Musk targets 440,000 Vera Rubin GPUs in one cluster by Q2 2027, up to 20GW
kimmonismus · x · 2026-10-21
On SpaceX's Q3 earnings call, Elon Musk laid out an aggressive AI compute buildout: a single training cluster with 440,000 NVIDIA Vera Rubin machines online by end of Q2 2027, roughly 13 GW of AI compute online by end of 2027 (about 10 GW added that year), and SpaceX aiming to add 20 GW of power and cooling infrastructure in 2027 — which he called the real constraint. He claimed the cluster could make Grok "potentially the best model."
More from Companies & People
- NVIDIA teases big Europe push in event intro, photos show — testingcatalog · 2026-10-21
- Jensen Huang takes stage at NVIDIA GTC Berlin, keynote livestreams now — nvidia · 2026-10-21
- Stanford's 2026 AI VC ranking: a16z tops with 22,509 points, Menlo Ventures climbs to No. 3 — deedydas · 2026-10-21
- Two solo-founder AI startups: always-on security engineer and sim-based verification — julianweisser · 2026-10-21
- Google releases free 1-hour AI engineering course, from first agent to autonomous systems — msharmas · 2026-10-21
- NVIDIA GTC 2026 kicks off in Berlin with supply and volatility on stage — kimmonismus · 2026-10-21