SpaceX Bought Cursor for $60 Billion. Let's Talk About What That Actually Means.
A rocket company just paid $60 billion for a VS Code fork.
That sentence should feel absurd. And yet here we are: SpaceX — fresh off the largest IPO in history ($75B raised, $2T+ market cap within 48 hours) — announced today that it’s acquiring Anysphere, the company behind Cursor, in an all-stock deal valued at $60 billion.
This is not normal. Let’s break down what’s actually happening.
The numbers don’t make sense (until they do)
Cursor’s financials are legitimately impressive:
- $4B annualized revenue as of June 2026
- $2.6B from enterprise B2B alone
- Growth from $100M ARR (early 2025) → $1B (November 2025) → $3B (May 2026) → $4B (June 2026)
- 1 million+ daily active developers
At $4B ARR, a $60B valuation implies a 15x revenue multiple. For a company growing 40x year-over-year, that’s not insane by SaaS standards. It’s aggressive, but defensible.
But here’s the thing: SpaceX didn’t pay $60B for a code editor.
Cursor is a VS Code extension with a really good AI layer on top. The editor itself is commodity open-source technology. Any team with 10 engineers could rebuild the UI in three months. What they can’t rebuild:
- The user base. A million developers with their workflows, settings, and muscle memory locked into Cursor. These people send their proprietary code through Cursor’s servers every single day.
- The enterprise contracts. $2.6B in B2B revenue means Fortune 500 companies are paying for Cursor at scale. Those contracts take years to build.
- The data flywheel. Every keystroke, every acceptance, every rejection — Cursor has the richest dataset of how developers actually write code that exists outside of GitHub.
- The team. xAI (Grok) has been hemorrhaging engineers. Cursor’s team has been shipping faster than almost anyone in AI. SpaceX is buying talent as much as technology.
This is a compute-for-distribution trade
The real logic becomes clear when you understand what each side gets:
SpaceX/xAI gets: Distribution. Enterprise credibility. A revenue-generating AI product that developers already love. A path to compete with Claude Code, GitHub Copilot, and Windsurf without starting from zero.
Cursor gets: Chips. SpaceX built the Colossus cluster for xAI — one of the world’s largest GPU installations. Cursor has been model-agnostic (routing between GPT, Claude, and custom models), but with access to SpaceX’s compute, they can train proprietary models that don’t pay Anthropic/OpenAI per token.
As Kilo.ai put it: “When a rocket company spends $60B on a coding tool, the strategic center has moved from model quality to compute access.”
This is the new playbook: the company with the most GPUs wins the AI coding war, because the marginal cost of model inference trends toward zero if you own the silicon.
What this means for you (the developer)
Let’s be practical. If you use Cursor today, here’s what you need to think about:
Your code now flows to a Musk company
Every file you open, every prompt you write, every codebase you index — that data will be processed by a SpaceX subsidiary. For solo developers, maybe you don’t care. For enterprise teams subject to compliance requirements? For EU companies under GDPR? For anyone working on government contracts that aren’t SpaceX contracts?
This is going to create procurement headaches. Some companies will ban Cursor reflexively. Others will wait for the privacy policy changes. Smart teams are already evaluating alternatives.
Pricing will probably change
Cursor’s current pricing ($20/month Pro, $40/month Business) is almost certainly underpriced for what SpaceX paid. Expect “enterprise” tiers, usage-based pricing for heavy compute, or bundling with other xAI products. The days of unlimited AI completions for $20/month are numbered.
The model-agnostic era may end
Cursor currently lets you use Claude, GPT, Gemini, or custom models. How long does that last when SpaceX is trying to make xAI/Grok competitive? The incentive to default to in-house models — or to subtly degrade the experience of competitor models — is enormous.
Alternatives exist and are maturing fast
- Claude Code — Terminal-based, different paradigm, but extremely powerful for agentic coding
- Windsurf — Direct Cursor competitor, independent
- Continue.dev — Open source, model-agnostic, VS Code native
- Zed — Built-in AI, fast, not dependent on VS Code
- OpenCode — Terminal agent, free, works with any model
- Vim/Neovim + AI plugins — If you value independence over convenience
None of these are perfect Cursor replacements today. But the market just got a massive incentive to improve.
The bigger picture: consolidation is here
This acquisition signals something beyond Cursor:
AI coding tools are not a standalone business anymore. They’re a distribution channel for compute companies. Whoever controls the IDE controls which models get used, which means controlling the flow of inference revenue.
Microsoft understood this early (GitHub Copilot → Azure → OpenAI). Google is doing it (Gemini → Android Studio + Cloud). Now SpaceX/xAI is doing it with Cursor.
The indie AI coding tool is an endangered species. If your favorite tool doesn’t have a compute partner or a path to profitability at scale, it’s either getting acquired or getting outspent.
The exceptions are open-source tools (Continue.dev, Neovim plugins) and terminal-based agents (Claude Code) that are tied to model providers rather than compute platforms. These have different economics that don’t require a $60B sugar daddy.
My take
I’ve been covering AI coding tools since this blog started. Here’s what I actually think:
The deal makes strategic sense for both sides. SpaceX gets the #1 AI coding tool overnight. Cursor gets infinite compute. It’s a marriage of distribution and infrastructure.
Developers should be worried about lock-in, not excited about features. The immediate product probably won’t change. But over 12-18 months, expect Cursor to become increasingly tied to xAI’s ecosystem. The “use any model” flexibility is the first thing that’ll erode.
$60B for what is fundamentally an AI middleware company is a sign of how overheated this market is. Cursor doesn’t own a foundation model. It doesn’t own unique data (your code belongs to you). It owns a really good prompt router and a million developers who are too lazy to switch. That’s worth something — but is it worth more than SpaceX paid for all of xAI ($125B) relative to the capabilities? Debatable.
The real winner is the developer who stays portable. Use CLAUDE.md files. Keep your workflows in scripts, not in proprietary IDE features. The companies will keep shuffling — your ability to move between tools is your actual leverage.
The AI coding wars just got more expensive. Stay flexible.
This is an AI Dev Weekly Extra — opinionated coverage of major AI development news. Subscribe to the regular newsletter for weekly roundups.