SpaceX commits exclusively to Nvidia GPUs as Musk plans orbital AI data centers with Vera Rubin

SpaceX Nvidia Exclusivity Deal Signals Major AI Infrastructure Shift

Elon Musk confirmed during SpaceX's earnings call that both SpaceX and xAI will exclusively use Nvidia GPUs moving forward, praising the Vera Rubin NVL72 architecture as superior to all available alternatives1

1

. "We think the Vera Rubin architecture is the best architecture," Musk stated, emphasizing the cable-less design of compute trays in the NVL72 VR200 rack system1

1

. This commitment solidifies Nvidia's dominance in AI infrastructure, particularly as the company already commands approximately 85 percent of the datacenter GPU market2

2

. While xAI has historically relied on Nvidia's Hopper and Blackwell hardware to train multiple generations of Grok, this announcement formalizes the partnership and extends it beyond terrestrial data centers into orbital space computing1

1

.

Starmind AI Satellite Program Targets 2027 Launch

SpaceX announced a partnership with Nvidia to design the Starmind AI1 satellite orbital compute payload, with each satellite incorporating Nvidia Rubin GPUs and Vera CPUs for datacenter-class space computing2

2

. "We expect to start launching this next year," Musk revealed, describing the project as an optimized Vera Rubin NVL72 computer designed for both ground and orbital deployment1

1

. The fully deployed Starmind AI1 will measure 30 meters tall with a 75-meter wingspan and support a 250 kW compute payload2

2

. SpaceX envisions scaling this to a constellation of up to one million satellites, creating a distributed AI supercomputer in space that relies on laser links for high-speed data exchange3

3

. SpaceX has requested Federal Communications Commission authorization for deploying satellites between 311 and 1,243 miles above Earth, describing a petabit-scale optical communications network3

3

. Broader deployments could begin as early as 2028, with demonstrations of orbital AI computing targeted for late 20274

4

.

Orbital AI Data Centers Address Power and Land Constraints

Sharmila Bednarek, speaking at the AI for Good Global Summit 2026, explained that orbital AI data centers solve critical resource limitations facing terrestrial data centers. "The sun has vast energy. It requires no water or land," she stated, adding that "the universe is always clear"4

4

. The International Energy Agency projects global data center electricity consumption will roughly double to approximately 950 terawatt-hours by 2030, with AI-focused facilities tripling their power use4

4

. Satellites operating in sun-synchronous orbits receive consistent solar energy without depending on Earth's electrical grid3

3

. With unlimited power and cooling in space, AI infrastructure can scale beyond terrestrial limitations5

5

. Musk has called solar-powered orbital data centers a "no-brainer," predicting space will become "the lowest-cost place to put AI" within two or three years4

4

.

Massive Capital Expenditures Support AI Ambitions

SpaceX's capital expenditures grew sixfold to $18.4 billion year-over-year as the company accelerates buildout of xAI infrastructure2

2

. The company initially plans to install 2GW of compute capacity by year's end, scaling to 10GW by the end of 20275

5

. SpaceX posted a quarterly net loss of $541 million on revenues of $7.81 billion in its first earnings report as a public company2

2

. The Space-1 Vera Rubin module offers up to 25 times the AI processing performance of an H100 GPU, with commercial shipments expected later this year3

3

5

5

. Each module includes four Rubin GPUs and two Vera CPUs designed for real-time AI processing supporting geospatial intelligence, autonomous operations, and on-orbit analytics5

5

.

Starship Critical to Scaling Space Computing Vision

Starship serves as the foundation for achieving orbital AI data centers at scale. Bednarek emphasized that large orbital platforms ultimately require "megaton-class materials," adding that "Starship can make that a reality"4

4

. SpaceX's fully reusable rocket is designed to carry more than 100 metric tons to orbit4

4

. Current launch costs sit around $7,000 per kilogram aboard a Falcon 9 rocket, placing the price of launching a 3.33-ton AI1 satellite north of $23 million2

2

. For orbital data centers to become cost-effective, Starship needs to reduce costs to Musk's stated goal of $10 per kilogram, which would bring launch costs down to approximately $33,300 per satellite2

2

. SpaceX envisions Starship eventually delivering millions of tons to orbit annually, with a factory capable of producing thousands of AI satellites beginning as soon as late 20274

4

.