SpaceX has reportedly secured another major customer for its rapidly expanding AI infrastructure business, signing a computing agreement with Reflection AI valued at approximately $6.3 billion. According to CNBC, the deal will provide Reflection AI with access to advanced computing resources beginning on July 1, with monthly payments of roughly $150 million continuing through 2029.
The agreement grants Reflection AI immediate access to NVIDIA’s latest GB300 chips, currently regarded as some of the most powerful processors available for training and running large-scale artificial intelligence models. Demand for high-performance AI computing has surged across the industry as companies race to develop increasingly sophisticated foundation models, autonomous agents, and enterprise AI applications.
Under the terms of the contract, either party can terminate the agreement with 90 days’ notice after the first three months of operation. While such flexibility is common in large infrastructure agreements, the scale of the deal highlights the growing importance of AI compute capacity as one of the most valuable resources in the technology sector.
The latest contract further strengthens the customer base of SpaceX’s Colossus AI data center business. The company has rapidly emerged as a major provider of AI infrastructure, leveraging massive investments in computing hardware and data center capacity. Existing customers reportedly include Anthropic and Google, both of which have signed agreements for AI computing resources. Cursor, the AI coding platform developed by Anysphere, is also listed as a customer while simultaneously being acquired by SpaceX in a separate transaction.
The expansion of SpaceX’s AI operations reflects a broader shift in the company’s business model. While SpaceX remains best known for launch services and the Starlink satellite network, investors increasingly view AI infrastructure as one of its most significant long-term growth opportunities. The company has invested heavily in advanced computing systems, positioning itself to compete with hyperscale cloud providers and specialized AI infrastructure operators.
For Reflection AI, access to NVIDIA GB300 hardware could provide a significant competitive advantage. The newest generation of AI accelerators offers substantial performance improvements over previous systems, enabling faster model training and more efficient inference at scale. As AI developers compete for limited access to cutting-edge chips, long-term compute agreements have become a critical strategic asset.
The deal also reinforces the growing concentration of AI infrastructure spending among a small group of technology companies. Analysts expect demand for large-scale compute capacity to continue rising sharply over the coming years as AI adoption expands across industries worldwide.
