SpaceX inks compute deal with Reflection AI, an open-source AI lab

SpaceX has secured a $150 million monthly agreement with Reflection AI, an open-source AI laboratory, to provide access to Nvidia's GB300 AI chips through its Colossus 2 data center in Memphis, Tennessee, starting July 2026 through 2029. This deal demonstrates SpaceX's expansion into AI infrastructure services and reflects growing demand for enterprise-grade GPU compute capacity.

SpaceX's partnership with Reflection AI represents a strategic monetization of its Colossus 2 data center infrastructure, positioning the aerospace company as a serious player in the competitive AI compute market. The $150 million monthly commitment—totaling roughly $5.4 billion over the contract period—underscores the exceptional scarcity and premium pricing of Nvidia's latest-generation GB300 chips, which are highly sought after by AI labs and enterprises building large language models and other compute-intensive applications.

This deal emerges amid a broader infrastructure arms race in AI development. Major cloud providers and specialized compute vendors are aggressively acquiring the latest Nvidia GPUs, creating bottlenecks that force AI companies to seek alternative suppliers. SpaceX's entry into this space leverages its existing expertise in large-scale infrastructure deployment and its independent supply chain advantages, including satellite capabilities that could theoretically support distributed computing resilience.

For the AI industry, this signals that dedicated data center operators—particularly those with unique infrastructure advantages—can command premium pricing for cutting-edge compute access. Reflection AI's willingness to commit to such a substantial multi-year contract suggests confidence in needing sustained, guaranteed capacity for its research initiatives. The deal also validates the Colossus 2 facility's technical capabilities and reliability from a third-party perspective.

Investors should monitor whether SpaceX announces additional compute contracts, as this could establish a meaningful new revenue stream for the company. The broader implication is that AI compute infrastructure consolidation continues, with non-traditional players disrupting established cloud providers' dominance.

• →SpaceX monetizes Colossus 2 data center through a $150 million monthly AI compute contract extending through 2029.
• →Reflection AI secures long-term access to scarce Nvidia GB300 chips, addressing acute GPU availability constraints in the AI market.
• →The deal validates SpaceX's infrastructure capabilities and signals the company's pivot toward recurring AI services revenue.
• →Premium pricing for enterprise GPU access reflects sustained undersupply of cutting-edge AI chips relative to demand.
• →Non-cloud-provider data center operators are gaining competitive advantages in serving AI workloads requiring dedicated capacity.