Colossus Phase 1 arithmetic: 100,000 Hopper GPUs divided by 64 GPUs per reference rack is about 1,563 rack-equivalents; divided by 122 days is about 12.8 rack-equivalents per day
~1,563 rack-equivalents; ~12.8/dayderived
| Value kind | derived — Derived values are guide calculations; their result is only as fixed as the stated inputs, scope, and method. |
|---|---|
| Scope | Arithmetic equivalent, not an observed rack count or sustained set-and-cable production rate. The 122-day denominator covers the entire build (shell, power, cooling), not rack-setting labor; the rack-arrival-to-training window was 19 days. |
| As of | 2024-10 |
| Source | Supermicro / ServeTheHome, Colossus case study · PDF p.1: 100,000 H100 GPUs and 122-day build; p.2: eight 8-GPU servers, 64 GPUs per rack |
| Derivation | Guide arithmetic: 100,000/64=1,562.5 rack-equivalents and 1,562.5/122=12.8 per day. It does not establish the actual rack mix or daily installation rate. |
| Review | checking…review by 2026-09-23 · fast cadence |
| Recorded changes | last 2026-08-24 · 4 revisions tracked |
| Claim id | gpu-racks-deployed-at-colossus-phase-1-implies |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.