Capital

SpaceX Closes Its $60bn Cursor Buy as Its Share Erodes

SpaceX completed its $60 billion all-stock purchase of Cursor-maker Anysphere on August 14, 2026, issuing 391 million Class A shares. Cursor's revenue doubled to 4 billion dollars in ARR over the same year its Ramp-tracked usage share fell from 41 to 26 percent.

CapitalBy Servola Tech Desk2026-08-144 min read

AI-assisted, edited by humans. | Editor-in-Chief: Leon Soliman | Editorial standards

SpaceX Closes Its $60bn Cursor Buy as Its Share Erodes

Key takeaways

SpaceX completed its acquisition of Anysphere, the maker of the AI coding tool Cursor, on August 14, 2026, according to SpaceX's own 8-K filing with the SEC. Cursor's common and preferred shareholders received 389,289,254 shares of SpaceX Class A common stock, plus 1,752,426 more for vested restricted stock units, priced at the seven-day volume-weighted average closing price immediately before closing, implying the previously agreed $60 billion valuation. Track the delivered share count against SpaceX's own share price in the weeks after close to see what Cursor's founders and employees actually received, not just the headline number from June.
The deal closed almost exactly two months after the definitive merger agreement was signed on June 16, 2026, itself the exercise of an option SpaceX secured on April 21, 2026 that let it either acquire Cursor outright for $60 billion or walk away for roughly $10 billion in combined breakup fees and deferred compute-services credit. SpaceX chose the larger number, which only makes sense if Cursor's standalone trajectory looked strong enough in June to justify a 15-times-ARR price tag.
That trajectory has a split personality: Cursor's annualized revenue roughly doubled to about 4 billion dollars over the twelve months to mid-2026, driven mostly by enterprise contracts with more than 50,000 business customers, roughly two-thirds of the Fortune 500. Over the identical period, Cursor's own usage share of corporate AI-coding spend, tracked by the expense-management platform Ramp, fell from 41 percent in June 2025 to 26 percent by May 2026, as developers shifted toward alternatives including Anthropic's Claude Code. Rising revenue and falling usage share are both true at once; the enterprise contracts behind the revenue number were mostly signed before the erosion accelerated, so the number SpaceX priced the deal on is a lagging indicator of Cursor's actual developer standing.
Cursor CEO Michael Truell framed the deal around compute, not distribution: 'SpaceX has exercised their option to acquire Cursor in an all-stock transaction with the goal of building the world's most useful AI models,' tying Cursor's future explicitly to access to SpaceX's Colossus supercomputer rather than to Cursor's existing product lead. That is a tell about which problem the acquisition is meant to solve: SpaceX is betting that more raw compute for model training can win back the developers Ramp's data shows are already leaving, a bet that has not yet been tested in the market Cursor was losing.
The deal lands four days after SpaceX's own June 12, 2026 Nasdaq debut, which raised 86.2 billion dollars including the greenshoe and valued the stock at 192.46 dollars a share by the close of the announcement day, and three months after the May 6, 2026 completion of the SpaceX-xAI merger that valued xAI at roughly 250 billion dollars, a deal struck after all eleven of xAI's original co-founders had already left the company. For any enterprise buyer standing up an AI coding tool policy, Cursor is no longer an independent vendor with its own roadmap incentives; it is a compute-strategy line item inside a company whose AI ambitions now span rockets, foundation models and developer tools at once.

A Two-Month Sprint From Option to Close

The mechanics of this deal moved fast by large-cap M&A standards. SpaceX first secured optionality on April 21, 2026: pay roughly 10 billion dollars combined in breakup fees and deferred computing-services credit to walk away, or exercise a right to acquire Anysphere outright for 60 billion dollars. On June 16, 2026, four days after SpaceX's own Nasdaq listing, SpaceX exercised the larger option and signed a definitive all-stock merger agreement through a wholly owned subsidiary named X67 Inc. Fewer than two months later, on August 14, 2026, SpaceX's own SEC filing confirms the merger closed, with Cursor's shareholders receiving stock priced off a trailing seven-day volume-weighted average rather than a fixed number.

That pricing mechanism means the final share count, just under 391 million SpaceX Class A shares once vested RSUs are included, was not fixed at signing; it moved with SpaceX's own stock price through the summer. A buyer paying in its own newly listed, still-volatile stock is making a bet on two things at once: that the target is worth the price, and that its own shares will hold their value long enough for the target's shareholders to feel fairly paid.

Revenue Up, Usage Share Down, at the Same Time

Cursor's own numbers going into the deal tell two different stories depending on which one a buyer chooses to price. Annualized revenue reached roughly 4 billion dollars by mid-2026, doubling over twelve months, with enterprise contracts covering more than 50,000 businesses and about two-thirds of the Fortune 500. That is the number that supports a 60 billion dollar, roughly 15-times-revenue valuation.

The other number comes from Ramp, the corporate card and expense platform that tracks what its own business customers actually spend money on: Cursor's share of tracked AI-coding tool spend fell from 41 percent in June 2025 to 26 percent by May 2026, a drop of fifteen percentage points in a single year, as spend shifted toward competitors including Anthropic's Claude Code. Enterprise contracts signed months or years earlier do not update in real time when developers start preferring a different tool day to day; the revenue line lags the usage line, which means the 4 billion dollar figure used to justify the price reflects where Cursor's customer relationships were, not necessarily where its product standing is today.

