In November 2025 Cursor’s maker, Anysphere, raised $2.3 billion at a $29.3 billion valuation. Many people asked whether a code editor could be worth that much. Nine months later SpaceX paid $60 billion for it in stock. The deal closed on August 14, 2026, making it the largest purchase of a venture-backed startup on record. So the moat was real enough for a buyer to pay double. Whether it holds inside SpaceX is a different question.
How Fast Cursor Grew
Cursor is a fork of VS Code with AI built into the editor itself. It reads the whole project, edits many files at once and now runs background agents that act without a developer in the loop. Its revenue curve is the steepest ever seen in business software. Annual recurring revenue crossed $100 million in January 2025, $500 million by June 2025, $1 billion by November 2025 and $2 billion by early 2026. In April 2026 the company told investors it expected to end the year above a $6 billion run rate.
| Date | Event | Valuation |
|---|---|---|
| Jul 2024 | Series A, $60M | $400M |
| Dec 2024 | Series B, $105M | $2.5B |
| Jun 2025 | Series C, $900M | $9.9B |
| Nov 2025 | Series D, $2.3B, ARR passes $1B | $29.3B |
| Apr 2026 | $2B round at about $50B in talks; xAI signs option to buy for $60B or pay $10B for a partnership | about $50B |
| Jun 16, 2026 | SpaceX exercises the option to buy | $60B |
| Aug 14, 2026 | Deal closes; about 389.3M SpaceX Class A shares issued; Cursor joins SpaceXAI | $60B |

What the Moat Turned Out to Be
Back in 2025 the bear case was simple. Cursor rented its brains from OpenAI, Anthropic and Google, paid them a large share of every dollar it made, and competed with those same labs’ own coding tools. That case was right about the pressure and wrong about the outcome.
Three things held up. First, distribution. Millions of developers used Cursor daily, and enterprise customers became the larger share of revenue. Second, Cursor’s own Composer models. Moving routine work onto them helped cut inference costs. Third, data. Every accepted or rejected edit is a training signal about how real engineers change real code. Many analysts said that, more than the revenue, was what SpaceX was buying.
Why SpaceX Wanted It
Musk’s xAI had strong models in Grok but no developer product with mass adoption. OpenAI had Codex and Anthropic had Claude Code, both growing fast. Buying Cursor gave SpaceXAI a ready user base and a feedback loop to train coding models. Cursor got access to a huge GPU fleet; the company now says it has the largest one in the world. The two had already worked together on Grok 4.5 and 4.6 before the deal closed.
The deal was paid entirely in SpaceX stock, issued days after SpaceX’s June 2026 IPO. Anysphere’s founders and backers, including Thrive, a16z and Nvidia, now hold a bet on Musk’s company instead of cash. If SpaceX shares fall, the real price falls with them.
Risks Now That Cursor Is Inside SpaceX
- Model neutrality. Cursor’s pitch was choice: use Claude, GPT or Gemini in one editor. Under SpaceX ownership, Anthropic and OpenAI may be slower to give it their best models on good terms.
- Enterprise trust. Large companies were Cursor’s growth engine. Some will hesitate to send their code to a Musk company and may move to GitHub Copilot, Claude Code or Codex.
- Price war. Model prices keep falling. That helps margins but also lowers the bar for rivals. The cuts are covered in OpenAI and Anthropic price cuts.
- Talent. Founder-led teams often thin out after a big sale, especially when payment is in the buyer’s stock.
What It Says About the AI Software Market
Cursor proved that an application built on someone else’s model can win, if it grows fast enough to fund its own models and gather its own data. It also showed that the winners in AI apps get bought by whoever owns compute. For the other side of that trade, see the Anthropic IPO, and for where the money for all this compute comes from, read our AI capex tracker. More on developer and software trends is at technologies.org.