At $150 million a month from July 1, 2026, through the end of 2029, a new computing agreement between SpaceX and open-source AI startup Reflection AI would add up to roughly $6.3 billion over its full life, according to CNBC.
Through the arrangement, Reflection gains immediate use of Nvidia GB300 chips and related hardware housed at SpaceX’s Colossus 2 facility in Memphis, Tennessee, according to Bloomberg. A 90-day exit clause, exercisable by either party once the initial three-month period has elapsed, is built into the contract.
Two former Google DeepMind researchers started Reflection, which has Nvidia among its backers and most recently carried a $25 billion valuation. No public frontier open-source model has emerged from Reflection yet, though the startup has cultivated ties with national security and government clients — among them participants in the Department of Energy’s Genesis Mission and various Pentagon AI programs.
The agreement is the latest in a series of compute deals SpaceX has struck as it positions itself as an AI infrastructure provider. Prior commitments from Alphabet’s Google and Anthropic to rent SpaceX computing capacity were valued at roughly $30 billion and $45 billion respectively, each running through approximately mid-2029. SpaceX has also struck agreements with Cursor, which it is now acquiring.
Reflection’s deal arrives amid growing interest in open-source AI models. When the Trump administration threatened to cut off foreign nationals from Anthropic’s services, the company pulled access to its most powerful models — a move that, according to Yahoo Finance, stoked wider debate about the vulnerabilities inherent in relying on closed AI providers.
“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 spokesperson said in a statement.
The arrangement creates an unusual loop: Nvidia put $800 million into Reflection, and Reflection will now run on Nvidia hardware that SpaceX acquired — making the chipmaker simultaneously an investor in and an indirect supplier to the same customer. “More compute means more runway to build the world’s best open models at scale,” a Reflection spokesperson said in a statement.
