04BBuildout · frontier site· cluster ledger reviewed 2026-10-14
Colossus 1 AI datacenter
SpaceXAI (SpaceX, formerly xAI) · Memphis, Tennessee. The ledger records it as operating, with 300 MW energized. Every figure below traces to a dated public source.
Ledger figures
- Operator
- SpaceXAI (SpaceX, formerly xAI)
- Location
- Memphis, Tennessee
- Status
- Operating
- Current power
- 300 MWest — estimated / secondary-source figure
- Planned power
- No public figure
- Accelerators
- ~230K GPUs (150K H100, 50K H200, 30K GB200)
- Announced
- No public date
- First power
- 2024-09
First ~100K H100 built in 122 days in 2024; a grid link upgraded toward 150 MW plus on-site gas turbines and Megapacks. Since May 2026 all of its compute capacity is contracted to Anthropic.
Power figures are IT load where trackers publish it and campus power where only the operator’s headline exists; each source below names its basis. A figure marked est rests on at least one source other than the operator’s own announcement.
Sources behind each figure
- Current power · est All of Colossus 1’s compute capacity contracted to Anthropic: more than 300 MW (over 220,000 NVIDIA GPUs) — Anthropic,
- Current power · est ≈340 MW IT (≈425 MW facility), unchanged since Nov 2025 (tracker estimate) — Epoch AI data-center directory,
- Accelerators · est 230K GPUs: 150K H100, 50K H200, 30K GB200 (Musk on X, Sep 25, 2026) — Tom’s Hardware (quoting Elon Musk on X),
- First power · est 100K H100s online Sep 2, 2024 (tracker timeline) — Epoch AI data-center directory,
- First power · est Initial ~100K-H100 cluster built in 122 days and live in mid-to-late 2024; the grid link (~8 MW at first, upgraded toward 150 MW) was supplemented by temporary gas turbines and Tesla Megapacks — Wikipedia — Colossus (data center),
Background reading: xAI Colossus on AI Wiki. AI Wiki is run by the same team as AI Compute Tracker.