What is Cursor?
Cursor is an AI-native code editor built by Anysphere that has become the fastest-growing developer tool in history. It's a fork of VS Code — meaning all your extensions, themes, and keybindings carry over — but rebuilt from the ground up around AI-assisted coding. Cursor reached $1 billion in annual recurring revenue within 17 months of launch, shattering the previous B2B record (Slack took 15 months to hit $100M). The company hit a $29.3 billion valuation by November 2025 with roughly 40-60 employees, producing an ARR-per-employee ratio of $8-12 million that has no real precedent in software. Over 50% of Fortune 500 companies have adopted it, with Shopify, OpenAI, Coinbase, and Instacart among named enterprise clients.
Key Takeaways
- $1B ARR in 17 months — fastest B2B software scaling in history, $29.3B valuation
- VS Code fork — all existing extensions, themes, and keybindings work out of the box
- Deep codebase indexing lets the AI understand your entire project, not just the open file
- Agent mode autonomously implements features across multiple files with terminal access
- Over 50% of Fortune 500 companies adopted; Shopify, OpenAI, Coinbase among named clients
Key Features
What separates Cursor from a standard editor with AI bolted on is how deeply the AI is integrated into the editing experience. Tab Autocomplete uses a proprietary technique called "speculative edits" — rather than generating each token from scratch, it feeds your existing code back as a speculation and only diverges at points where changes are needed, enabling ~1,000 tokens/second on large models. Cmd+K lets you edit or generate code inline using natural language. Chat provides a conversational interface that's aware of your entire codebase through local indexing — you can ask questions about code you haven't opened. Agent Mode takes a task description and autonomously edits files, runs commands, reads errors, and iterates until the task is complete. Behind the scenes, a Shadow Workspace runs a hidden VS Code instance with language servers that lint-check AI output for type errors before surfacing it to you — something no competing tool replicates.
Pricing Plans (2026)
Cursor offers three tiers. The Hobby plan ($0/mo) includes 2,000 code completions, 50 slow premium requests, and access to the cursor-small model — enough to evaluate the tool meaningfully. Pro ($20/mo) removes completion limits, provides 500 fast premium requests per month (with unlimited slow requests), and unlocks all models including Claude Sonnet, GPT-4o, and Cursor's custom models. Business ($40/user/mo) adds centralized team billing, admin controls, enforced privacy mode that ensures zero data retention by model providers, and SAML SSO for enterprise environments.
Cursor vs GitHub Copilot
The comparison comes down to depth versus distribution. Cursor indexes your entire codebase locally, enabling AI suggestions that understand project-wide context — not just the file you have open. Its speculative editing engine, Shadow Workspace lint-checking, and agent mode represent genuine technical differentiation. GitHub Copilot works as a plugin within your existing editor (VS Code, JetBrains, etc.) and offers broader surface area — code completions, chat, PR summaries, code review, and Workspace — backed by GitHub's 100M+ developer distribution. Copilot is the enterprise default because procurement teams already have GitHub contracts; Cursor wins developers who try both and prefer the deeper AI integration. Many teams use both: Copilot for its GitHub integration features and Cursor as the daily coding environment.
The .cursorrules File and Why Enterprises Lock In
Power users know that Cursor's stickiest feature isn't the AI quality — it's the `.cursorrules` file. Committed to a repo's root directory, this file gives the AI standing instructions: preferred libraries, architectural constraints, naming conventions, testing patterns, and how to handle specific code patterns. The AI follows these rules across every interaction — completions, chat, and agent mode.
At team scale, this becomes a codifiable form of institutional knowledge. A new engineer's Cursor output matches the team's patterns from day one because the rules file encodes conventions that would otherwise take weeks to absorb through code review. Copilot has no equivalent persistent-context mechanism per repository. An Upwork case study showed a 25% increase in PR volume and 100% increase in average PR size among engineers using Cursor — numbers that suggest the tool doesn't just make individual developers faster, it changes how much scope a single engineer can own.
Cursor in the Remote Talent Context
Cursor adoption has become a meaningful signal in the fractional engineering market. On Pangea, developers who list Cursor proficiency tend to be the same ones who stay current with modern tooling across the board — it correlates with engineers who are invested in their own productivity. For companies hiring fractional engineers, the practical benefit is that a developer comfortable with Cursor's agent mode and codebase indexing can ramp up on an unfamiliar codebase significantly faster than one working without AI assistance. The `.cursorrules` file also matters for fractional work specifically: a well-maintained rules file means a contractor can produce code that matches your team's conventions without extensive onboarding. That's a real cost reduction for project-based engagements.
The Bottom Line
Cursor represents a bet that AI-native development environments will replace traditional editors, and its growth trajectory suggests that bet is paying off. It's technically superior to Copilot for interactive coding, but Copilot's distribution advantage means both will coexist for the foreseeable future. For companies hiring through Pangea, Cursor proficiency signals a developer who takes tooling seriously and is likely to ramp up faster on your codebase. For developers, learning to work effectively with AI-native editors is becoming a career-defining skill.
