Cursor (Anysphere) Valuation 2026: $29.3B & $2B ARR

Cursor's parent company Anysphere is valued at $60B as of June 2026 — that's the all-stock price SpaceX (now merged with xAI) agreed to pay to acquire it on June 16, 2026, with the deal expected to close in Q3. That's up from a $29.3B Series D in November 2025 and reported April 2026 talks above $50B — less than four years after the four MIT founders shipped the first build of a VS Code fork with AI baked in.

That's the short answer. The longer answer is more interesting — because the round-by-round math of how a free side project became a ~$30B company in under 40 months explains a lot about how AI valuations are actually getting set in 2025 and 2026.

Cursor AI Valuation and Funding: The Round-by-Round Math

Cursor's parent Anysphere has raised approximately $3.4B across six priced rounds since 2022, with valuation climbing from ~$8M post-money at YC to $9.9B in 30 months and then $29.3B by November 2025. The Series D — $2.3B at $29.3B post-money — was co-led by Accel and Coatue, and as of April 2026 the company was reportedly in oversubscribed talks for $2B+ above a $50B valuation, co-led by Andreessen Horowitz and Thrive Capital with Nvidia as a strategic investor. The pace of mark-ups, not the absolute valuation, is what separates Cursor from every other AI infrastructure bet in this cycle.

Mark-up math: $60M → $60B is a ~1,000x step-up across roughly 34 months. The Series C to Series D jump alone was ~3x in five months, and the SpaceX/xAI acquisition price doubled the November 2025 mark again in about seven.

The exit cap on that curve arrived faster than anyone modeled: on June 16, 2026 — days after SpaceX's blockbuster IPO and less than two months after first floating the deal — SpaceX agreed to acquire Anysphere for $60B in an all-stock transaction, folding Cursor into the xAI subsidiary to anchor an enterprise AI-coding push against OpenAI and Anthropic. The deal is expected to close in Q3 2026 subject to regulatory approval. See the AI Valuations dashboard for the live comp set.

How Cursor's ARR Reached $2B Faster Than Any Other SaaS

Cursor crossed $100M ARR in January 2025, roughly 21 months after the IDE launched in March 2023. That broke the previous record held by Wiz (18 months from $1M to $100M ARR) for application-layer SaaS — and the ramp kept compounding: $500M by June 2025, $1B by November 2025, and past $2B annualized by February 2026, with internal projections reportedly above $6B for end of 2026. The growth shape is unusual: bottoms-up developer adoption that converted to team and enterprise contracts inside large customers without a traditional outbound sales motion.

Who Founded Cursor and Why MIT Matters

Anysphere was founded in 2022 by four MIT undergraduates: Michael Truell (CEO), Sualeh Asif (CTO), Aman Sanger, and Arvid Lunnemark. The company went through Y Combinator's S22 batch and shipped Cursor publicly in March 2023, three months after the GPT-3.5 ChatGPT launch made code-completion-as-a-product viable in a way Copilot hadn't yet shown at scale. Truell was 21 when he raised the Series A; he was 24 at the time of the $9.9B Series C and remains one of the youngest CEOs of a decacorn-valued private company in history — now running a business priced at $29.3B.

Cursor AI Valuation in Context: How It Compares to GitHub Copilot and Windsurf

At $29.3B post-money on roughly $1B ARR, Cursor's November 2025 Series D priced at ~29x trailing ARR — a multiple that's high by historical SaaS standards (5-15x) but in line with frontier AI infrastructure pricing in 2025-2026, and one the $2B+ ARR run-rate by February 2026 has already grown into. Compared to direct competitors, Cursor sits between Microsoft-bundled distribution (Copilot) and venture-funded standalone independents (Windsurf/Codeium, Replit, Sourcegraph). The full competitive layout:

Note: Windsurf agreed to a Google licensing deal in mid-2025 after a proposed $3B OpenAI acquisition collapsed, materially changing competitive dynamics in the second half of 2025. See AI Valuations Dashboard for the live valuation table.

Cursor's Customer Base and Why Enterprises Pay For It

Cursor's customer list reads like a who's-who of AI-native and post-2020 enterprise software: Stripe, Shopify, Instacart, OpenAI, Perplexity, Mercedes-Benz, Samsung, Johnson & Johnson, US Foods, and reportedly more than half of the Fortune 500 in some form of paid usage. The pricing model is simple: $20/seat/month for Pro, $40/seat/month for Business, and custom enterprise tiers above 100 seats with SSO and audit logging. Cursor has not disclosed seat count directly, but a $2B+ run rate at blended per-seat pricing implies a paid base in the millions of seats globally.

Why Cursor's Valuation Math Is Different From Other AI Companies

Three things make Cursor's $29.3B price tag actually defensible — not the multiple itself, which is rich, but the underlying revenue shape that makes it worth pricing that way at all:

What the Anysphere Rounds Tell Us About AI Valuations in 2025-2026

Cursor's $29.3B Series D on roughly $1B ARR — a ~29x trailing multiple — is now the implicit benchmark for AI infrastructure companies that ship product to developers. Six months later, Lovable, Replit, Bolt, and Codeium all repriced against the Cursor comp. That's how a single round resets a category. The same dynamic applies to the broader AI application layer: when investors see usage-based revenue compounding 10-20% monthly with real enterprise logos on it, they will pay 30-40x trailing ARR. Without those signals, multiples collapse fast — see the 2022 SaaS reset.

The harder question is whether the $29.3B price — let alone the reported $50B+ — survives the next 24 months. Three things have to hold for that multiple to compress only modestly rather than collapse: (1) ARR growth has to stay above 8% monthly; (2) Cursor has to extend product surface area beyond the IDE into agents, terminal, and CI; (3) gross margin has to climb from the current ~40% to at least 60% as inference economics improve and Cursor moves more workloads onto cheaper open models. None of those are certain. All three are plausible.