Musk Exercised the Call Option

How SpaceX turned a $10 billion Cursor partnership into a $60 billion acquisition four days after the largest IPO in history, and why a rocket company is now building the full AI developer stack

Author

Justin Wright
June 22, 2026 • Est. Reading Time: 6 minutes

Happy Monday!

Four days after the largest IPO in history, SpaceX made the largest acquisition of a venture-backed startup in history.(source: CNBC)

Several weeks ago in this newsletter, we wrote about SpaceX paying $10 billion for what we called "a call option on the fastest-growing software company in history." The structure gave SpaceX the right, but not the obligation, to acquire Cursor for $60 billion. We wrote that the deal was designed to close after SpaceX's June IPO, when publicly traded stock could replace cash.

On Tuesday, four days after raising $75 billion in the largest IPO in financial history, SpaceX exercised the option. Cursor is being acquired for $60 billion in an all-stock transaction, the largest acquisition of a venture-backed startup on record. SpaceX shares surged 14% on the announcement, briefly pushing its market cap past $2.9 trillion and making it the fourth most valuable company in the United States.

But the Cursor that SpaceX is buying is not the Cursor it optioned in April. And that gap is the real story.

SpaceX exercised its $60 billion call option on Cursor four days after the largest IPO in history. The deal came with a twist: on the same day, Cursor announced Origin, an agent-first code hosting platform and direct GitHub competitor. SpaceX is now building a full vertical stack from compute to code hosting. But Cursor's market share dropped from 41% to roughly 26% in the past year as Claude Code captured about half the AI coding market. The question is whether SpaceX bought Cursor for the editor or for what comes after it.

TL;DR

The Option We Called

In April, SpaceX structured the Cursor deal as a call option: $10 billion for a partnership that gave SpaceX the right to acquire Cursor for $60 billion later in the year. We wrote at the time that the structure was designed to delay the acquisition until after the IPO, when SpaceX could pay in stock rather than cash.

That is exactly what happened. SpaceX listed on Nasdaq on June 12 at $135 per share, raising $75 billion at a $1.77 trillion valuation. Four days later, it exercised the option. Fortune noted that SpaceX's surging stock "paid for the $60 billion Cursor acquisition in just a few hours of trading." The company's shares had gained enough value in a single morning of post-IPO trading to cover the entire purchase price.

The four Cursor co-founders, Michael Truell, Sualeh Asif, Aman Sanger, and Arvid Lunnemark, will each be worth an estimated $2.7 billion when the deal closes in Q3.

The Cursor That Changed

Here is what shifted between April and June: when SpaceX optioned Cursor, the company had roughly $2 billion in annualized revenue and was projecting $6 billion by year end. Six weeks later, ARR is approximately $2.6 billion. Four months to add $600 million is still strong by any normal standard. But Cursor was doubling every three to five months through 2025. The growth rate is decelerating.

The reason is Claude Code. According to Ramp spending data, Cursor's market share in AI coding tools dropped from 41% in June 2025 to roughly 26% by May 2026. Claude Code now commands about half the market. Anthropic, Cursor's core AI supplier, launched a competing product and took the majority of the market Cursor helped create.

That competitive squeeze is what pushed Cursor into SpaceX's arms. Building proprietary models to reduce Anthropic dependence requires compute that a startup cannot easily afford. SpaceX has Colossus, with 330,000 or more Nvidia GPUs. The two companies have been jointly training a coding model on Colossus for months, and that model is expected to ship inside both Cursor and Grok Build in the near term.

We flagged this exact scenario in the April newsletter: "If SpaceX trains competitive coding models on Colossus, Cursor could swap its Anthropic dependence for in-house models." That swap is now underway.

Cursor's Market Position: Then and Now

Metric

April 2026 (Option)

June 2026 (Acquisition)

ARR

~$2B

~$2.6B

Market share

~35% (declining)

~26%

Claude Code market share

Growing

~50%

Fortune 500 penetration

Expanding

64%

Growth trajectory

Doubling every 3-5 months

Decelerating

Model dependency

Claude (Anthropic)

Transitioning to joint SpaceX model

The Full Stack Play

The acquisition alone would be a significant story, but what makes this week different is what Cursor announced on the same day.

