SpaceX just bet $60 billion on a 25-year-old's coding startup because Elon Musk's own AI can't keep up with the competition.

The Summary

The Signal

This isn't a standard acquisition. It's a $60 billion option play with a $10 billion floor, structured like Musk is hedging against his own company's weakness. SpaceX gets the right to buy Cursor later this year for $60 billion, but if the deal doesn't close, they pay $10 billion for whatever Cursor builds using SpaceX's compute. That's not how confident buyers structure deals. That's how you buy insurance when you know you're behind.

The reason is brutal and specific. Bloomberg reports that xAI has spent months trying to convince businesses to use Grok for coding, but Musk's own employees sometimes refuse to use it. When your internal team won't touch your product, you don't have a go-to-market problem. You have a product problem.

"Neither Cursor nor xAI has proprietary models that can match the leading offerings from Anthropic and OpenAI."

Meanwhile, the competition is moving fast. Google directed its "strike team" to help agentic AI tools catch up, and Sam Altman reportedly declared a "code red" at OpenAI last year, shutting down Sora to focus on ChatGPT's superapp and its own Codex. The AI coding race isn't theoretical anymore. It's the new search war, except instead of crawling web pages, you're automating the people who build the web pages.

Cursor brings distribution that xAI doesn't have. The company has "leading product and distribution to expert software engineers," according to SpaceX's statement. Translation: developers actually use it. SpaceX brings compute. Colossus is a million H100-equivalent training supercomputer powered by 200,000 Nvidia GPUs. Cursor gets to train bigger models without raising another round. SpaceX gets to put that compute to work on something developers might actually adopt.

The deal's structure reveals what both sides need:

  • Cursor needs compute to compete with Anthropic and OpenAI's coding tools
  • SpaceX/xAI needs a product that developers will actually use
  • Both need this to work before the IPO roadshow starts

The IPO timing matters. TechCrunch notes this is happening right before SpaceX's planned public offering, which makes the optics stranger. You typically clean up your story before going public, not add $60 billion question marks to it. Unless the question mark is actually the answer. If SpaceX can show investors it has a credible path in AI coding, a market projected to replace a significant chunk of software engineering work, that's a better IPO story than "we own an AI company that our own people won't use."

The human angle here is Michael Truell, who went from Google intern at 18 to running a company with a $60 billion price tag at 25. That's not just fast. That's the kind of velocity that only happens when you're building the tool that eliminates the job you used to have. Cursor isn't just helping developers code faster. It's training on their work to make them optional.

The Implication

Watch how many developers are still using Cursor in six months. If this deal works, you'll see it in retention numbers and enterprise contracts, not press releases. If it doesn't, SpaceX pays $10 billion for models that might be obsolete by the time the IPO closes.

For anyone writing code for a living, this is the starting gun. The biggest AI coding deals aren't being done by software companies anymore. They're being done by companies that need software, have compute, and can't wait for the market to solve their problems. When SpaceX builds its own coding stack instead of buying from OpenAI or Anthropic, it's saying the same thing Google and Meta already figured out: you can't outsource the tools that build your company. The agent economy doesn't have subcontractors.

Sources

Business Insider Tech | Fortune Tech | Bloomberg Tech | TechCrunch AI | The Verge AI