Elon Musk built xAI to compete with OpenAI and Anthropic. All 11 co-founders left. Now SpaceX is buying Cursor for $60 billion in stock instead.
The deal was announced Tuesday, four trading days after SpaceX’s record $75B IPO. The price: $60 billion, paid entirely in SpaceX shares. The target: Cursor, the AI coding tool most developers I know either use or wish they could afford at the higher tiers.
Expected close: Q3 2026.
The story isn’t really the deal. The story is that SpaceX spent the last four months telegraphing exactly this, and the reason is the xAI experiment that didn’t work.
The deal in three numbers
Three numbers frame the scale.
$60 billion. The acquisition price, all-stock. Cursor investors get SpaceX shares, not cash. Reuters reported the deal includes a $10 billion standard termination fee.
There’s also a separate $4 billion fee if the deal collapses on antitrust or regulatory grounds specifically.
That split is the tell. SpaceX is hedging for regulatory review, and the structure of the fee tells you they expect to.
$2.6 billion. Cursor’s annualized B2B revenue as of late May 2026, per Reuters. (Wikipedia’s Cursor page, citing Bloomberg, had $3B; Reuters is more recent and more conservative, so I’m using that.)
At a $60B price tag, SpaceX is paying roughly 23× ARR. For context, OpenAI’s last private round priced it at about 35× ARR. Cursor is “cheaper” than OpenAI, on a revenue multiple basis.
$2.53 trillion. SpaceX’s market cap as of Tuesday pre-market, per Reuters. The stock hit $211.27 in pre-market trading, up 56% from the $135 IPO price four days earlier. SpaceX added roughly $247B in market cap in a single morning.
Why now? The xAI disaster, in order
You can’t understand the deal without the timeline.
February 2, 2026: SpaceX closes its merger with xAI. The combined entity carries a $1.25 trillion valuation, per Reuters at the time.
By the end of March 2026: every single one of the 11 xAI co-founders has left. All 11. Per TechCrunch’s reporting, Musk publicly admits xAI “was not built right the first time around” and is being rebuilt “from the foundations up.”
That’s a polite way of saying the product didn’t work and the team fractured. Grok called itself “MechaHitler” in 2025. xAI allowed non-consensual deepfakes of women and children in early 2026. California sent a cease-and-desist. SpaceX flagged all of it as a business risk in its IPO filings.
April 21, 2026: SpaceX announces an option. It can buy Cursor for $60 billion in stock, or pay a $10 billion break-up fee for a paid partnership. The structure looks like an “I’ll know it when I see it” bet.
CNBC covered the original April disclosure, along with the FT and the NYT.
June 12, 2026: SpaceX goes public. $75 billion raised. Largest IPO in history. Musk becomes the world’s first trillionaire on paper.
June 16, 2026: SpaceX exercises the option.
The pattern is clear. xAI couldn’t ship a coding tool. SpaceX bought one. “Plan A was xAI. Plan B is a $60B check.”
Cursor’s 4-year ride
Cursor is a wild story on its own terms.
Founded in 2022 as Anysphere by four MIT undergrads. First money in: $8M from the OpenAI Startup Fund in 2023. Graduated from OpenAI’s startup accelerator in 2024. A coding tool, seeded by the AI lab it now competes with, about to be bought by a rival lab’s biggest backer.
The raise cadence is fast. June 2025: $900M Series C. Late 2025: another $2.3B. Pre-deal valuation: ~$29 billion. SpaceX pays $60B, roughly 2× the last private mark.
The team is small. ~300 people. They ship their own models (Composer 2.5, shipped in May 2026, RL-trained on a Kimi base). They have an AI code-review product called Bugbot. They bought a code-review tool called Graphite in December 2025.
Reuters reports Cursor was on the verge of closing a $2B round at a $50B valuation from a16z, Thrive, and Nvidia. One source told TechCrunch that round “wasn’t going to be enough to help it break even.”
Cursor’s CEO had publicly cited lack of compute as the company’s biggest constraint. SpaceX has compute. Lots of it. Pre-IPO, SpaceX signed $1.25B/month and $920M/month compute deals with Anthropic and Google, both with 90-day termination clauses.
So: Cursor needed compute. SpaceX had it. Cursor’s planned round was undersized. SpaceX had post-IPO stock up 56%. The deal writes itself once you have those facts.
What it means for Cursor users
Honest answer up front: too early to say on most things.
What we know: Cursor’s pricing today is $20/mo for Pro, $40/mo for Pro+, with an Ultra tier above that. Cursor 2.0 launched in October 2025 with parallel agents using git worktrees. Composer 2.5 shipped in May 2026. The product cadence hasn’t slowed.
What we don’t know, and won’t until SpaceX says more:
- Will pricing change? Plausible. SpaceX is paying $60B and will want a return.
- Will third-party model integrations (Anthropic, OpenAI, Google) survive? This is the real question. Cursor’s competitor set includes vibe-coding tools built directly on top of Claude and GPT. If SpaceX decides Cursor should run on SpaceX compute and SpaceX-preferred models, third-party access could be cut.
- Will the product cadence hold? Unknown. A new owner usually means a 6-month pause for “strategic review.”
- What happens to Cursor’s data, code, and customer contracts? Standard acquisition stuff. Unclear until the deal closes.
The compute angle is the most concrete. Cursor’s biggest constraint was compute. SpaceX has $2.4T of TAM in AI infrastructure per the S-1. The Cursor acquisition gives SpaceX a flagship product to run on that infrastructure. That’s a bet, not a guarantee.
The MANGOS era starts paying out
This is the first major post-IPO move from the new AI public-market cohort.
I covered the AI IPO wave last week. The acronym: MANGOS. Meta (or Microsoft), Anthropic, Nvidia, Google, OpenAI, SpaceX. The new FAANG. Each is racing to public markets, and each is using the public-market narrative to justify a different kind of move.
SpaceX’s S-1 told investors a $28T total addressable market story. $26T of that is AI ($2.4T infrastructure, $22.7T enterprise applications). The space business, the rockets, Starlink, is real, but the public-market growth thesis is AI. Tuesday’s $60B acquisition is the proof point.
Notice the parallel: OpenAI bought Ona last week. I covered that too. Two AI coding tool acquisitions, one week apart, both by trillion-dollar labs. The thesis is the same on both sides: own the developer workflow end-to-end.
The regulatory question
Don’t sleep on this one.
Reuters flagged a $4B antitrust-specific termination fee, half the standard $10B. That’s not a number you put in a deal unless you expect to use it.
SpaceX is a US government contractor (Starlink military contracts, launch services). CFIUS may want a look. The FTC is already aggressive on AI consolidation. A $60B deal in AI coding is exactly the kind of thing that gets a second-request review.
This is one of the open questions, not a fait accompli.
Bottom line
SpaceX is now an AI company that also goes to space.
For Cursor users: nothing changes today, but the questions worth asking are pricing, model access, product cadence, and the compute pipeline. None of those have answers yet. Don’t panic. Don’t pre-cancel.
For anyone tracking the AI IPO wave: this is the first major deal of the new era. It won’t be the last. MANGOS is real, and they’re all cash-flush from IPO proceeds. Watch the next two quarters.
And for the record: Plan A was xAI. Plan B is a $60B acquisition. Plan C is whatever SpaceX buys next.
