SpaceX exercises its $60B Cursor option four days after IPO, closing the largest AI dev-tools acquisition on record
SpaceX signed a definitive merger agreement on 2026-06-16 to acquire Anysphere, the company behind Cursor, for $60B in stock. The deal exercises the April option tied to a Colossus training arrangement and closes four days after the SpaceX IPO.
SpaceX signed a definitive merger agreement on 2026-06-16 to acquire Anysphere, the company behind Cursor, in an all-stock deal valued at $60 billion. The transaction exercises the option SpaceX disclosed in April, which paired the $60B acquisition price with a $10B alternative break-up fee tied to a collaborative training arrangement on the Colossus supercomputer. Four days after the SpaceX IPO that took the merged aerospace and AI conglomerate public at a $75B valuation, the company spent roughly a 3.4% IPO-valuation dilution to acquire the largest distributed AI coding agent on the market.
What shipped
The deal uses a reverse triangular merger structure with X67 Inc., a wholly owned SpaceX subsidiary, as the merger vehicle. Cursor will survive as a wholly owned SpaceX subsidiary at close. Each share of Anysphere common and preferred stock converts into SpaceX Class A common stock at a ratio set by the volume-weighted average closing price of SpaceX shares over the seven trading days preceding close. Completion is targeted for Q3 2026 subject to regulatory approval. Cursor CEO Michael Truell told CBS News the goal is “building the world’s most useful AI models,” language that frames the agentic-coding stack as a model-development problem first and a distribution problem second.
The April option came with structural commitments that read clearly in retrospect. Cursor was already training Composer 2.5, its in-house frontier coding model, on tens of thousands of GPU-equivalents on xAI’s Colossus by the time SpaceX disclosed the option. The collaborative training arrangement was the gate; the acquisition is the upgrade from “Cursor borrows Colossus capacity” to “Cursor’s roadmap reports to SpaceX.” The $10B break-up fee structure was never about whether SpaceX would buy Cursor, only about which of the two operational endpoints the relationship would land on.
Where this lands in the market
Cursor’s market share among AI coding tools fell from 41% in June 2025 to about 26% in May 2026 per Ramp spending data, a slide that coincided with Claude Code’s rise as the bring-your-own-key default for shops that wanted Anthropic without the marketplace markup and Codex’s June 2026 push into role plugins and the Sites surface. SpaceX is buying a category leader on the back half of its growth curve, but it is also buying the largest distributed install base of any AI coding agent and roughly $4B in annualized revenue, and that mix is where the strategic case actually sits.
The Anthropic dependency is the loudest unresolved variable. Cursor’s defaults have leaned on Claude through most of 2026, and Cursor’s enterprise contracts assume Claude availability. Composer 2.5 was already a first-party hedge, but the merger with a SpaceX division that includes Grok and the SpaceXAI training stack now puts model-tier ownership inside the same corporate parent. The clean read is that Cursor’s first-party model program accelerates and the Anthropic share of inference quietly shrinks. The messier read is that Cursor’s enterprise customers, many of whom standardized on Cursor specifically to consume Claude through a familiar editor, did not sign on for a SpaceX-owned model lane.
Microsoft considered a bid and declined to file one per CNBC, and Cursor previously rebuffed two OpenAI approaches. The signal in the rejections is that Cursor’s board treated GitHub Copilot Desktop and Codex as adjacent products rather than acquirers worth selling to, and waited for an acquirer with a model stack and distribution outside the existing AI editor field. SpaceX fits that profile, but it brings the regulatory exposure the merged xAI brought with it; the DOJ probe of xAI’s unpermitted gas turbines opened the same day (2026-06-16) is the active reminder.
What’s worth watching
- Composer 2.5 timeline. SpaceX pitched IPO investors a $26T AI TAM and an AI division that needed to compete with Anthropic and OpenAI. The first deliverable that proves the merger worked is a Composer 2.5 release that lands competitive on coding benchmarks against Claude Code’s Anthropic backbone and Codex’s GPT-5.5 lineage. Watch for the model release date relative to Q3 close.
- Anthropic contract status. The catalog already covered the export-control episode that suspended Fable 5 and Mythos 5; the additional unresolved question is whether Cursor’s existing Anthropic enterprise contracts survive the merger. If Cursor pulls Claude defaults inside Q4 2026, that is the signal SpaceX is rebuilding the inference lane to its own stack faster than the install base expects.
- Regulatory drag through close. The Q3 2026 target depends on antitrust clearance for a deal that consolidates a category leader under a SpaceX division already drawing DOJ scrutiny on energy infrastructure. The risk is less that the deal breaks than that closing slips into Q4, at which point the $10B break-up fee becomes the operative price rather than the headline number.
The plain reading is that SpaceX bought distribution and revenue, not technology. Cursor’s structural lead was always the editor surface and the install base, and what SpaceX gains is the line from xAI’s Colossus compute through Cursor’s enterprise distribution to a frontier coding model SpaceXAI can ship under its own brand. Whether that lane outruns the Anthropic dependency the install base assumed at purchase is the question the next two quarters will answer.