Cursor

Zero to $1B ARR. Four MIT dropouts. Under three years.

In 2022, Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger turned down lucrative offers from Google, Meta, and Jane Street to leave MIT and build something nobody had built before: an AI that doesn't just suggest the next line of code, but genuinely understands what you're trying to build — and helps you build it. The result is Cursor, the fastest B2B company in history to reach $1 billion in annual recurring revenue, and the code editor that 1 million developers now open every single day.

SaaStr called it "the fastest value creation in B2B history." Jensen Huang named it his favorite enterprise AI service. OpenAI reportedly offered to acquire it for billions — and was turned down.

What Makes Cursor Different

Cursor is a fork of VS Code — familiar enough that developers adopt it in minutes, different enough that switching back feels like losing a limb. The AI layer is not bolted on. It is the foundation. Every feature — Tab autocomplete, Cmd+K inline edits, Composer agent, Codebase Chat — is built around the idea that the editor should understand your entire project, not just the file you have open.

• Tab Autocomplete — predicts not just the current line but the next logical edit across your file, learning your personal coding patterns as you go; the fastest autocomplete in any AI editor
• Composer / Agentic Mode — describe a feature in plain English; Cursor plans, writes, and edits across multiple files simultaneously like a senior engineer executing a sprint ticket end-to-end
• Cmd+K Inline Edit — highlight any block of code, describe the change, get a clean diff in under a second
• Codebase Chat — index your entire repository and ask questions in natural language; Cursor returns answers grounded in your actual code, not generic documentation
• Bugbot — always-on automated debugging agent integrated with GitHub; reviews pull requests, flags issues, and suggests fixes at $40/seat/month
• Cloud Agents — long-running autonomous agents that execute multi-step tasks in the background without requiring the editor to stay open
• MCP Support — connect any Model Context Protocol server to give agents access to external tools, databases, and APIs
• Cursor Marketplace — a library of community-built agents, rules, and skills that extend Cursor's capabilities for specific stacks and workflows

Model Agnostic by Design

Cursor routes tasks to Claude (Anthropic), GPT-4o and o-series (OpenAI), Gemini (Google), and its own internally optimized models depending on context. Switching models is a single dropdown — no new interface, no configuration, no workflow disruption. This architecture is why NVIDIA and Google made strategic investments in the Series D: a model-agnostic editor with 1M+ daily users is one of the highest-value inference distribution channels in the industry.

Who Uses Cursor

• 1 million+ daily active developers as of December 2025
• 50,000+ engineering teams globally
• Stripe, Ramp, Brex, Rippling, Samsara, OpenAI, Figma, Datadog, and the majority of the Fortune 500
• 9,900% year-over-year ARR growth — the fastest in enterprise software history

Funding & Valuation

• Seed (2023): $8M led by OpenAI Startup Fund; angels: Nat Friedman (ex-GitHub CEO), Arash Ferdowsi (Dropbox co-founder)
• Series A (August 2024): $60M led by a16z at $400M valuation
• Series C (June 2025): $900M led by Thrive Capital at $9.9B valuation; $500M ARR milestone
• Series D (November 2025): $2.3B co-led by Accel and Coatue at $29.3B valuation; strategic checks from NVIDIA and Google; $1B+ ARR milestone
• Total raised: $3.365B as of December 2025