A Compute Bet, Not a Distribution Bet

Cursor CEO Michael Truell's own statement on the deal did not mention market share, competitors, or product roadmap. It centered on compute: the goal, in his words, was 'building the world's most useful AI models,' with Cursor gaining access to SpaceX's Colossus supercomputer as part of the combination. That framing matters because it tells a buyer what SpaceX believes it is fixing. If the problem were distribution, SpaceX would likely have talked about enterprise sales, partnerships, or bundling. Talking about compute instead signals a bet that Cursor's usage-share slide is a model-quality problem, one that more training capacity can reverse.

That bet is unproven. Anthropic's Claude Code, the tool cited as drawing developers away from Cursor, is built by a foundation-model lab with its own compute advantages, not a thinner competitor Cursor can simply out-spend on training. Buyers evaluating Cursor as a vendor now inherit a wait-and-see position: the product they are paying for today is backed by a compute investment whose payoff, if any, will not be visible for at least one more product cycle.

What This Means for Enterprise Buyers of AI Coding Tools

The deal closed four days after SpaceX's own Nasdaq debut on June 12, 2026, an IPO that raised 86.2 billion dollars including the overallotment option and left the stock at 192.46 dollars a share by the close of the day SpaceX announced the Cursor deal. It also closed three months after the May 6, 2026 close of the SpaceX-xAI merger, which valued xAI at roughly 250 billion dollars, notably after all eleven of xAI's original co-founders had already departed the company before the merger completed. Cursor is now the third major AI asset folded into SpaceX inside a single year, alongside a public listing and a foundation-model merger.

For any organization with Cursor seats on its books, the practical shift is one of incentive structure, not features shipped this week: Cursor's product roadmap now competes for attention and capital inside a company whose primary businesses are rockets and satellite internet, evaluated by a leadership team that has just spent 60 billion dollars betting compute can fix a usage-share problem rather than buying the market leader outright. Renewal conversations from here should ask what specifically the SpaceX compute access is expected to change in the product, and on what timeline, rather than treating the acquisition itself as a stability signal.

Frequently asked questions

How much did SpaceX pay for Cursor, and in what form?

SpaceX paid entirely in stock: Cursor's common and preferred shareholders received 389,289,254 shares of SpaceX Class A common stock, plus 1,752,426 additional shares for vested restricted stock units, priced using the seven-day volume-weighted average of SpaceX's stock immediately before the deal closed on August 14, 2026. The share count implied the previously agreed 60 billion dollar valuation set when the option was exercised in June.

Why did Cursor's revenue double while its usage share fell?

Cursor's roughly 4 billion dollars in annualized revenue by mid-2026 comes mostly from enterprise contracts with over 50,000 businesses, agreements that were signed and locked in before usage patterns shifted. Ramp's spend-tracking data shows Cursor's share of corporate AI-coding spend fell from 41 percent in June 2025 to 26 percent by May 2026 as developers moved toward alternatives like Anthropic's Claude Code, a day-to-day usage shift that lags behind multi-year enterprise contracts.

What did Cursor's CEO say the deal was actually for?

Michael Truell framed the acquisition around compute rather than market position, saying the goal was building better AI models with access to SpaceX's Colossus supercomputer. He did not cite market share, distribution, or competitive positioning, which signals SpaceX is betting that more training compute, not more sales reach, is what Cursor needs to compete with rivals like Claude Code.

How does this fit into SpaceX's broader 2026 moves?

The Cursor deal closed four days after SpaceX's own June 12, 2026 Nasdaq debut, which raised 86.2 billion dollars, and three months after the May 6, 2026 completion of the SpaceX-xAI merger that valued xAI at roughly 250 billion dollars. Cursor is the third major AI-related transaction SpaceX has closed within about a year, alongside going public and absorbing a foundation-model company.

A 60 billion dollar price tag reads as confidence from the outside, but the two numbers behind this deal point in opposite directions: revenue that doubled because enterprise contracts take months to sour, and usage share that fell by fifteen points because individual developers change tools in an afternoon. SpaceX priced Cursor on the slower-moving number and is now betting that raw compute, not the product decisions that lost usage share in the first place, will close the gap with Claude Code. Buyers with Cursor seats on the books inherited that bet the moment the deal closed; the honest question for the next renewal cycle is not whether Cursor still has 50,000 enterprise customers, but whether the developers inside those enterprises are still the ones opening it every day.

SpaceXCursorAnyspherexAIM&AAI Coding Tools

Dynatrace Pays $915M to Buy AI Evaluation

Dynatrace is buying AI observability startup Arize for $915 million and telling investors it will cost 175 basis points of margin. Here is why that number matters.

2 min read

Capital

SoftBank Priced DigitalBridge Before Vantage Reset

SoftBank agreed to buy DigitalBridge for a fixed $4 billion in December 2025. Seven months later, Reuters reports DigitalBridge's own Vantage Data Centers stake is exploring a $100 billion IPO. Why the re-rating windfall goes to the buyer, not the sellers who already voted yes.

4 min read

Decisions

The Third Pivot Was the Warning Magic Leap Missed

Magic Leap filed a WARN notice for 193 layoffs effective 1 October 2026, its fourth strategic pivot after $4 billion raised. The lesson is the pattern, not the pivot.