI facility's environmental performance.
How often should PUE be measured?
Continuous or frequent energy monitoring provides much more useful information than occasional manual readings.
Operators can use:
- facility meters
- UPS measurements
- PDUs
- DCIM platforms
- building-management systems
Trend data helps identify seasonal variation and the effect of infrastructure changes.
Common PUE mistakes
- Comparing different measurement boundaries. The numbers may not be equivalent.
- Using one short-term reading. Weather and load can distort the result.
- Assuming low PUE means low carbon. Energy source matters.
- Ignoring IT utilisation. Efficient facilities can still contain idle servers.
- Chasing PUE without considering resilience. Operational requirements still matter.
- Treating PUE as a complete sustainability score. It measures only one aspect of performance.
A practical PUE example
Suppose a facility uses 12,000 MWh over a year and its IT equipment uses 8,000 MWh.
PUE = 12,000 ÷ 8,000 = 1.50
The difference of 4,000 MWh represents facility energy outside the measured IT load.
If improvements reduce total annual facility use to 10,800 MWh while IT energy remains 8,000 MWh:
PUE = 10,800 ÷ 8,000 = 1.35
The supporting overhead has been reduced substantially.
Frequently asked questions
Is a PUE of 1.2 good?
A PUE of 1.2 represents very efficient facility overhead by typical industry standards, but the measurement boundary, climate, load and facility type should still be considered.
Does PUE include servers?
Yes. IT equipment energy is the denominator and is also part of the total data centre energy in the numerator.
Does PUE measure carbon emissions?
No. It measures energy efficiency at the facility level. Carbon impact depends on factors such as the energy source and is considered through other measures.
Why is PUE higher in older data centres?
Older facilities may use less efficient cooling and power systems and may have layouts that make airflow management harder. However, age alone does not determine efficiency.
Can improving server utilisation reduce PUE?
It can influence the ratio, particularly where facility overhead remains relatively fixed, but server utilisation is an IT-efficiency issue that PUE does not directly measure.
Who created PUE?
The Green Grid introduced PUE as an industry metric, and it has since become formalised through international data centre KPI standards.
Learn data centre sustainability and efficiency
The Certified Data Centre Environmental Sustainability Specialist (CDESS) covers PUE, CUE, WUE, energy management, carbon emissions, cooling efficiency, sustainable energy and environmental reporting.
For broader facility knowledge, the Data Centre Foundation Certificate (DCFC) introduces the power, cooling and infrastructure systems that influence PUE.
Explore our full Data Centre training portfolio.
Further reading
Found this article useful? Add ExperTrain as a Preferred Source on Google to help surface more of our training guides, articles and learning resources.
.png){kind=small}
