At 3:58 PM Pacific on Tuesday, SpaceX posted to X. The message was short. It said that SpaceX and Cursor were now "working closely together to create the world's best coding and knowledge work AI." Within an hour, reporters at TechCrunch, Bloomberg, CNBC, and Fortune had filed the rest of the story. Cursor's board had agreed to a structure that, by year-end, gives Musk a choice. Pay ten billion dollars for the joint work. Or pay sixty billion dollars and take the company.
Every other AI company that might have wanted to buy Cursor now has to wait until January to find out whether they still can.
The deal is the climax of a three-month sequence that, in hindsight, reads like a slow-motion takeover. In February, Musk merged SpaceX with xAI in a transaction he valued at $1.25 trillion. In October 2025, Cursor's co-founder and CTO Arvid Lunnemark left the company. In March, Andrew Milich and Jason Ginsberg, two of Cursor's most senior engineering leaders, followed him to xAI. They now report directly to Musk. The week before the announcement, xAI began renting GPU capacity to Cursor. On Tuesday, Musk bought the option on what was left.
What the Deal Actually Says
The terms are unusual. SpaceX is not buying Cursor right now. Instead, it has purchased two things at once.
The first is a joint development commitment. SpaceX and Cursor will combine Cursor's product and its distribution into expert software engineers with Colossus, the xAI supercomputer that Musk has publicly described as having the compute equivalent of one million Nvidia H100 chips. Cursor's team will train its in-house model, Composer, on that hardware. If the collaboration ends there, SpaceX will pay Cursor $10 billion.
The second is a call option. At any point before the end of 2026, SpaceX can cancel the collaboration fee and instead acquire Cursor outright for $60 billion. The option is exclusive. Any competing acquisition offer that arrives during that window is effectively blocked.
Cursor had been in advanced talks to raise $2 billion in a private round that would have valued it at roughly fifty billion dollars, co-led by Andreessen Horowitz and Thrive Capital with Nvidia as a strategic investor. That round, reported less than three weeks ago, is now irrelevant to the question of who will own Cursor. The question has been replaced with a different one: will Musk pull the trigger?
Cursor CEO Michael Truell framed the announcement as a scaling story. In a post to X, he wrote that he was "excited to partner with the SpaceX team to scale up Composer." Cursor's official statement acknowledged the real constraint. "We've wanted to push our training efforts much further, but we've been bottlenecked by compute," the company wrote. The statement added that the new partnership will allow Cursor's team to use xAI's Colossus infrastructure to dramatically scale up the intelligence of its coding models.
How Cursor Became Worth $60 Billion
Cursor's valuation has moved faster than any startup in enterprise software history. The progression:
| Round | Date | Valuation | Lead Investors |
|---|---|---|---|
| Series A | Aug 2024 | $400 million | OpenAI Startup Fund |
| Series B | Jan 2025 | $2.6 billion | Andreessen Horowitz, Thrive Capital |
| Series C | May 2025 | $9 billion | Thrive Capital |
| Series D | Nov 2025 | $29.3 billion | Thrive Capital |
| Series E (in talks) | Apr 2026 | $50 billion | Andreessen Horowitz, Thrive, Nvidia |
| SpaceX option price | This week | $60 billion | SpaceX |
The acceleration was pulled by revenue. Cursor crossed one hundred million dollars in annual recurring revenue in January 2025. It hit five hundred million by June. It cleared one billion in November. By February 2026, that figure had doubled again to $2 billion in ARR. More than half of the Fortune 500 now pays for Cursor seats. The trajectory from zero to two billion dollars in annualized revenue in roughly three years has no precedent in enterprise software. Slack took five years to reach its first billion. Zoom took nine.
A second lever came in November 2025, when Cursor shipped Composer, its first in-house inference model optimized for code generation. Before Composer, every Cursor request routed to a third-party model (primarily Claude or GPT) and every token cost Anysphere, Cursor's parent company, money that flowed out of its gross margin. Composer changed the unit economics. It also gave Cursor something it had never owned before: a model that could, in principle, be trained to beat the frontier labs on the narrow problem of writing software.
Training Composer is what required Colossus. And Colossus is what Musk controls.
Why Musk Wants Cursor and Why He Wants It This Way
The deal structure tells you what SpaceX is actually buying. It is not just Cursor's product. It is a way of shaping the coding-AI market without having to commit $60 billion today.
xAI and Cursor do not currently have proprietary models that match the frontier offerings from Anthropic or OpenAI. Cursor's in-house Composer is fast and tuned for autocomplete, but on complex reasoning tasks it still routes many requests to Claude or GPT. Grok, xAI's model, has struggled to compete on the reasoning benchmarks that matter for code. Neither company can win the coding-agent race on model quality alone.
