SpaceX has expanded its footprint in the artificial intelligence infrastructure market by securing a multi-billion-dollar compute agreement with Reflection AI, an open-source AI startup. Under the terms of the deal, Reflection AI will pay $150 million monthly starting July 1, 2026, through 2029. The contract, valued at up to $6.3 billion, grants the startup immediate access to Nvidia’s cutting-edge GB300 AI chips and supporting hardware housed at SpaceX’s Colossus 2 data center in Memphis, Tennessee. Both parties retain the option to terminate the agreement with a 90-day notice after an initial three-month period.

This partnership represents a significant milestone for Reflection AI, which was founded in 2024 by two former Google DeepMind researchers. The startup is positioning itself as a leading open-weight alternative to closed-source AI giants like OpenAI and Anthropic. The massive infrastructure commitment comes at a time of heightened interest in open-source AI models, particularly following recent U.S. government restrictions on closed-source models such as Anthropic’s Fable and Mythos. Reflection AI intends to leverage this newly acquired computational power to accelerate the development of high-performance open models at scale.

While substantial, the $6.3 billion agreement is smaller than SpaceX’s existing compute partnerships. The aerospace giant currently holds massive contracts with Anthropic and Google, valued at $1.25 billion and $920 million per month, respectively. These deals are also slated to run through July 2029, though SpaceX leadership has noted the flexibility of these timelines due to cancellation clauses.

The Colossus data center hosting these operations was originally constructed by xAI, the artificial intelligence venture founded by Elon Musk that has since been integrated into SpaceX. As SpaceX’s internal AI initiatives shifted focus, the company pivoted to monetize its extensive inventory of high-demand Nvidia chips by leasing them to prominent external AI laboratories. This strategy has effectively positioned SpaceX as a major broker of raw computational power in the global AI race.

Key Takeaways

• Reflection AI has committed up to $6.3 billion in a compute deal with SpaceX to access Nvidia GB300 chips at the Colossus 2 data center.
• The agreement highlights a growing industry shift toward open-weight AI models, especially amid increasing regulatory scrutiny on closed-source alternatives.
• SpaceX continues to establish itself as a major AI infrastructure provider, leveraging hardware originally acquired for its xAI division to serve major tech clients.

Editor’s Analysis & Impact

This deal underscores a critical pivot in the AI landscape, highlighting both the monetization of hardware infrastructure and the rising prominence of open-source AI. By leasing out its massive Colossus 2 data center, SpaceX is successfully hedging its bets; rather than relying solely on internal AI breakthroughs, it is generating steady, multi-billion-dollar revenue streams from external developers. For Reflection AI, securing Nvidia’s state-of-the-art GB300 chips provides the computational runway necessary to challenge closed-source giants like OpenAI. Furthermore, as governments increasingly scrutinize proprietary, closed-loop AI systems, open-weight models are gaining strategic traction. This massive infrastructure commitment suggests that the future of frontier AI development may not belong exclusively to a handful of tech monopolies, but will be increasingly distributed and open.

Frequently Asked Questions

Q: What is the total value and duration of the deal between SpaceX and Reflection AI?
A: The deal is valued at up to $6.3 billion, with Reflection AI paying $150 million per month from July 1, 2026, through July 2029.

Q: What hardware and facilities are involved in this agreement?
A: Reflection AI will gain access to Nvidia's latest GB300 AI chips and supporting hardware located at SpaceX's Colossus 2 data center in Memphis, Tennessee.

Q: Why is Reflection AI focusing on open-weight AI models?
A: Reflection AI advocates for open-weight models as safer, more transparent, and cost-effective alternatives to closed-source models, especially as regulatory bodies begin restricting certain proprietary AI systems.