SpaceX 600亿美元收购Cursor：AI编程战争进入新阶段 | SpaceX $60B Cursor Acquisition: AI Coding Wars Enter New Phase
2026-06-18 | WDSEGA
2026年6月16日，SpaceX向SEC提交文件，确认以6000亿美元估值换股收购Cursor母公司Anysphere。
这是AI编程工具赛道迄今最大的一笔收购，也是SpaceX第一次把触角伸进软件开发工具链。
为什么是Cursor
Cursor在AI编程工具里的位置，有点像2012年的GitHub。
数字：
- 月活用户超过 500万（2026年Q1数据）
- 年化收入 $300M+（主要靠Pro订阅 $20/月）
- 开发者净推荐值（NPS）72（软件工具类顶尖水平）
- 2025年完成了 $150M 估值 $2.5B 的B轮
SpaceX看中Cursor的不是收入，是入口。
SpaceX要做什么
马斯克同时拥有 SpaceX 和 xAI（Grok的母公司）。这两家公司目前是独立运营的，但Cursor的收购让这个局面变得微妙。
最可能的整合路径：
- Grok + Cursor：xAI的模型通过Cursor直接触达500万开发者，绕过OpenAI/Anthropic的API依赖
- SpaceX工程栈：SpaceX内部的软件工程工具链（火箭遥测、星链调度、发射模拟）逐步迁移到Cursor + Grok
- 星舰操作系统：长远看，马斯克要把Cursor做成”星舰上的IDE”——宇航员/工程师在轨维修、调试、重写代码
这个逻辑听起来疯狂，但马斯克做过更疯狂的整合（Tesla + SpaceX + Neuralink + xAI 四家公司共享部分AI基础设施）。
对AI编程赛道的影响
Anthropic是最大输家。
Cursor是Claude API最大的第三方客户之一（开发者用Cursor时，默认调用Claude）。收购完成后，Cursor会逐步切换到Grok——这意味着Anthropic每年损失 $50M+ 的API收入。
OpenAI反而受益。
Cursor被收购后，很多开发者会担心”马斯克控制了我的IDE”，转而投向GitHub Copilot（OpenAI背书）或Windsurf（Anthropic投资）。OpenAI的Copilot市场份额可能因此上升5-8个百分点。
Google是唯一观望者。
Gemini Code Assist的用户量级远小于Cursor，但通过Google Cloud捆绑销售正在暗中增长。Cursor被收购后，Google Cloud的客户会多一个问题：”你们有没有类似Cursor的东西？”——这可能会逼Google加速收购或自研。
$60B估值贵不贵
横向对比：
| 公司 | 估值 | 年化收入 | P/S |
|---|---|---|---|
| Cursor（本次） | $60B | ~$300M | 200x |
| GitHub（2018年微软收购） | $7.5B | ~$200M | 37.5x |
| Figma（2022年Adobe拟收购） | $20B | ~$400M | 50x |
| Unity（2021年峰值） | $70B | ~$1.1B | 63x |
200x P/S看起来夸张，但要考虑：
- 收入增速：Cursor年化收入从2025年Q1的 $80M 增长到2026年Q1的 $300M+，年增速 >300%
- 马斯克溢价：马斯克系公司的估值从来不能用传统倍数算（Tesla P/S 曾经到过 20x）
- 战略价值：SpaceX买的不只是Cursor，是500万开发者的使用习惯和xAI的模型分发渠道
所以 $60B 在战略上说得通，在财务上很贵。
开发者的担心
Cursor被收购后，最常被问到的问题：
“我的代码会不会被用来训练Grok？”
Cursor官方回应（6月17日）：”数据隔离政策不变，代码不会用于训练任何模型。”
但这个承诺在马斯克系公司里的可信度……历史记录不太好。Twitter（现在X）被收购后就改了API政策和数据使用政策。
实际建议：如果你在做敏感项目（军工、金融、医疗），现在可以考虑迁移到Windsurf或GitHub Copilot。
结论
这次收购的本质是：马斯克在搭建一个从模型（xAI）到编辑器（Cursor）到部署环境（SpaceX算力）的全栈AI编程生态。
目标不是跟Anthropic或OpenAI竞争，是让”用Grok写代码”变成一件自然的事——就像现在”用Claude写代码”一样自然。
收购预计在2026年Q3完成。在此之前，Cursor的产品路线暂时不会变。但Q3之后，事情会开始变得有趣。
SpaceX $60B Cursor Acquisition: AI Coding Wars Enter New Phase
On June 16, 2026, SpaceX filed with the SEC confirming a $60 billion all-stock acquisition of Anysphere, the parent company of Cursor.
