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

SpaceX has signed a compute agreement with open-source AI lab Reflection AI valued at up to $6.3 billion, providing access to Nvidia's latest chips from its Memphis-area data center starting July 1, 2026, the companies confirmed to TechCrunch. [1] Under the terms, Reflection AI will pay $150 million a month for immediate access to Nvidia's GB300 AI chips and supporting hardware at SpaceX's Colossus 2 facility through 2029. [1] Either party can terminate the contract with 90 days' notice after the first three months. [1] The deal is smaller than SpaceX's existing compute agreements with Anthropic and Google, which pay $1.25 billion and $920 million per month respectively, both also running through July 2029. [1] Reflection AI, founded in 2024 by two former Google DeepMind researchers, is positioning the contract as a validation of its open-weight AI strategy. [1] The startup releases trained model parameters publicly, an approach it frames as an alternative to closed frontier labs such as Anthropic and OpenAI. [1] Interest in open-weight models has grown following the U.S. government's ban of Anthropic's closed models, Fable and Mythos. [1] "Recent events highlight how important open source is to the AI ecosystem, with more nations and enterprises recognizing the risks and costs associated with exclusively depending on closed models," a Reflection AI spokesperson said. [1] "Our deal with SpaceXAI signals Reflection's strategic importance within the frontier AI ecosystem, and more compute means more runway to build the world's best open models at scale." [1] The Colossus data center was originally built by xAI, the artificial intelligence company founded by Elon Musk that is now part of SpaceX, for its own internal AI projects. [1] SpaceX has since leveraged its chip inventory to rent compute capacity to external AI labs as its in-house pursuits have faltered. [1] SpaceX operates three divisions: Space, Connectivity via Starlink, and Artificial Intelligence, which runs Grok, X, and data centers. [6] The company acquired xAI in 2026 and announced a joint venture with Tesla to build Terafab, a large semiconductor fabrication plant. [6] Nvidia, whose GB300 chips power the Colossus 2 deployment, controlled more than 80% of the market for GPUs used in training and deploying AI models as of 2025 and provided chips for over 75% of the world's TOP500 supercomputers. [5] The company originally focused on GPUs for video games before investing over a billion dollars to develop its CUDA software platform, which enabled GPUs to run massively parallel programs for compute-intensive applications. [5] In 2025, driven by global demand for AI data center hardware, Nvidia became the first company to surpass $4 trillion and $5 trillion in market capitalization. [5]