TL;DR
Cursor’s parent company Anysphere is in talks to raise more than $2 billion at a $50 billion valuation, co-led by Andreessen Horowitz and Thrive Capital with Nvidia participating as a strategic investor. The round would nearly double the $29.3 billion price tag it hit in November 2025, five months ago, and is reportedly already oversubscribed. Behind the number is a revenue curve almost nobody in B2B software has matched: $0 to $2 billion ARR in roughly three years.
The round, by the numbers
Bloomberg first reported the talks on April 17, and CNBC, TechCrunch, TechStartups, and TFN have all corroborated the core terms in the days since. Here’s what’s on the table:
| Detail | Value |
|---|---|
| Round size | $2B+ (oversubscribed) |
| Post-money valuation | ~$50B |
| Co-leads | Andreessen Horowitz, Thrive Capital |
| Strategic participant | Nvidia |
| Prior valuation | $29.3B (Series D, Nov 2025) |
| Time since last round | ~5 months |
| Round number | ~6th in roughly 30 months |
The most striking line in that table is the gap between rows four and five. AI coding has been the single largest sub-category of Q1 venture spending, and Cursor’s curve is what the headline number is being justified on. Anysphere closed its Series D at $29.3B in November. It’s now reportedly raising at a price that’s 70% higher, less than half a year later, on terms that are already oversubscribed. That doesn’t happen unless the revenue is real.
An ARR curve with no quiet period
I’ve been following AI coding tools on this blog since Cursor was a nights-and-weekends download for Anthropic API power users. What the company has done in the last 15 months compresses a decade of normal SaaS growth into a year and a quarter.
The milestone timeline, as reported across multiple trackers:
- Jan 2025 — $100M annualized revenue
- Jun 2025 — $500M
- Nov 2025 — $1B
- Feb 2026 — $2B
ARR doubled in three months, on a base that was already a billion. TheNextWeb summarized the trajectory as “zero to $2 billion ARR in roughly three years,” which it called the fastest-scaling B2B software ramp on record. Several investors I know who passed on Cursor’s early rounds have been, let’s say, very quiet lately.
The company is reportedly telling new investors it expects to exceed a $6 billion annualized run rate by the end of 2026. Triple current revenue in eight months. Every time I type that number I want to hedge it, but it’s what’s in the deck.
Why Nvidia is in the round
Andreessen Horowitz co-leading a top-5 AI deal is the least surprising thing about this story. Thrive Capital is a repeat backer of fast-scaling AI infra. The interesting name is Nvidia.
Nvidia participating as a strategic investor in an AI coding IDE has nothing to do with GPU billings. Cursor runs inference through OpenAI, Anthropic, and its own Sonic/Composer models. The point is distribution. Every Fortune 1000 engineering org adopting Cursor is an org where Nvidia wants to be the default hardware for local fine-tuning, on-prem inference, and the internal agent infrastructure those engineering teams will be asked to ship next. Getting cap-table adjacency to the fastest-growing dev tool since Slack is cheap insurance against Intel and AMD making inroads in that segment.
The competitive map, drawn honestly
A year ago, most people framed this market as Cursor vs. GitHub Copilot. That framing is now obsolete. The real map looks like this:
| Tool | Backer / Parent | Position |
|---|---|---|
| Cursor | Anysphere (independent, now ~$50B) | Standalone IDE, default for AI-first shops; rated #1 in JetBrains’ 2026 dev survey |
| GitHub Copilot | Microsoft | ~4.7M paid seats, enterprise default via MS bundles |
| Claude Code | Anthropic | Terminal + SDK, strong with devs already on Claude |
| Windsurf | Cognition (acquired remainder mid-2025 after OpenAI deal collapsed) | IDE-style, now part of Devin’s parent company |
| Amazon Q Developer | AWS | AWS-native, enterprise lock-in play |
| Gemini Code Assist | Google Cloud-native, free tier as wedge |
The reason Cursor can justify a $50B valuation at this stage is that it’s the one fully independent IDE player left at scale. Windsurf went through a chaotic split last summer, with Google licensing the tech and hiring the CEO and Cognition acquiring the remaining product and team. GitHub is Microsoft. Q and Gemini are hyperscaler bundles. If you’re a CTO who doesn’t want your coding IDE owned by whichever cloud you’re trying to avoid lock-in with, Cursor is approximately the only answer, and that’s a moat worth real money even as it narrows.
The flip side of that position is the one Fortune’s Jessica Mathews pointed out back in March: Cursor’s best product remains a thin client on top of Anthropic’s and OpenAI’s frontier models. If Anthropic pushes Claude Code harder into the IDE slot, or if OpenAI tightens API terms in a way that hurts third-party IDEs, the margin shape changes. The ~$6B projected run rate by year-end implies the founders don’t think that’s going to happen, or that they’ll out-ship whoever tries.
What the $2B actually funds
Rounds this size don’t exist to extend runway. Cursor is likely cash-flow positive at $2B ARR with the kind of margins hosted dev tools typically print. The money goes somewhere else.
Based on how Anysphere has been recruiting and what Michael Truell has said publicly, three buckets account for most of it.
The first is proprietary models. Cursor already ships Composer as a first-party model for agent loops, alongside the Sonic preview model that gets routed through the editor. A credible in-house frontier effort would reduce dependence on Anthropic and OpenAI and defend gross margin. That’s an expensive project: training runs, a multi-year GPU commitment, and the research hiring to match. Nvidia on the cap table is not incidental here.
The second is agentic coding infrastructure. Cursor 3 reoriented the product around managing agents rather than writing code keystroke-by-keystroke. The infrastructure to run long-horizon agents across a customer’s repo (sandboxed VMs, tool servers, review workflows) is a compute bill nobody’s publicized yet, but it’s not cheap.
