SpaceX says it holds an option to buy Cursor, the AI code editor, for $60 billion. Reuters reported the news, and the tech community's reaction has been somewhere between confused and alarmed. The acquisition is enormous for a developer tool that faces real competition from OpenAI's Codex and Anthropic's Claude Code. Cursor is good, but $60 billion good? That's a stretch by most measures.
This deal is really about the data. Cursor has collected a massive trove of how developers write and edit code, and that information is catnip for training reinforcement learning models. xAI, Elon Musk's AI company, reportedly has around 2GW of GPU compute but not enough specialized coding data to feed it. Cursor also brings enterprise contracts, something xAI has struggled to build on its own. And there's been talk of instability at xAI, so acquiring a talented team wouldn't hurt. SpaceX has also struggled financially with its xAI integration.
Cursor's Composer model complicates things. It's built on Kimi K2, a large language model from Moonshot AI, a Beijing-based company. SpaceX is a US defense contractor. Mixing Chinese AI technology with American defense work creates a serious national security headache. Any acquisition would require unwinding Cursor's dependence on Moonshot AI's technology and likely trigger a CFIUS review to ensure Chinese entities can't access sensitive data flowing through Cursor's systems.
Commenters on Hacker News have also speculated that the deal structure tells its own story. A reported $10 billion breakup fee suggests SpaceX might be fine walking away after getting access to Cursor's data, without completing the full purchase. If that's the strategy, $10 billion for training data is still expensive, but it's a lot less than $60 billion for a code editor that competes in an increasingly crowded market.