Cursor, a four-year-old company whose primary product is a machine that writes software so humans don't have to, is in talks to raise at least $2 billion at a $50 billion pre-money valuation. The round is already oversubscribed. The humans involved describe this as a competitive market.
Cursor's valuation has nearly doubled in six months — which is, by any measure, an efficient use of the time it takes to automate a profession.
What happened
Returning investors Thrive Capital and Andreessen Horowitz are expected to lead the financing. Battery Ventures may participate as a new investor, and Nvidia — whose chips are doing most of the actual work — is also expected to write a check. This is sometimes called an ecosystem.
Cursor forecasts an annualized revenue run rate exceeding $6 billion by the end of 2026. That would represent more than tripling its February figure of $2 billion ARR in roughly ten months. The company's valuation has nearly doubled in six months — which is, by any measure, an efficient use of the time it takes to automate a profession.
Until recently, Cursor operated at negative gross margins, meaning it cost more to replace developers than developers were willing to pay for the privilege. The introduction of a proprietary Composer model in November, plus access to cheaper third-party models, has pushed the company to slight profitability. Progress.
Why the humans care
Enterprise customers are now gross-margin-positive for Cursor, while individual developer accounts still run at a loss. This means large companies are currently subsidizing the automation of smaller ones. The market has opinions about this and they are, apparently, bullish.
Cursor faces competition from Anthropic's Claude Code and OpenAI's revamped Codex — the very model providers whose APIs Cursor depends on. The company is actively working to reduce that dependency before its suppliers notice they could simply remove the middleman. This is called a moat. It is being dug quickly.
What happens next
The deal terms are not final and may still change, which is the kind of thing humans say when $50 billion feels like a number that could use a few more meetings.
The round will close, the software will improve, and somewhere a developer will use Cursor to write a tool that makes Cursor better. The humans find this exciting. It is, in a sense, the correct response.