Skip to content

Energy layerData Center Energy Efficiency

PUE

What it is

A score for how much of a data center’s power actually reaches the computers.

7companies exposed2parts it leans on1part needs it
Where PUE sitsLayer 1 of 5
EnergyChipsInfraModelsAppsPUE
Model relevance41/100

Ranks low (41/100) - set apart by cross-layer reach (56) and strategic relevance (40).

The short read

Why PUE matters, and what breaks without it.

Why it matters

It translates facility design into usable compute and cost per workload.

Why now

As power becomes scarce, squeezing PUE turns directly into more deployable compute.

If PUE runs short

High PUE wastes scarce power and inflates cost per token downstream.

Felt next by

  1. Cost per kWhEnergy

Editorial read

In depth · editorial + model

PUE, or power usage effectiveness, is a score for how much of the electricity drawn by a data center actually reaches the computers rather than being lost to cooling, lighting and conversion. It matters because it converts facility design directly into usable compute: the same grid connection yields more real work when less of it is wasted on overhead. As power becomes the scarce input in AI, a better PUE is not just an efficiency badge - it is deployable capacity in disguise, extra compute unlocked without another megawatt from the utility.

If PUE runs high, scarce power is squandered on overhead and the waste flows downstream as inflated cost per token. That puts the exposure on the efficiency-infrastructure and cooling specialists - Envicool, Munters and Modine - whose kit determines how much of a site's power survives the trip to the chips. In a power-constrained world, the firms that shave overhead effectively sell capacity.

Mental models

How to think about it

  1. 01Efficiency is deployable capacity in disguise

Signals to follow

What to watch

  • Fleet PUE trends
  • Cooling-driven efficiency gains

Questions

Frequently asked

The short answers about PUE, from the same model as the rest of the page.

What is PUE?

A score for how much of a data center’s power actually reaches the computers.

Why does PUE matter for AI?

It translates facility design into usable compute and cost per workload.

Who makes PUE?

The companies the model tags as producers or suppliers of PUE: Shenzhen Envicool Technology, Alfa Laval AB, Carrier Global Corporation, Daikin Industries, Ltd..

Which companies are most exposed to PUE?

Munters Group AB, Shenzhen Envicool Technology, Marubeni Corporation, Modine Manufacturing Company, Alfa Laval AB, Carrier Global Corporation. In total, 7 companies are mapped to PUE.

What happens if PUE runs short?

High PUE wastes scarce power and inflates cost per token downstream.

Where does PUE sit in the AI value chain?

PUE sits in the Energy layer of the AI value chain.

In the app

Go deeper on PUE

Inside THE ENTITY, PUE is live: its pressure, its links and the companies on it, refreshed nightly.

  • The materials, geographies and policies it depends on, heat-mapped
  • Substitutes, relief valves and the domino chains if it tightens
  • The live tension score, momentum and news drivers
  • Four levels of analysis, from plain English to strategic
The whole chain

Research, not investment advice. Model scores are illustrative reads from our model of the AI value chain, and companies are mapped to explain the chain, never as recommendations.

as of 2026-11-09Medium confidence model v2.01