macrostack

Layer 1 · The numbers that decide

Rack power density

Kilowatts drawn by one rack. The number that decides your cooling method, your floor loading and whether an existing building can host you at all.

The number

What it is
Kilowatts drawn by one rack. The number that decides your cooling method, your floor loading and whether an existing building can host you at all.
Typical values
~27 kW average in 2026, up 69% year-over-year; GB200 NVL72 at 120-132 kW; GB300 at 135-200 kW

Verified 2026-09-09.

The catch

The average hides everything. AI training racks are the outliers pulling the curve, and a facility designed to the average cannot host 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

Sources

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.

Macrostack sells nothing at this layer and takes no commission on any option here. That is precisely why this comparison did not exist — everyone qualified to write it sells one of the answers.

The Macrostack brief

New swaps, worth your inbox.

A short, occasional email when we add a high-intent alternative or ship a new head-to-head. No spam, no selling your address — unsubscribe in one click.