Cursor reports a $60B all-stock deal with SpaceX

• Cursor said it agreed to a $60 billion all-stock deal with SpaceX, and Cursor's announcement framed it as the start of "significant improvements" to the product.
• According to Rohan Paul's filing summary, SpaceX plans to fold Anysphere into a merger vehicle called X67 Inc., with Cursor surviving as a wholly owned SpaceX subsidiary if the deal closes.
• The consideration is stock, not cash, and that same filing summary says Cursor shareholders convert into SpaceX Class A shares using a $60 billion implied Cursor equity value and SpaceX's 7-day VWAP.
• Closing is targeted for Q3 2026, but kimmonismus's filing note and RayFernando1337's reply both point out the usual blockers still apply, including antitrust review and shareholder approval.
• The acquisition lands just as Cursor is teasing more first-party model ambition, with Yuchenj_UW arguing the company had moved beyond the "wrapper" label and Nick Dobos's Compile teaser pointing to a larger model release within weeks.

You can read Cursor's announcement in its own words, scan Reuters' deal report, and check the filing breakdown for the merger mechanics. The weirdly useful extra detail is that the paperwork names the merger shell, prices the stock exchange off SpaceX's 7-day VWAP, and leaves enough time before closing for Cursor to keep shipping its own roadmap.

Cursor's announcement was short. The filing details are where the story gets concrete.

The merger is structured as a stock-for-stock deal, not a cash exit. According to Rohan Paul's summary of the filing, SpaceX subsidiary X67 Inc. merges into Anysphere, Cursor survives the merger, and becomes a wholly owned SpaceX subsidiary after closing.

That same summary says shareholders receive SpaceX Class A stock based on a $60 billion implied equity value and the 7-trading-day volume-weighted average SpaceX closing price before closing. It also pegs the expected close to Q3 2026, subject to HSR expiration, court or legal clearance, shareholder approval, and standard no-material-damage conditions filing summary.

The immediate engineering read is simple: SpaceX gets a coding-agent distribution channel, while Cursor gets a much larger compute backstop.

Rohan Paul described the combination as SpaceX buying both enterprise reach and a feedback loop, citing Cursor's place inside daily developer workflows and an estimated $2.6 billion in annualized B2B revenue, via a Reuters report. Yuchenj_UW pushed the same idea from the model side, arguing in his post that Cursor had already become capable of post-training its own coding models.

A follow-up from the same thread sketches the division of labor people now expect: xAI or SpaceX keeps pre-training base models, while Cursor handles coding-specific post-training with its own recipe and data. That is still a community read, not an announced product plan, but it lines up with why several replies treated the deal as more than a distribution acquisition.

The price tag looks huge until you compare it with the business people think SpaceX is buying.

The public numbers floating around are not perfectly aligned. Deedy Das called Cursor a $4 billion run-rate business growing 7x year over year, while Rohan Paul's Reuters-based thread cited about $2.6 billion in annualized B2B revenue. Even with that spread, the common point is that Cursor had already become one of the biggest commercial wins in applied AI.

Aaron Levie, CEO of Box, put the sharper version in his post: Cursor proved that a company could win by owning distribution, acting as a model router, deciding when to lean on frontier providers, and only training where that improved the product. That is a much more specific story than "VS Code fork plus Claude wrapper," which is exactly the framing Yuchenj_UW argued the company had outgrown.

The last useful wrinkle is timing: Cursor is not heading into this deal as a finished product.

A Compile teaser cited by Nick Dobos claims Cursor's next model is roughly Opus and GPT-5.5 class in size, trained from scratch rather than on a Kimi base, and used 10 to 20 times more compute than Composer. Separately, mattlam_ said the event also included Cursor Mobile and Origin, which he described as an agent-native git platform.

If those launches land before the merger closes, SpaceX is not just buying the current editor. It is buying into a product line that is still expanding across models, clients, and developer workflow surfaces.