Musk says SpaceX may run Nvidia VR72 at 250kW, report says
SpaceX CEO Elon Musk stated on X that the company is "cautiously optimistic" about running its version of Nvidia's VR72 AI system at an average power level of 250 kilowatts. He described this potential milestone as a "big deal" that would also apply to ground-based systems, crediting a "close working partnership and co-design" with Nvidia's engineering team. If achieved, this sustained power…
Key points
- Musk says SpaceX may run Nvidia VR72 at 250kW average power, potentially 10-25% above industry averages.
- Google will pay SpaceX $920 million per month for 110,000 Nvidia GPUs from October 2026 to June 2029.
- Anthropic will pay $1.25 billion per month for Colossus 1 data center output through May 2029.
Industry insider Aaron Burnett, founder of Mach33, noted that 250kW of average power could result in battery-supported peaks closer to 275kW. Burnett estimated this target is 10% to 25% above typical third-party reported averages for VR72 systems. He highlighted the technical challenge of heat rejection, stating that SpaceX's StarMind team would need a plan to continuously manage the heat produced by such a sustained load.
These comments highlight SpaceX's expanding AI infrastructure role. The company has disclosed major compute agreements, including a deal where Google will pay $920 million per month from October 2026 through June 2029 for access to 110,000 Nvidia GPUs and related equipment. Additionally, Anthropic has an agreement to pay $1.25 billion per month through May 2029 for output from SpaceX's Colossus 1 data center near Memphis.
The story so far
2 episodes →- Musk says SpaceX may run Nvidia VR72 at 250kW, report saysthis story
SPCX Stock Climbs Overnight: Musk Teases ‘Big Deal’ Nvidia AI Milestone — Industry Insider Flags Up To 25% Higher Power
finance.yahoo.com · 1 October 2026Loading the full article…
This text was published by finance.yahoo.com and written by Deepti Sri. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- Sea deploys Nvidia Vera Rubin chips to scale AI agents · 1 src
- AMD’s AI chip deals push revenue up 107% but leave valuation at 229x · 4 src
- Microsoft says Meta will not buy new chips to scale Muse agent · 1 src
- SpaceX launches Google AI chips into orbit on Falcon 9 rocket · 8 src
- Jacobs launches digital twin platform for AI data centers on NVIDIA Omniverse · 1 src
Comments
via GitHub Discussions