SpaceX swallows Cursor for $60 billion: Musk wants more than just an IDE

SEC Filing Finalized: From Option to Formal Merger

In the early hours of June 16, SpaceX submitted a merger agreement to the U.S. Securities and Exchange Commission — its wholly-owned subsidiary X67 Inc. will merge with Anysphere, with Cursor continuing as a wholly-owned subsidiary of SpaceX. All issued common and preferred shares of Cursor will be automatically converted into SpaceX Class A common stock based on an implied equity value of $60 billion, with the share exchange ratio referencing the VWAP (volume-weighted average price) for the seven consecutive trading days prior to closing. The transaction is expected to close in Q3 2026.

This actually had been foreshadowed earlier. Back in April, SpaceX officially announced it had obtained an acquisition option for Cursor — either buy it outright for $60 billion or spend $10 billion to lock in deep cooperation. At the time, Musk left himself an opening, essentially turning the acquisition risk into a "controllable experimental cost." Two months later, he directly exercised the option.

From $29.3 billion to $50 billion to now $60 billion, Cursor’s valuation has doubled in the past six months. Bear in mind, the Anysphere team only has about 150 people — this valuation implies a per capita valuation of $400 million, a ceiling that no other startup in history has ever reached.

What Does $60 Billion Buy? Not a Tool, but a Developer Entry Point

On the surface, a rocket manufacturer like SpaceX buying a code editor seems baffling. But if you factor xAI into the equation — remember, xAI was merged with SpaceX back in February — the logic becomes clearer.

Musk himself admitted: "Grok is behind in coding." This is blunt truth. In the AI programming race, OpenAI has Codex, Anthropic has Claude Code, Microsoft has Copilot, and xAI has yet to produce a comparable product. What Cursor holds in its hands is the rarest asset in this track:

• Over 1 million paying users
• 64% of the Fortune 1000 are its customers
• ARR has exceeded $1 billion, with internal forecasts aiming for $6 billion by the end of 2026
• Over 30,000 NVIDIA developers use it intensively on a daily basis

More importantly, Cursor doesn’t just have user volume — it has high-quality training data from real-world work scenarios. No amount of scraped code from the internet can compare to watching an experienced engineer actually debug, refactor, and write tests in a large-scale enterprise codebase — this is the goldmine for training the next generation of programming models.

For xAI, rather than spending years building another IDE to fight for users, it’s smarter to buy the entry point outright. Integrating Grok into Cursor instantly gives them a distribution channel to over a million developers and a continuous feedback loop.

Cursor Also Struggled to Refuse: Compute Anxiety Is Real

This isn’t a one-sided “being acquired” — Cursor itself has strong motivation.

AI programming tools are a peculiar business — traditional SaaS has near-zero marginal cost per additional user, but AI tools burn real compute every time they generate a line of code. Cursor was in negative gross margin for a long time in its early days, sustained only by funding. It wasn’t until the launch of proprietary model Composer at the end of 2025, along with a “compute scheduling system”: cheap models for simple tasks, top-tier models for complex full-scale refactoring, that its enterprise operations turned profitable.

But in March this year, an embarrassing story surfaced — some developers intercepted proxies and discovered that Cursor’s hyped “cutting-edge proprietary model” Composer 2 was essentially a fine-tuned version of Moonshot’s open-source Kimi K2.5. This indicated Cursor’s steps on the proprietary model path were shaky, still dependent on others.

Cursor’s merger announcement was candid:

“We’ve always wanted to push our training work further, but have been limited by compute resources. Through this collaboration, our team will leverage xAI’s Colossus infrastructure to dramatically enhance our models’ intelligence.”

Colossus is xAI’s supercomputing cluster deployed in Memphis, equivalent to a million H100-scale capacity. For Cursor, aligning with the SpaceX/xAI giant tree means compute will no longer be the limiting ceiling.

Musk’s AI Puzzle: Full Stack Control from Ground to Space

Looking at SpaceX’s actions over the past year, this acquisition is far from random. Musk’s goal is a complete loop from models to compute to application entry point, and then to energy and network:

| Layer | Asset | |-------|-------| | Application entry point | Cursor (newly added) | | Model | xAI / Grok / Composer | | Ground compute | Colossus supercomputer | | Global network | Starlink | | Future compute | Space-based data centers (planned) | | Chips | In-house chip factory (planned) |

Musk once said in an internal letter to SpaceX staff: “Only after deploying data centers in space to better harness solar energy will humanity become a multi-planetary species.” This sounds sci-fi, but paired with his ongoing space data center plan, the ambition already spills beyond Earth.

