SpaceX acquires Cursor (Anysphere) for $60 billion in stock

The most striking feature of this deal is that almost none of it is cash. SpaceX is buying Cursor entirely with its own freshly-minted Class A stock, days after a Nasdaq debut that floated only about 4% of its shares ^[1]. That tiny float is the whole trick: when a company sells only a sliver of itself, scarcity alone can push the price up, and SpaceX's stock jumped roughly 50% from its $135 offering price, briefly carrying the company past a $2.9 trillion valuation and ahead of Amazon ^[2]. A richly-valued share is now a currency, and the company is spending it.

The deal's structure quietly compounds this. The exchange ratio — how many SpaceX shares each Cursor owner receives — is set by the volume-weighted average of SpaceX's closing price over the seven trading days before the deal closes ^[3]. In plain terms: the higher SpaceX's stock trades into closing, the fewer shares it has to print to cover the same $60 billion. The incentive to keep the stock elevated through Q3 is baked into the contract.

What makes this remarkable is the contrast with the underlying business. SpaceX lost $4.9 billion on $18.7 billion of revenue in 2025, with Starlink the only consistently profitable part, and the Guardian notes the purchase does not draw on the IPO proceeds at all ^[4]. A company burning billions a year is acquiring a $60 billion company without spending a dollar of operating cash. The reaction split along exactly this fault line: euphoria on X greeted the deal as the largest software acquisition in history, while Reddit's finance and technology communities were openly cynical, framing it as creating paper value out of thin air to buy real companies and reaching for Cisco-era dot-com comparisons.

Strip away the financial spectacle and the strategic logic is clean. After absorbing xAI, SpaceX already had the two heavy ingredients of an AI company: compute, in the form of the Colossus supercluster, and a frontier model effort in Grok ^[3]. What it lacked was a product developers actually open every day. Cursor is that product — used by more than half the Fortune 500 and over a million daily active users ^[11]— and acquiring it hands SpaceX the developer workflow and usage analytics that a raw model cannot generate on its own ^[5][6].

The deeper prize is positional. A code editor like Cursor has historically been neutral ground: it lets developers route requests to Anthropic's Claude, OpenAI's GPT, or Google's models interchangeably. The moment SpaceX owns that surface, the neutral ground stops being neutral. SpaceX has already begun jointly training a model to release inside both Cursor and Grok Build ^[8], and IDC's Arnal Dayaratna argues the real constraint Cursor faced was GPU access — precisely what xAI can now resolve ^[5]. This is why the most clear-headed commentary on developer YouTube framed the deal not as buying an editor but as a land grab for the agentic interface itself: whoever owns the surface where developers work owns distribution for whatever model sits behind it.

The collateral damage points at Anthropic. Cursor has been one of Anthropic's largest Claude customers, so any redirection of that traffic toward Grok hits Anthropic's coding revenue directly ^[7]. Owning the application layer, in other words, is also a way to starve a rival's distribution.

For the CIOs who actually deploy Cursor, the deal trades one set of certainties for a pile of open questions. Coding agents are now used by 64% of the Fortune 500, which makes Cursor's governance posture a board-level concern rather than a developer preference ^[5]. Moor Insights' Jason Andersen warns that enterprise customers will be very concerned about model choice and governance, and specifically whether a SpaceX-owned Cursor can keep pointing at models other than Grok ^[5]. A tool whose entire value proposition was model-agnosticism is now owned by a company that makes one of the models.

Data handling sharpens the worry. InfoWorld's analysts flag zero-data-retention concerns, and note a detail many enterprise buyers miss: Cursor's standard Privacy Mode already stores some code data, and strict zero retention is a legacy setting rather than the default ^[5]. Source code is among the most sensitive assets a company holds, and a change of ownership is exactly the moment procurement teams reopen those contracts. Andersen also points to competitors like AWS's Kiro as the natural fallback if Cursor's neutrality erodes ^[5].

This anxiety was loudest among Cursor's own users. The developer community greeted the news with a mix of rename jokes and genuine unease, and the realist read circulating among them was sober: expect a near-term compute boost, expect leadership churn, and don't count on model-agnosticism surviving beyond about a year. The people most dependent on the product are the ones least sure it will stay the product they chose.

The bull case assumes Cursor is ascendant. The numbers complicate that. CNBC reports Cursor's market share fell from 41% to 26% between 2025 and 2026 as Anthropic's Claude Code became dominant, and TechCrunch has reported the company was struggling to break even ^[1][8]. SpaceX is paying $60 billion — roughly double the $29.3 billion post-money valuation from Cursor's November 2025 Series D ^[9]— for an asset whose competitive position was eroding, not strengthening, at the moment of sale.

The corporate history reads as a warning too. Per AI Business, both OpenAI and Microsoft had previously expressed interest in Cursor and did not close a deal ^[6]. When the two most natural strategic buyers — each with deeper coding-AI experience than SpaceX — pass, a $60 billion all-stock bid from a rocket company invites the question of whether SpaceX is buying conviction or buying at the top. The favorable termination clauses, including a structure CNBC describes as roughly a $1.5 billion fee plus $8.5 billion in compute, suggest both sides wanted optionality more than certainty ^[1].

The optimists are not unanimous, but they are specific. Hargreaves Lansdown's Matt Britzman concedes Cursor lacks OpenAI and Anthropic's scale yet praises its capital-efficient coding models ^[3], and Vital Knowledge's Adam Crisafulli frames the buy as a jolt for an underperforming Grok rather than a trophy ^[10]. Read together, the analysts are not arguing that $60 billion is cheap — they are arguing that compute and distribution might make it look cheap in hindsight. Whether that bet pays off depends entirely on a stock price that, for now, is sustained by a 4% float.