Power Usage Effectiveness (PUE)
Data Center & FacilitiesA standard efficiency metric for data centers, the ratio of a facility's total energy use to the energy actually delivered to its computing equipment.
Eleveight AI's facility achieves a PUE of 1.1, well below typical industry levels, through high-altitude air cooling, roughly eleven months of free cooling a year, and power delivery engineered to keep overhead low under dense, sustained GPU load.
Overview
Power Usage Effectiveness, or PUE, captures how much of a data center's electricity actually does useful computing work versus how much is spent on overhead such as cooling, power conversion, and lighting. It is expressed as a ratio: a PUE of 2.0 means a facility draws two watts for every watt that reaches the servers, while a PUE approaching 1.0 means almost all energy goes to compute. Lower is better.
How it works
PUE is calculated by dividing the total energy entering a facility by the energy delivered to the IT equipment alone. The gap between the two is overhead, dominated in most facilities by cooling. Efficient designs shrink that gap through better cooling approaches, such as liquid cooling, smarter airflow, and minimizing losses in power distribution, so a larger share of every kilowatt reaches the processors that do the work.
Why it matters
For GPU facilities, where power is both the largest operating cost and the binding constraint on how much hardware a site can support, efficiency has real consequences. A lower PUE means more of the available power budget goes to compute rather than to cooling it, improving both economics and the amount of useful capacity a fixed electrical supply can sustain. It is a key measure of how well a facility is engineered.
Use cases
- Benchmarking data center efficiency
- Planning power capacity for GPU density
- Evaluating facility sustainability
- Comparing colocation and hosting options