The third is the enterprise sales motion. 70% of the Fortune 1000 is a customer-count claim, more than a seat-count claim. Converting those footholds into full-company rollouts is a field-sales problem, and field sales is what burns the kind of money a $2B round implies.
The IPO question
Anysphere has been consistently described as on track to be one of the headline AI IPOs of the next 18 months. The Information reported in March that Cursor was weighing a listing in the same window as OpenAI (~Q4 2026) and Anthropic (recently valued at $380B).
At $50B post-money, the private market is pricing Cursor at roughly 25× current ARR and ~8× projected year-end ARR. Those are aggressive multiples by historical SaaS standards but defensible for a category leader compounding at this rate, in the same broad neighborhood that fast-growing infra companies have commanded in recent IPO windows. The public-market question is whether that growth curve can survive Claude Code and Copilot compressing pricing in the mid-market segment that’s been Cursor’s volume driver. My guess: yes if Composer (and whatever follows it) becomes genuinely competitive at tab completion and short-loop agents, no if first-party models stall.
What this means for developers
A few practical takeaways for anyone working with Cursor day to day.
Pricing stability for Cursor’s current tiers is more likely after this round. A well-funded market leader doesn’t need to chase cheaper per-seat prices to stay in the game. Don’t expect a free-tier fire sale; do expect enterprise-plan expansion.
Lock-in is also creeping up. Cursor 3’s agent orchestration is more differentiated than tab completion ever was. If your team builds out review flows, custom agents, and repo-specific configurations in Cursor, you’re putting down roots that don’t move cleanly to Copilot.
Expect more first-party model news in the next six months or so. Composer is already live and the Sonic preview hints at where they’re going, but a round of this size without a near-term frontier-model story attached would be hard to defend at the price.
If you were holding off comparing Cursor to Claude Code or Copilot, don’t any longer. The market isn’t getting simpler. We have a detailed Cursor vs. Claude Code vs. Windsurf breakdown and a separate real-cost comparison of Cursor vs. GitHub Copilot if you want numbers before you pick.
The honest caveat
This is a breaking-news story built on reporting from Bloomberg, CNBC, and secondary outlets, with no Anysphere press release behind it. The round isn’t closed. The ARR milestones are sourced from investor briefings that made their way to the press, and Anysphere has not published audited financials. “Oversubscribed” is a term of art that means different things at different points in a process. None of these data points are wrong in any way I can verify, but they should be read with the usual hedging that applies to pre-close venture rounds.
What’s not hedged is the direction. Six rounds in roughly 30 months, each one at a materially higher price, with the current lead investor list being a16z + Thrive + Nvidia, looks less like venture exuberance than like a tool engineers keep paying for.
FAQ
How much is Cursor raising?
More than $2 billion, according to reporting by Bloomberg and CNBC. The round is co-led by Andreessen Horowitz and Thrive Capital, with Nvidia participating. It is reportedly already oversubscribed, though final terms and close date are not yet public.
What is Cursor’s new valuation?
Approximately $50 billion post-money. That’s up from a $29.3 billion Series D closed in November 2025, about a 70% jump in five months.
Who owns Cursor?
Cursor is the product of Anysphere, a company founded in 2022 by four MIT graduates: Michael Truell (CEO), Sualeh Asif, Arvid Lunnemark, and Aman Sanger. Major investors include Andreessen Horowitz, Thrive Capital, Benchmark, and (if the new round closes) Nvidia.
What is Cursor’s ARR?
Annualized revenue reportedly crossed $2 billion in February 2026, up from $1 billion in November 2025 and $100 million in January 2025. The company projects a $6 billion run rate by the end of 2026, per investor briefings cited in press reports.
Is Cursor going to IPO?
Anysphere has been publicly weighing a public listing in the Q4 2026 window, alongside OpenAI and Anthropic. Nothing is filed. A round of this size and the accompanying investor lineup suggest the option is being kept open, but there is no concrete timeline.
Why is Nvidia investing in Cursor?
Strategic distribution more than direct revenue. Cursor doesn’t buy Nvidia GPUs directly at meaningful scale; it runs inference through frontier-model providers. But the Fortune 1000 engineering teams using Cursor are the same teams building out on-prem AI infrastructure, and cap-table adjacency to the fastest-growing dev tool in that segment is worth money to Nvidia regardless of where the immediate revenue lands.
Bottom line
A $50B valuation for an AI coding IDE is a number that would have sounded absurd twelve months ago. Today the revenue math actually supports it. Cursor is on pace to triple ARR this year, 70% of the Fortune 1000 is already a customer, and the company now has Nvidia’s distribution muscle pointed at its enterprise motion. The open question is whether Anthropic’s Claude Code, GitHub Copilot, or a Cognition-led Windsurf revival can compress Cursor’s margin fast enough to matter before the IPO window opens.
Sources
- Bloomberg: AI coding startup Cursor in talks to raise $2 billion in funding. The original report on April 17, 2026 (paywalled).
- CNBC: AI startup Cursor in talks to raise $2 billion at $50B valuation. Independent corroboration on April 19, 2026.
- TechStartups: Cursor in talks to raise $2B as revenue surges. ARR milestone timeline and competitor map.
- TheNextWeb: Cursor/Anysphere $2B funding at $50B valuation. Growth metrics and founder background.
- Fortune: Cursor’s crossroads, the rise of a $30B AI startup. Context on Cursor’s model-dependency risk.
- Anysphere on Wikipedia. Founder list, round history, company background.