SpaceX itself is preparing for an IPO with a target valuation of $1.75 trillion. Adding an AI asset like Cursor with both aggressive cash flow and growth completely changes the pitch to secondary markets — this is no longer just a rocket company, but a "technology giant with complete core AI capabilities."

What Does This Mean for Developers: Will Cursor Change?

This is the developer community’s biggest concern. So far, a few details are known:

1. The agreement includes a “lock-up” clause: Cursor commits not to be acquired by any other company for one year, blocking the possibility of OpenAI or Anthropic swooping in;
2. Preferential use of xAI’s Colossus for training, meaning future Composer versions will be deeply tied to xAI’s infrastructure;
3. Cursor’s current multi-model routing capability (users can freely switch between OpenAI, Anthropic, xAI, Google models) should remain in the short term — it’s core to its product strength, but whether Grok and Composer will be pushed by default in the long term is uncertain.

There’s a delicate tension here: one of Cursor’s moats is model neutrality — developers can pick the most suitable model for their task. If it becomes xAI’s direct offspring, will Claude and GPT still have equal standing inside Cursor? This is a key point to watch.

For enterprise clients, trust commitments like “Zero Data Retention” and SOC 2 Type 2 certification may face re-evaluation. When Cursor’s parent company becomes xAI/SpaceX, Fortune 1000 companies previously worried about their code being used for model training may reassess. This could be Cursor’s biggest business headache in the near term.

AI Programming Industry Landscape: Fully Entering the Era of Giants

Cursor being absorbed marks the closure of the independent AI programming startup window. Look at the current table:

• GitHub Copilot: Microsoft ecosystem, king of distribution, $10/month volume play
• Claude Code: Anthropic’s own product, terminal-native, beloved by senior back-end engineers
• Codex/ChatGPT: OpenAI official, direct model connection advantage
• Cursor: now under SpaceX/xAI, enterprise-level dominance
• Windsurf: price slasher, $15/month targeting small and mid-size developers

Behind each sits a massive platform or model provider. The era of independent IDEs is over, because compute and model costs inevitably keep any neutral player in long-term losses.

Silicon Valley’s hottest word right now is “Vibe Coding” — the 2025 term of the year, with search heat up 6700%. Stripe’s CEO revealed that 5% of the company’s code merge requests are entirely generated by AI agents (dubbed Minions), with humans only reviewing. But this productivity surge hasn’t been evenly distributed among all developers:

The real beneficiaries are the longest-serving, most experienced senior architects. They have entire system blueprints in their heads, and now, like orchestra conductors, they direct AI through “vibes.” Newcomers who rely too heavily on AI may produce code that appears to run but is riddled with bugs, leading to a “Vibe Coding Hangover.”

It’s a rather harsh trend — technical judgment will fully outweigh coding speed.

An Observation: Should Developers Be Concerned?

In the short term, Cursor’s acquisition has little impact on daily use — product iterations will continue. But if you rely heavily on Cursor for enterprise development, a few suggestions:

• Keep backup options for model switching: In case Cursor starts leaning toward Grok, you should be able to switch back to Claude or GPT at any time;
• Watch for changes in data policies: Privacy terms will likely be updated after closing, so enterprise legal teams should get involved early;
• Don’t lock your entire workflow into a single tool: High switching costs are Cursor’s moat, but also your exposure to risk.

Incidentally, for teams needing to call multiple models for comparative testing, aggregation platforms like OpenAI Hub (openai-hub.com) — one key connecting GPT, Claude, Gemini, DeepSeek and other mainstream models, compatible with OpenAI format and directly accessible in China — can serve as a backup at the model layer. Especially now, when IDE ownership changes and model routing policy may shift, decoupling model supply is a prudent move.

Final Thoughts

$60 billion for a 150-person company sounds crazy. But placed within the “model + compute + application entry point” three-layer AI competition framework, this deal is in fact the most rational — Musk is plugging his shortfall in the most direct way.

AI companies are shifting from pure software firms into heavy infrastructure enterprises. Who trains the model, where the compute comes from, and which entry point the users use — the boundaries between these three are breaking down. Cursor is the first major chip to land in this restructuring, but it certainly won’t be the last.

Who’s next? Windsurf? Replit? Or some new unicorn making enterprise knowledge work agents? This “application entry point battle” has only just begun.

References

• SpaceX Plans $60 Billion Acquisition of Cursor Developer Anysphere - linux.do: Community discussion and information summary on this merger