What they both have is distribution and compute. Cursor sits in the editor window of more than one million paying developers. Colossus, after the xAI merger, houses the single largest unified GPU cluster outside the hyperscalers. Putting them together gives Musk something Anthropic and OpenAI do not yet have: a vertically integrated coding stack where the product, the distribution, the fine-tuning data, and the compute all live inside one company.
The split between collaboration and acquisition lets SpaceX test the thesis cheaply. If the joint training work pushes Composer past Anthropic's and OpenAI's coding models, Musk can exercise the option and absorb Cursor into the xAI group ahead of the SpaceX IPO. If the work disappoints, ten billion dollars is the price of a multi-year lockup on a company that had been fielding acquisition approaches from every frontier lab, including OpenAI, which Cursor's board had already rejected.
For the IPO, the structure is convenient. SpaceX filed confidentially with the SEC on April 1 and is targeting a $1.75 trillion valuation with a raise of roughly seventy-five billion dollars. An active, signed partnership with the fastest-growing enterprise software company in the world is a talking point for the roadshow. A sixty-billion-dollar acquisition that sits on the balance sheet during the IPO window is a less clean story.
The Other Side
The deal has skeptics, and their objections are specific.
The most pointed critique comes from Tim Fernholz at TechCrunch, who observed that "Neither Cursor nor xAI has proprietary models that can match the leading offerings from Anthropic and OpenAI." The partnership's entire thesis depends on that changing. If Composer, trained on Colossus, still cannot beat Claude Sonnet or OpenAI's Codex on code benchmarks a year from now, the $10 billion will have purchased distribution at a price that the market will not support at IPO.
The second objection is about governance. Cursor's founders have so far retained operational control through every round. A sixty-billion-dollar acquisition, however it is structured, collapses that arrangement. Fortune's coverage raised the open question directly: the most interesting thing about the deal is whether the Cursor founders still hold the levers that matter once Musk is paying their bills.
The third objection is from the customer side. A meaningful fraction of Cursor's Fortune 500 users are enterprises with explicit compliance policies against sending code to Musk-affiliated infrastructure. JPMorgan, which has independently disclosed that it is tracking whether its 65,000 engineers use AI tools, is among the buyers that will now have to ask whether Cursor still qualifies. Cursor has not yet clarified how customer data will be segmented under the new partnership.
For context, the previous LDS article on Cursor's funding round Cursor's Valuation Just Doubled in Five Months. The New Round Is Already Oversubscribed at Fifty Billion. laid out the commercial case. The case for a sixty-billion exit two days later rests on whether compute access alone is worth the markup over what Andreessen Horowitz was willing to underwrite on Monday.
What Happens Next
Four things will get decided in the next 90 days.
Whether the Cursor fundraise closes at the leaked $50 billion or gets marked up further by the SpaceX announcement. Whether enterprise customers raise new compliance objections or quietly accept the partnership. Whether Anthropic and OpenAI respond by accelerating their own coding products, since the door to buy Cursor is now closed. And whether Musk commits to the option or leaves it unused, sending a signal about how much of Cursor he actually wants.
The window to buy, under the terms announced Tuesday, runs through the end of 2026. After that, Cursor is either a SpaceX subsidiary or a $60 billion startup that turned down the offer.
The Bottom Line
Musk did not buy Cursor on Tuesday. He paid ten billion dollars to make sure nobody else could. The partnership locks in Colossus as Cursor's training compute, adds two of Cursor's senior engineers as xAI reports, and hands SpaceX an eight-month window to decide whether the full $60 billion is worth spending. Cursor, a company that had been days from closing a fifty-billion private round on its own terms, has agreed to a structure in which its ownership is now optional.
What this confirms is that the frontier of AI competition has moved off the model benchmark and onto the distribution layer. Cursor does not win because Composer is better than Claude. Cursor wins because it is the tool a million developers open every morning. Buying the tool, or merely leasing the right to buy it, is now a move that competes with building a better model.
As Cursor itself put it in the Tuesday statement: "We've been bottlenecked by compute." Every startup in this market has been. Tuesday was the day one of them admitted that the price of being unbottlenecked was the option to be owned.
Sources
- SpaceX is working with Cursor and has an option to buy the startup for $60B (TechCrunch, April 21, 2026)
- SpaceX says it can buy Cursor later this year for $60 billion or pay $10 billion for 'our work together' (CNBC, April 21, 2026)
- SpaceX Has Deal for Right to Acquire Cursor for $60 Billion (Bloomberg, April 21, 2026)
- SpaceX strikes a $60 billion deal for Cursor (Fortune, April 22, 2026)
- SpaceX and Cursor strike partnership that might end in a $60 billion acquisition (Engadget, April 21, 2026)
- SpaceX secures option to buy AI coding startup Cursor for $60B (The Next Web, April 22, 2026)
- SpaceX Obtains Option to Buy Cursor for $60 Billion (Gizmodo, April 21, 2026)
- Why SpaceX just made a $60 billion bet on AI coding ahead of historic IPO (Teslarati, April 21, 2026)