Layer 1 · The numbers that decide
PUE — power usage effectiveness
Total facility power divided by IT power. 1.0 is perfect; every 0.1 above it is overhead you pay for and never compute with.
The number
- What it is
- Total facility power divided by IT power. 1.0 is perfect; every 0.1 above it is overhead you pay for and never compute with.
- Typical values
- 1.4-1.6 air · 1.1-1.2 direct-to-chip · 1.02-1.10 immersion
Verified 2026-09-09.
The catch
PUE says nothing about how much useful work the IT load did. A facility running idle GPUs at PUE 1.1 is efficient and wasteful at the same time. It is a facility metric being widely misused as a performance one.
What this is powering
A GB200 NVL72 rack draws 120–132 kW and a GB300 pushes 135–200 kW, against a 2026 average rack of about 27 kW. The silicon decision and the power decision are the same decision, made eighteen months apart.
Layer 2 — Silicon · GB200 NVL72 · The 150 W inference option
Others in the numbers that decide
Want this whole layer as one table?
The power and cooling options on these pages as a single CSV — lead time, capex, carbon, maturity, density supported, PUE, the catch and what each one decides — ready to sort, filter or load into a notebook.
Get the energy table — free for research and evaluation
Four short lines and the download starts at once (CC-BY-NC-4.0, built from these pages the moment you ask). What you type is used only to answer the request and is never sold. Prefer Hugging Face? Request it there.
Using it inside a product, a client report or anything you sell? That needs a commercial licence.