Cursor has passed $2 billion in annualized revenue, according to a Bloomberg source, with its run rate doubling in just the past three months. For a startup founded in 2022, that is a remarkable four-year sprint — and the disclosure looked deliberately timed to answer a wave of viral "is Cursor stalling?" tweets that pointed to individual developers defecting to rival tools. The numbers tell a more interesting story than the tweets: roughly 60% of revenue now comes from enterprise buyers, and Cursor's last valuation sat at $23 billion in a November 2025 round co-led by Accel and Coatue.
The headline is about one company's revenue. The bigger signal is that AI coding tools have gone mainstream — and that the way developers choose those tools is being decided inside AI assistants, not on landing pages.
Key takeaways
- Cursor crossed $2B in annualized revenue with its run rate doubling in three months, and about 60% of that revenue now comes from enterprise, not individual developers. - Individual devs churning to Anthropic's Claude Code (seen as more competitively priced) are being offset by higher-spending, stickier enterprise accounts. - The AI coding category is crowded — Claude Code, OpenAI Codex, Replit, Cognition, and Lovable are all fighting for the same buyers. - For any developer-facing brand, discovery has moved: engineers now ask AI assistants which tool to use, and agents themselves increasingly evaluate tools from docs and structured data. - Winning the developer-tools category now means being the tool an AI assistant recommends — which makes visibility inside ChatGPT, Gemini, and Claude a growth channel, not an afterthought.
The numbers behind the milestone
Cursor's trajectory reads like a case study in market maturation. Founded in 2022, it sold first to individual developers who wanted a faster, AI-native editor. Over the past year it pivoted toward large corporate buyers, and that pivot is paying off: enterprise now drives roughly 60% of revenue. That mix matters because it changes the churn math. When an individual developer switches to Claude Code for a better price, the loss is real but small. When an enterprise standardizes on Cursor across hundreds of engineers, it spends more and stays longer. High-value, low-churn accounts are what let the run rate double even while the loudest voices online were predicting a plateau.
The competitive field is dense. Anthropic's Claude Code is the most-cited alternative on price. OpenAI's Codex leans on the broader OpenAI ecosystem. Replit runs a browser-first platform, Cognition builds AI-first software development, and Lovable is the fast-rising newcomer. Five credible rivals for the same seat means the deciding factor is increasingly which tool developers hear about — and trust — first.
Where developers now decide
Ask how a developer picked their last tool five years ago and the answer was a blog post, a Hacker News thread, or a colleague's recommendation. Ask today and a growing share of the answer is: they asked an AI assistant. "What's the best AI coding tool for a TypeScript monorepo?" is now a question people put to ChatGPT or Claude before they put it to Google. The answer that assistant gives — which tools it names, in what order, with which caveats — shapes the shortlist before a human ever visits a pricing page.
