Summary: TechCrunch reported that Cursor was negotiating at least a $2 billion raise at a $50 billion pre-money valuation, although terms were not final. The company’s annualized revenue had reached $2 billion in February, and enterprise gross margins had turned positive after adding Composer and cheaper model routing. The article says in-house models and orchestration reduce dependence on suppliers that also compete with Cursor.
TechCrunch has learned from four people with knowledge of the discussions that Cursor, the AI coding startup, is negotiating a raise of at least $2 billion at a $50 billion pre-money valuation. The round is already oversubscribed, although nothing is final yet and the terms could still shift. Converted to a pre-money basis, the price tag sits about seventy percent over the post-money valuation the company reached five months earlier.
Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger started the company in 2022, while all four were still studying at MIT. The product itself — an AI-native code editor and agent workspace whose users run from individual developers up to, in growing numbers, large enterprise teams — goes head-to-head with Anthropic’s Claude Code and OpenAI’s overhauled Codex. Read that roster of rivals and you know exactly where investors believe value is pooling in the AI stack: not only in the model layer, but in the surface that orchestrates model work across real codebases.
A Second Doubling: $29.3B to $50B
November 2025 brought the last round: Cursor took in $2.3 billion at a $29.3 billion post-money valuation, a deal that landed a mere five months after the one before it. A close at $50 billion pre-money would leave the company worth nearly twice what it was six months prior — three large rounds inside roughly a year. Across 2026’s fundraising landscape, that kind of speed appears in just two lanes: AI coding and foundation models.
Oversubscription carries real weight here. It signals that the company, not the market, is deciding how much of the deal to hand out — demand never had to be chased — and it leaves incumbent holders Thrive and a16z with the leverage to defend their stakes even as the checks get bigger.
The Money Behind the Round: Thrive, a16z, Battery, Nvidia
Going by the reporting, Thrive and Andreessen Horowitz, both already investors, are lined up to co-lead; Battery Ventures looks set to come aboard as a fresh name; and Nvidia is expected to put up a check with strategic intent. Approached for the story, Cursor and Battery declined to comment, while Thrive, a16z, and Nvidia gave no reply.
The most telling detail is Nvidia’s possible seat at the table. AI coding ranks among the quickest-growing wells of inference demand — agentic coding runs tokens through models nonstop rather than one prompt at a time — and when a chipmaker buys equity in one of its thirstiest token consumers, it fastens the demand side to its own ecosystem. Watch for the pattern to spread: silicon investors bankrolling the application layers that ensure their compute keeps getting used up.
ARR Hits $2B as Gross Margin Flips Positive
Bloomberg’s early-March report put Cursor’s annualized revenue at $2 billion as of February, a level that had doubled within three months; the company’s own projection has the annualized run rate clearing $6 billion before 2026 ends — no less than a threefold climb across ten months. Hardly any corner of software has ever compounded at that speed, and nothing has done it at this size.
The bigger structural change sits in margin. Cursor formerly sold below cost at the gross level — that well-known squeeze on AI applications, where serving one more user through inference costs more than the user ever pays back. Launching its in-house Composer model in November 2025 and shifting slices of traffic onto cheaper third-party models, Kimi among them, has since nudged the company to slightly gross-margin positive overall. Enterprise deals now carry positive gross margin; individual developer accounts remain loss-making. That split is why the reporting flags an enterprise-sales surge: financially, what counts is growth that arrives as signed procurement contracts, not viral uptake among developers.
The Case for an In-House Model: Avoid Being Swallowed by Your Supplier
TechCrunch frames the move this way: depending less on external model providers is insurance against getting swallowed by those same suppliers, with Anthropic the clearest danger. Claude Code doubles as Cursor’s closest competitor even while Cursor’s product leans hard on Anthropic’s models. Seen that way, Composer and the routing of traffic to budget models go beyond cost engineering — they amount to a survival plan. Keep the orchestration layer in-house and attach a capable model of your own, and Cursor stops passing along someone else’s intelligence and stands as infrastructure itself.
The wave of capital in developer tooling is also lapping at the public markets: Vercel has said outright that it is IPO-ready, with AI agent deployments driving its revenue. Eye-watering private valuations and IPO groundwork argue the same case: what developers will pay for AI is software’s most dependable growth curve at the moment, and the tooling layer is keeping a larger slice of it than the market projected a year back.