Cursor, 2026: the sixty-billion-dollar code editor
In 2016, a sixteen-year-old at the Horace Mann School in the Bronx created a programming competition called Halite, which eventually drew ten thousand contestants across three seasons. His name was Michael Truell, and he had six years before he would start building Cursor.
In 2022, Truell and three MIT classmates — Sualeh Asif, Arvid Lunnemark, and Aman Sanger, whom he had met at Horace Mann — founded a company called Anysphere and started building a code editor. Their initial funding was eight million dollars, from a startup fund backed by OpenAI and a handful of angels that included Nat Friedman, the former CEO of GitHub. Friedman had presided over GitHub when Copilot launched; he recognized what the four founders were building.
The founding insight was narrow and decisive: every code editor that existed could read a file. Cursor would read the whole codebase. GitHub Copilot predicted what a developer would type next. Cursor was told what needed to change and then changed it — locating the relevant files, making the edits across multiple locations, running the tests, reading the error output, and iterating, without the developer specifying the path. The user interface was, deliberately, not very different from Visual Studio Code. The AI layer underneath had no precedent.
The revenue numbers that followed were not the kind the developer-tools market had seen before. The company crossed a hundred million dollars in annual recurring revenue in January 2025, five hundred million by June, and a billion by the end of the year — a pace that had taken Salesforce nine years to reach from its own founding. By May 2026, Cursor was generating three billion dollars in annualized revenue. Clients included Nvidia, Adobe, Uber, and Shopify.
The acquisition began as a partnership. In April 2026, SpaceX and Cursor announced a collaboration in which the code editor would gain access to SpaceX’s GPU cluster — by that point the largest privately held AI compute fleet in the world. Two months later, after SpaceX’s IPO, the partnership became a purchase. The price was sixty billion dollars in SpaceX stock: 389.3 million Class A shares. Cursor’s statement at close was precise about the appeal: SpaceX was “building the computing capacity needed to scale intelligence far beyond what exists today,” and a code editor generating three billion dollars in annual revenue needed more compute than most software companies could own.
Truell was twenty-five when the deal closed on 14 August 2026. The transaction was described, credibly, as the largest acquisition of a venture-backed startup in history.
TECO, in 1962, gave MIT programmers a way to write programs; Bell Labs built ed so that Unix could write itself; Emacs added programmability so the editor could become whatever its user needed. Each assumed the programmer knew what they wanted and needed help expressing it. Cursor was the first to assume the programmer might only need to know the direction. SpaceX, apparently, found that distinction worth sixty billion dollars.
Sources
- Cursor (company) — Wikipedia — founding story, revenue milestones ($100M→$3B ARR), SpaceX acquisition details.
- SpaceX Completes $60 Billion Cursor Acquisition — Bloomberg — closing date, stock terms (389.3M Class A shares), GPU compute rationale.
- SpaceX officially closes its Cursor acquisition — TechCrunch — partnership announcement, IPO context, acquisition close.
- Michael Truell — Wikipedia — biographical background, Halite competition, Horace Mann School, MIT education.