SpaceX pays $60bn in stock for Cursor, six times the price of staying just partners

SpaceX closed its purchase of Cursor on August 14, 2026, paying $60 billion entirely in stock for the AI coding platform built by Anysphere ^[1][2]. The price was set back in April, when the two sides announced a partnership that handed SpaceX a choice: buy Cursor for $60 billion later in the year, or pay $10 billion to formalise the joint work ^[3][4].

It took the option that costs six times as much ^[5]. The mechanism that made that arithmetic tolerable is the one worth studying: SpaceX went public in mid-June, and only after the IPO did it exercise the purchase option, structuring the consideration entirely in stock ^[6][7]. A $10 billion partnership fee is a cash number and would have been paid out of the balance sheet. A $60 billion acquisition, settled in freshly listed shares, is paid by shareholders through dilution. That is the practical difference between the two branches of the April agreement, and it explains why the expensive branch was the one available.

Regulatory clearances ran on schedule, the merger took effect in mid-August, and Cursor now operates as a wholly owned subsidiary ^[8]. From announcement to close was roughly four months, and from listing to close roughly two ^[9].

What is not disclosed matters. The reported terms give a headline value and a form of payment and nothing else: no share count, no exchange ratio, no reference price, no Cursor revenue figure against which $60 billion could be tested ^[16]. Until those appear, the dilution taken by anyone who bought SPCX in June is not calculable from the public record.

The strategic case is compute plus distribution. Cursor's own framing is that access to SpaceX's computing resources, described as the world's largest fleet of GPUs, will let it train more powerful models that are cheaper to run ^[10], with the expected result being better tools at lower prices for customers ^[11]. That is a promise rather than a reported outcome; the only thing offered as early evidence is the release of Grok 4.6 ^[12]. Cursor's side of the trade is that it has moved from simple code completions to agents handling substantial real-world tasks ^[17], and it brings tens of thousands of organisations including a majority of the Fortune 500, a base described as distribution and training data for Grok and coding agents such as Grok Build ^[14]. The stated aim is to close competitive gaps with rival AI coding products by combining product, developer reach and training infrastructure ^[15].

This is the third leg of a pattern. SpaceX integrated xAI earlier in 2026, has prioritised large terrestrial clusters such as Colossus, and is pursuing orbital data centres meant to get around power, land and cooling limits on Earth ^[13], while positioning Cursor as the interface through which that capacity gets applied ^[18].

Three things to watch. First, the S-4 or equivalent disclosure: share count and exchange ratio will convert $60 billion from a press release number into a measurable cost per existing holder. Second, whether Cursor's prices actually fall, since cheaper and better was the case made for the deal ^[11]. Third, the share price itself. Equity is now this company's stated acquisition currency ^[2][6]; if it weakens, the next $60 billion option gets considerably harder to exercise on the same terms.