SpaceX has transformed itself from an AI company betting on xAI's success into a data-center landlord collecting rent from multiple frontier AI labs. By leasing compute power at its Colossus facility to competitors like Reflection AI, Google, and Anthropic, SpaceX stands to generate tens of billions in revenue regardless of which company's model wins—a strategy similar to McDonald's shift from selling burgers to owning real estate.
Summaries like this, in your inbox every morning.
Sign up free →What happened
SpaceX signed a $150 million（約240億円） per month compute lease with Reflection AI for use of Nvidia GB300 chips at Colossus 2 in Memphis, TN, potentially generating around $6.3 billion（約1兆円） if the lease runs its full term. Earlier, Google agreed to pay SpaceX approximately $920 million（約1500億円） per month for 32 months (around $30 billion（約4.8兆円） total), and xAI disclosed a deal with Anthropic that could bring SpaceX as much as $45 billion（約7.2兆円） in revenue.
Why it matters
SpaceX has shifted from betting on xAI's success to profiting as a landlord regardless of which AI company wins. Like McDonald's real-estate model, SpaceX built Colossus infrastructure for xAI's Grok but now rents it to competitors—meaning the parent company makes money whether xAI, Google, Anthropic, or Reflection AI succeeds or fails. This decouples SpaceX's financial fate from any single AI model's performance.
What to watch
Colossus now comprises more than 220,000 Nvidia GPUs including H100, H200, and next-generation Blackwell-class accelerators. Nvidia also benefits at every layer—supplying the hardware to SpaceX and investing in the companies that rent from SpaceX (Anthropic, Reflection AI, xAI)—making it a dominant player across the AI economy regardless of which lab produces the dominant model.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
