SpaceX Commits Its AI Buildout to NVIDIA Vera Rubin
Elon Musk said during SpaceX's August 4 earnings call that the company had decided to build its AI infrastructure exclusively on NVIDIA and called Vera Rubin the best architecture for the job. Axios separately reported Musk's projection that SpaceX could reach up to 10 gigawatts of computing power by the end of 2027.
SpaceX plans to build its next phase of AI infrastructure around NVIDIA hardware. During the company's August 4, 2026 earnings call, CEO Elon Musk said SpaceX had decided to build exclusively on NVIDIA because it viewed the Vera Rubin architecture as the best available AI computer. Musk also posted a shorter first-party statement on X saying SpaceX had committed to NVIDIA GPUs.
Axios reported the earnings-call statement alongside SpaceX's first public-company quarterly results. The report says SpaceX generated $7.81 billion in revenue for the quarter ended June 30, including $2.56 billion from its AI division, and cut its net loss to $541 million. Musk projected that SpaceX could have up to 10 gigawatts of computing power by the end of 2027. Those figures are company statements and projections, not completed capacity.
A supplier decision with execution risk
The decision gives NVIDIA a prominent role in SpaceX's combined terrestrial and orbital-compute plans. Vera Rubin is NVIDIA's next rack-scale platform for AI training and inference, and SpaceX is framing the architecture as the basis for its buildout rather than merely another accelerator option.
The word "exclusively" should be read in the context of Musk's stated forward plan. It does not erase SpaceX's disclosed interest in developing or manufacturing chips over the longer term, and the retrieved evidence does not establish a supply contract size, purchase price, or delivery schedule. The 10-gigawatt figure is also a target covering total computing power, not a verified NVIDIA order.
For infrastructure teams, the material signal is architectural concentration: standardizing on one GPU platform can simplify software and networking decisions, but it also makes delivery timelines, power availability, cooling, and supplier capacity central dependencies. SpaceX's projection is unusually large, so progress should be evaluated against disclosed deployments rather than the target alone.
Key Points
- 1Musk said on August 4 that SpaceX had decided to build its AI infrastructure exclusively on NVIDIA.
- 2He identified NVIDIA's Vera Rubin platform as the architecture SpaceX considers best for the buildout.
- 3Axios reported a company projection of up to 10 gigawatts of computing power by the end of 2027.
- 4The 10-gigawatt figure is a forward target, not a disclosed GPU order or completed deployment.
Scoring Rationale
A major AI infrastructure operator's stated decision to standardize on NVIDIA Vera Rubin is commercially material, and the projected 10-gigawatt scale is substantial, but the capacity and timing remain forward-looking company claims.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