At its inaugural Compile conference in San Francisco on June 16, Cursor unveiled Origin: a code hosting platform and direct GitHub competitor. Built on technology from Cursor's December 2025 acquisition of Graphite, Origin is designed from the ground up for AI agents. The demo showed 22.6 commits per second inside a single repository, with an AI-powered conflict resolution engine that handles the parallel branch merges AI agents generate constantly.

Origin ships this fall and the waitlist opened on Tuesday.

This is the play that transforms the acquisition from "Musk buys a code editor" into something structurally different. SpaceX is now building a full vertical developer infrastructure stack:

  • Compute: Colossus, 330,000+ GPUs, already leased to Anthropic and Google for $75 billion in contracted revenue.

  • Models: jointly trained coding model shipping in Cursor and Grok Build.

  • Editor: Cursor, deployed in 64% of the Fortune 500.

  • Code hosting: Origin, an agent-first GitHub competitor.

From silicon to repository, the full vertical. No other company in AI has this stack.

Why Origin Matters More Than Cursor

We previously noted that GitHub saw 275 million code commits per week by mid-2026, on pace for 14 billion for the year, up from one billion in all of 2025. Most of that growth is AI-generated code.

GitHub was built for human developers committing code a few times a day. Origin is built for AI agents committing code dozens of times per second. If the future of software development is AI agents writing, testing, and merging code autonomously, the repository layer needs to be rebuilt for that reality. Cursor is betting it can own that layer before GitHub adapts.

The strategic logic runs deeper than performance. If you control where code lives, you control the data that trains the next generation of coding models. Cursor's editor generates interaction data about how developers work. Origin would generate data about how code evolves. Combined with Colossus compute, that is a self-reinforcing loop: better data trains better models, which produce better code, which generates more data.

What This Means for Practitioners

For developers using Cursor, the acquisition creates both opportunity and risk. The jointly trained SpaceX model could close the gap with Claude Code. But the transition away from Anthropic's models will involve a quality delta that users will feel. Watch the first post-acquisition model release closely.

For engineering leaders, Origin is the announcement that warrants attention. If your organization is evaluating AI coding infrastructure, the decision is no longer just which editor to use. It is whether your code hosting platform is built for a world where agents generate the majority of commits. GitHub is not going away, but the competitive pressure to modernize just intensified.

For anyone watching the industry, the arc from Newsletter #15 to today is worth studying. In six weeks, SpaceX went from call option to IPO to acquisition to full-stack developer infrastructure company. The company that could not build a competitive coding model bought the product instead, and is now positioning to own every layer of the developer workflow from compute to code storage.

The Bottom Line

Eight weeks ago, we called the SpaceX-Cursor deal a call option. This week, Musk exercised it. But the acquisition is not the most important thing that happened on June 16, Origin is.

A rocket company is now building the infrastructure layer for AI-native software development: compute, models, editor, and code hosting. Whether that stack can compete with Claude Code's 50% market share depends on whether SpaceX can train models that match Anthropic's quality. The product was always the moat. Now the question is whether the full vertical stack becomes one also.

In motion,
Justin Wright

If the company that controls where AI-generated code is stored also controls the compute that trains the models that generate that code, have we just described the most valuable feedback loop in software?

Food for Thought

Quick Hits

  • Altman, Amodei, and Hassabis appeared together at the G7 summit in France, the first time all three rival AI lab CEOs met with world leaders at once. Amodei and Hassabis called for a US-led AI coalition. No binding commitments emerged. (CNBC)

  • EU AI Act transparency obligations take effect August 2, 2026. Chatbots must disclose they are AI, deepfakes need machine-readable watermarks, and non-compliance penalties reach 7% of global annual revenue. (Axis Intelligence)

  • Anthropic's Project Glasswing partners found over 10,000 high- or critical-severity software vulnerabilities in its first month using Claude Mythos. Bug-finding rates increased by more than 10x. (Anthropic)

  • Goldman Sachs projects $7.6 trillion in cumulative AI infrastructure capex from 2026 to 2031. Compute accounts for $5.1 trillion of that total. (Goldman Sachs)

If you haven’t listened to my podcast Mostly Humans: An AI and business podcast for everyone yet, new episodes drop every week!

Episodes can be found below - please like, subscribe, and comment!