Cursor’s rise is turning the AI coding market into one of the industry’s most expensive battlegrounds.
Cursor, the AI-powered coding editor that helps programmers write, debug, and ship code faster, just reported $3 billion in annualized revenue (the run-rate it’d hit if its current pace held for a full year). In February, that number was $2 billion. It added $1 billion in two months. It’s projecting $6B+ by the end of 2026.
For context: it took Salesforce over a decade to reach $3 billion. Cursor did it in roughly two years.
Here’s what happened
- More than 3,000 customers now pay at least $100,000 each per year.
- Cursor shipped Composer 2.5 this week, its latest model, partially trained on a SpaceX data center.
- SpaceX’s IPO is expected June 12; a Cursor acquisition could close roughly 30 days after.
The SpaceX deal didn’t come out of nowhere. This spring, xAI began renting computing power from its Colossus supercomputer to Cursor for model training. Two of Cursor’s most senior engineering leaders left to join xAI, where they report directly to Musk.
Then in April, SpaceX announced it had secured the right to acquire Cursor for $60 billion outright, or pay a $10 billion fee and walk away. Cursor’s valuation trajectory tells the whole story: $2.5 billion eighteen months ago, $9 billion last May, $29.3 billion by November, and now $60B as a final price tag. I’m sure the team will gladly take $60 billion or $10 billion in cash!
Why this matters
Greg Brockman said it best: the model alone is no longer the product.
The massive success of ChatGPT when it first launched kinda made that obvious. And now, as Chat-based apps have somewhat saturated, the apps around the models are more important than ever.
Cursor is proof there’s still real money in the layer above the models, i.e. the tools that make people’s jobs faster.
But it’s also proof that there’s still value in training your own models. Basically, you need both (app and model) to work. Developers don’t really care which model is under the hood, so long as it’s really good and really cheap. The Pareto curve of frontier capable quality and affordable to run is key.
Our take
Caveat to the above: Cursor still sells access to Claude and GPT models even as Anthropic and OpenAI are now competing directly with it for developers. The SpaceX deal is partly an escape hatch: proprietary compute, a consistent path off its own competitors’ APIs, and a $60B payday.
In a way, this is a kind of team-up of necessity: xAI needs a popular coding platform, as the dev market is now the crown jewel for the AI industry to capture. And Cursor needs a hyper-scale benefactor to keep up with the intense competition on product and model development. If you can’t beat ‘em, join up Avengers Assemble style and try again!
Editor’s note: This content originally ran in the newsletter of our sister publication, The Neuron. To read more from The Neuron, sign up for its newsletter here.