SpaceX announced a deal on April 22, 2026, that gives it the option to acquire Cursor, the AI coding platform developed by Anysphere Inc., for $60 billion by the end of the year. The structure is not a standard acquisition. It’s an option with a floor: if SpaceX decides not to buy, it owes Cursor a $10 billion collaboration fee. The deal was accompanied by an agreement giving Cursor access to GPU infrastructure operated by SpaceX’s xAI unit, making compute the central resource in the transaction.
Two numbers define the deal. Sixty billion dollars is the option price, the amount SpaceX would pay to own Cursor outright. Ten billion dollars is what SpaceX owes if it opts out. That floor matters. It means Cursor receives a substantial financial commitment regardless of whether SpaceX exercises the option.
The compute dimension is equally significant. Cursor will gain access to GPU infrastructure from xAI, SpaceX’s AI unit. According to sources cited by Business Insider and corroborated by multiple outlets, Cursor plans to use those GPUs to train Composer 2.5, its next-generation coding model. This claim comes from anonymous sources and has not been officially confirmed by Cursor or SpaceX.
The deal’s origin matters too. According to Bloomberg, as reported by SiliconANGLE, Cursor had reportedly been in talks to raise a $2 billion funding round at a valuation exceeding $50 billion. That round was reportedly scrapped once the SpaceX partnership was structured. The reported sequence, a conventional VC raise abandoned in favor of a compute-backed option deal, is the story’s most telling detail.
Why this matters for AI investors and market strategists
The deal’s structure deserves attention independent of its size. Acquisition options are not rare in corporate dealmaking, but they’re uncommon as the primary instrument in venture-stage AI transactions. What’s different here is the mechanism funding Cursor’s operations in the interim: not a cash injection from investors, but GPU access from a strategic acquirer’s infrastructure arm.
That’s a trade-off with real consequences. Cursor gets compute it needs to train competitive models. SpaceX gets an option on one of the most widely adopted AI coding platforms before traditional investors can close at a comparable price. xAI’s GPU cluster becomes the bargaining chip rather than a balance sheet entry.
The scrapped $2 billion round tells the deeper story. Cursor’s reported decision to abandon a conventional raise at a $50-plus billion valuation suggests the SpaceX deal offered something the VC market couldn’t: certainty of compute access on a timeline that matters for model development, not just capital on a cap table.
What to watch
SpaceX has until year-end 2026 to exercise the acquisition option. The $10 billion floor means the clock isn’t costless. Watch for: any Cursor model releases using xAI GPU infrastructure (which would signal deep integration), competitive responses from GitHub Copilot or Replit as Cursor’s ownership trajectory clarifies, and any filings or disclosures that provide primary source confirmation of the deal terms beyond current T3 reporting.
TJS synthesis
The SpaceX/Cursor deal is most useful as a signal, not a data point. When a well-capitalized infrastructure player can offer GPU access in lieu of capital, and that offer beats a $2 billion VC round, it reveals something specific about where AI application-layer companies sit in the current market: their limiting constraint is compute, not cash. Any investor or strategist evaluating AI developer tool companies should now ask not just “what’s the valuation” but “who controls their compute, and on what terms.”