This is the largest acquisition in the AI coding tools赛道 to date, and SpaceX’s first move into the software development toolchain.
Why Cursor
Cursor’s position in AI coding tools is roughly like GitHub’s in 2012.
The numbers:
- MAU: 5M+ (Q1 2026)
- ARR: $300M+ (mainly $20/month Pro subscriptions)
- Developer NPS: 72 (top-tier for software tools)
- 2025: $150M Series B at $2.5B valuation
SpaceX isn’t buying Cursor for revenue — it’s buying the entry point.
What is SpaceX Planning?
Musk simultaneously owns SpaceX and xAI (parent of Grok). These operate independently — for now. The Cursor acquisition changes the calculus.
Most likely integration path:
- Grok + Cursor: xAI’s models reach 5M developers directly through Cursor, bypassing OpenAI/Anthropic API dependency
- SpaceX engineering stack: Internal software tooling (rocket telemetry, Starlink scheduling, launch simulation) gradually migrates to Cursor + Grok
- Starship OS: Long-term, Musk wants Cursor to become “the IDE on Starship” — astronauts/engineers doing on-orbit repairs, debugging, rewriting code
Sounds crazy? Musk has done crazier integrations (Tesla + SpaceX + Neuralink + xAI sharing some AI infrastructure).
Impact on AI Coding Landscape
Anthropic is the biggest loser.
Cursor is one of Claude API’s largest third-party customers (developers using Cursor default to Claude). Post-acquisition, Cursor will gradually swich to Grok — meaning Anthropic lises $50M+/year in API revenue.
OpenAI actually benefits.
Post-acquisition, many developers will worry “Musk now controls my IDE” and switch to GitHub Copilot (OpenAI-backed) or Windsurf (Anthropic-invested). OpenAI’s Copilot market share could gain 5-8 points from this.
Google is the only spectator.
Gemini Code Assist’s user base is far smaller than Cursor’s, but Google Cloud bundling is quietly growing. Post-acquisition, Google Cloud customers will ask: “Do you have a Cursor alternative?” — which may push Google to acquire or build.
Is $60B a Fair Valuation?
Peer comparison:
| Company | Valuation | ARR | P/S |
|---|---|---|---|
| Cursor (this deal) | $60B | ~$300M | 200x |
| GitHub (Microsoft 2018) | $7.5B | ~$200M | 37.5x |
| Figma (Adobe 2022) | $20B | ~$400M | 50x |
| Unity (2021 peak) | $70B | ~$1.1B | 63x |
200x P/S looks extreme, but consider:
- Revenue growth: Cursor ARR grew from $80M in Q1 2025 to $300M+ in Q1 2026 — >300% YoY
- Musk premium: Valuations of Musk-associated companies never made sense by traditional multiples (Tesla once traded at 20x P/S)
- Strategic value: SpaceX isn’t just buying Cursor — it’s buying 5M developers’ habits and xAI’s model distribution channel
So $60B makes strategic sense, financial sense is stretch.
Developer Concerns
Post-acquisition, the #1 asked question:
“Will my code be used to train Grok?”
Cursor official response (June 17): “Data isolation policy unchanged. Code will not be used to train any model.”
But the credibility of this promise at a Musk-owned company… the historical record isn’t great. Twitter (now X) changed API policy and data use policy post-acquisition.
Practical advice: If you’re working on sensitive projects (defense, finance, healthcare), consider migrating to Windsurf or GitHub Copilot now.
Conclusion
The core of this acquisition: Musk is building a full-stack AI coding ecosystem from model (xAI) to editor (Cursor) to deployment (SpaceX compute).
The goal isn’t competing with Anthropic or OpenAI — it’s making “writing code with Grok” as natural as “writing code with Claude” is today.
Deal is expected to close Q3 2026. Until then, Cursor’s product roadmap won’t change. But post-Q3, things will get interesting.