The cabinet of servers in which the buildout is counted, powered and cooled. Chapter 4.
From The AI Compute Race. Read the book
Where the book uses it
- Chapter 4 — One Rack
How a datacenter is actually built, from a converted factory in Memphis to a greenfield campus in Texas, and what the clock from first ground to first power really measures.
Sources that use it
- NVIDIA, “Hardware — NVIDIA DGX GB Rack Scale Systems User Guide” (GB200 NVL72: 72 GPUs, 36 CPUs, 18 compute trays, 9 NVLink switch trays; approximately 120 kW), accessed 29 August 2026. In the Sources
- NVIDIA, “NVIDIA DGX SuperPOD: Next Generation Scalable Infrastructure for AI Leadership — Reference Architecture Featuring NVIDIA DGX H100 Systems,” RA-11333-001 V11, 22 September 2023 (of the single-scalable-unit rack layout of Figure 2, which draws four DGX H100 systems in each of eight racks: “In this example, power consumption per rack exceeds 40 kW,” page 7; “Rack design can support one, two, or four DGX H100 systems per rack, so that the rack layout can be modified to accommodate different data center requirements,” page 4; NVIDIA states no per-rack figure but this one example, and neither it nor the user guide uses the words air-cooled of the rack; copy held, md5 c178af4b0161a5e3bcc4bafe2a92477a). In the Sources
Related terms
From the glossary of The AI Compute Race.
How to cite this page
Dryburgh, Lee S. “Rack.” In The AI Compute Race. DRYBURGH. https://dryburgh.com/vol1/glossary/rack (accessed 2026-11-01).
BibTeX
@misc{dryburgh-rack-2026,
author = {Dryburgh, Lee S.},
title = {{Rack}},
howpublished = {In \emph{The AI Compute Race}, DRYBURGH},
year = {2026},
url = {https://dryburgh.com/vol1/glossary/rack},
note = {Accessed 2026-11-01}
}