Cursor

Cursor is, by any objective measure, the best code editor ever built. It takes the world's most popular IDE — Visual Studio Code — and rebuilds it from the foundation up around AI. Tab completion that reads your mind. Multi-file refactoring that actually understands your codebase. Agents that run in cloud VMs, test their own code with a browser, and submit pull requests with screenshots and video artifacts.

And yet. Claude Code — a terminal tool with no graphical interface, no tab completion, no syntax highlighting beyond what your shell provides — is generating five times the revenue. A tool that looks like it belongs in 1985 is outselling the most polished AI IDE on the market.

That tension is the most interesting story in developer tools right now.

Cursor launched in 2023 as a fork of VS Code with AI superpowers. The pitch was simple: instead of bolting AI onto an editor as a plugin (like GitHub Copilot), build the editor around AI from day one.

The founding team — Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger, four MIT classmates — bet that developers would pay for an editor that truly understood their codebase. They were right. Within two years, Cursor hit $500M in annual recurring revenue and a $29.3B valuation. Half the Fortune 500 adopted it. Over 90% of developers at Salesforce now use it.

The core experience centers on three capabilities:

Tab Completion — Cursor's autocomplete is eerily good. It doesn't just complete the current line; it predicts the next several lines based on context from your entire project. Developers on X call it "god-tier" and they're not exaggerating.

Composer — The flagship feature. Describe what you want in natural language, and Cursor edits multiple files simultaneously with a structured diff view. It's powered by Cursor's own frontier model — a coding-specific AI trained with reinforcement learning and codebase-wide semantic search. Most turns complete in under 30 seconds.

Cloud Agents — The newest addition. Remote AI agents run in isolated virtual machines with full development environments. They clone your repo, make changes, test them in a built-in browser, and produce merge-ready pull requests complete with screenshots, videos, and logs. BugBot, the automated code reviewer, finds issues in your PRs and auto-fixes them — 35% of its fixes get merged without any human intervention.

Cursor's pricing tells a story about the economics of AI-powered tools:

- Free: Basic features, limited AI usage - Pro ($20/mo): Full features with a credit pool worth roughly 225 Claude Sonnet requests - Pro+ ($60/mo): 3x the credits - Ultra ($200/mo): 20x the credits, plus background agents - Teams ($40/user/mo): Centralized billing and admin

The shift from flat request limits to a credit-based system in mid-2025 was controversial. Credits deplete at different rates depending on which AI model you use — a Claude Sonnet request costs more credits than a Gemini request. The community felt it was less transparent, and costs could spike unexpectedly during heavy coding sessions.

Background agents — the long-running cloud agents that can work on substantial tasks over hours — are locked behind the Ultra tier at $200/month. That's a meaningful barrier for individual developers who want the most powerful feature.

Here's the number that should keep Cursor's board up at night: Claude Code is at $2.5B in annual revenue. Cursor is at $500M. Both cost roughly $20/month at their base tier.

Claude Code is a terminal application. You type commands. It reads your files, makes edits, runs your test suite, and explains its reasoning. There's no graphical interface. No drag-and-drop. No visual diff viewer. It looks, frankly, primitive.

But developers are choosing it 5-to-1 over the polished IDE experience.

The X discourse captures the divide perfectly. Christopher Karani frames it: "Terminal vs IDE isn't the real axis. It's about control surface area — Claude Code exposes raw reasoning, Cursor abstracts it. Better for exploration, worse for production where you want guardrails."

Translation: Cursor makes AI feel familiar by wrapping it in IDE conventions. Claude Code lets you see the AI thinking in real time, and developers value that transparency more than comfort.

Many power users pair both — Cursor for daily editing and tab completion, Claude Code for autonomous complex tasks. That's the pragmatic answer, but it raises the question: if Cursor's unique value is tab completion and visual diffs, is that a $29B business or a feature?

Cursor 2.0, released in February 2026, signals where the company is heading: away from "editor with AI" and toward "agent platform that happens to have an editor."

The agent-centric interface redesign puts agents, not files, at the center of the experience. Cloud agents with computer use — AI that can actually see and interact with the software it builds — is a genuine leap. BugBot Autofix turning code review into an automated pipeline is the kind of feature that justifies enterprise pricing.

The risk is that Cursor is chasing the same destination as Claude Code and OpenAI's Codex from a more constrained starting point. An IDE is a heavy artifact. A terminal agent is lightweight and composable. As AI agents get more autonomous, the IDE's visual affordances matter less.

Cursor is an exceptional product. If you write code daily and want the smoothest possible AI-assisted editing experience, there is nothing better. The tab completion alone saves hours per week. Composer's multi-file editing is category-defining. Cloud agents with self-testing are genuinely novel.

But the market is telling a story that Cursor's features don't fully explain. The most successful AI coding tool has no interface at all. Developers, it turns out, don't want AI wrapped in familiar metaphors. They want raw reasoning they can steer.

Cursor's challenge isn't building a better editor. It's that editors might not be the right abstraction for AI-native development. The $29B question is whether they can pivot fast enough to find out.

Use Cursor if: You code daily, value visual feedback, and want the best tab completion on the market.

Skip it if: You prefer terminal workflows, need unlimited AI usage without credit anxiety, or want maximum autonomy from your AI coding partner.