Cursor's $10 Billion Call Option

How SpaceX's Cursor deal reveals that Grok couldn't crack coding, and why a rocket company outbidding Microsoft for a code editor tells you everything about where AI value is consolidating

Author

Justin Wright
April 27, 2026 • Est. Reading Time: 5 minutes

Happy Monday!

On Monday, SpaceX announced a deal to either acquire Cursor, the AI coding startup, for $60 billion by the end of this year or pay $10 billion for what the company called "our work together." The deal came together so fast that it preempted a $2 billion fundraise Cursor was days away from closing at a $50 billion valuation. Prospective investors were caught off guard.

The structure matters more than the headline number. This is not an acquisition, it is essentially a call option. SpaceX is paying $10 billion for the right, but not the obligation, to buy Cursor later. That $10 billion flows to Cursor regardless. If SpaceX exercises the option, it pays an additional $50 billion. If it walks away, Cursor keeps the money and its independence.

The reason Musk is willing to pay the most expensive cover charge in tech history tells a more interesting story than the deal itself.

SpaceX struck a deal giving it the right to acquire Cursor for $60 billion or pay $10 billion for collaboration. The deal pairs Cursor's AI coding product with SpaceX's Colossus supercomputer and is timed to close after SpaceX's June IPO. Microsoft looked at buying Cursor first and passed. The strategic logic: Grok has not cracked the coding market despite xAI's merger with SpaceX, and Cursor is the fastest-growing SaaS company ever at $2 billion in annualized revenue. Musk is buying what he could not build.

TL;DR

Why SpaceX Needs a Code Editor

SpaceX merged with xAI in February in a deal valued at $1.25 trillion. The stated reason was building "orbital data centers." The practical reason was positioning SpaceX as an AI company ahead of a June IPO targeting a $1.75 trillion to $2 trillion valuation. The IPO filing claims a total addressable market of $28.5 trillion, with $22.7 trillion coming from enterprise AI.

There is one problem with that pitch: xAI's flagship product, Grok, has not established itself in the AI category where companies are actually paying real money at scale: coding tools. On paper, Grok 4 scores 75% on SWE-bench, competitive with GPT-5.4 at 74.9% and Claude Opus 4.6 at 74%. But benchmarks do not capture the real gap. Claude powers Cursor, Windsurf, and Claude Code. It dominates the developer tooling ecosystem that engineers actually use daily. Grok has competitive raw scores but no foothold in the workflows where developers spend money.

Cursor is where they spend it. The company hit $500 million in annualized revenue in June 2025, crossed $1 billion by November, and reached $2 billion by February 2026. It is forecasting more than $6 billion by year end. That makes it the fastest-scaling SaaS company ever, outpacing Wiz, Deel, and Ramp. Its 25-year-old CEO Michael Truell built a product so dominant that Microsoft, which owns GitHub Copilot, looked at acquiring Cursor before the SpaceX deal and passed.

Cursor's Growth Trajectory

Milestone

Date

Time from Previous

$500M ARR

June 2025

-

$1B ARR

November 2025

5 months

$2B ARR

February 2026

3 months

$6B+ ARR (projected)

End of 2026

10 months

$29.3B valuation (Series D)

November 2025

-

$50B valuation (pre-SpaceX)

April 2026

5 months

The Call Option Logic

SpaceX is delaying the potential acquisition until after its IPO this summer. Acquiring a $60 billion company before a public listing would require updating confidential financial filings and would further complicate the roadshow. After the IPO, SpaceX can finance the purchase using publicly traded stock rather than cash.

The $10 billion collaboration fee is the price of that delay. It buys SpaceX immediate access to Cursor's engineering team while pairing them with Colossus, SpaceX's supercomputer with the equivalent compute of one million Nvidia H100 chips. The bet: combining Cursor's product and distribution with SpaceX's compute will produce AI coding models that outperform anything Cursor could build with third-party infrastructure.

This is where it gets strategically interesting. Cursor currently relies heavily on Claude as its underlying model. If SpaceX trains competitive coding models on Colossus, Cursor could swap its Anthropic dependence for in-house models. That would give the combined entity vertical integration from compute to model to product, the same stack logic that has made Apple's approach so durable.

The risk is equally clear. Cursor's magic is not just the model, it is the product layer: the context awareness, multi-file reasoning, and workflow integration that makes developers prefer it over Copilot despite Microsoft's distribution advantage. If Musk pushes to replace Claude with Grok-derived models before they are genuinely better, the product could lose the quality edge that drove its growth.

What Microsoft's Pass Tells You

Microsoft looked at buying Cursor before the SpaceX deal and chose not to proceed. Microsoft has GitHub Copilot with 4.7 million paying subscribers, up 75% year over year. Acquiring Cursor would have been a defensive move to consolidate the AI coding market. But the price was apparently too steep for what Microsoft saw as a competitor to something it already owned.

The irony is that Cursor's growth suggests it was pulling developers away from Copilot, not just expanding the market. SpaceX, by contrast, has no coding product at all. Musk is not defending market share; he is buying his way into a market where xAI has failed to gain organic traction. That explains why a rocket company was willing to pay a price that the world's largest software company was not.

What This Means for Practitioners

For developers, the immediate question is whether SpaceX ownership eventually changes the model layer underneath Cursor. If Cursor swaps Claude for xAI models, the product experience could shift meaningfully. During the collaboration phase, expect business as usual.

For AI companies, the deal validates a specific thesis: value is consolidating around products that own the user relationship, not the models underneath. Cursor is worth $60 billion not because it built a frontier model but because it built the product layer that makes frontier models useful to the people willing to pay. Anthropic builds Claude while Cursor collects the revenue.

For anyone watching the broader market, this is the most expensive admission of failure in AI history. Musk merged xAI with SpaceX, built one of the world's largest supercomputers, and launched multiple Grok models. None of it produced a coding product that developers chose to use. He is paying $10 billion for the option to buy the company that did, built by a 25-year-old CEO who understood something Musk's operation missed: in AI coding, the product, and not the model, is the moat.

The Bottom Line

Musk just paid $10 billion for the right to buy a code editor. That sounds absurd until you consider that Cursor generates $2 billion in annualized revenue, doubles every few months, and has become the default tool for professional software development. Grok could not crack this market and Copilot is losing ground to it.

The SpaceX-Cursor deal is not really about coding. It is about where AI value actually lives. Increasingly, the answer is not in the model but in the product that sits between the model and the user. Cursor built that product, now Musk wants to own it. And he is willing to pay $10 billion just to keep the option open.

In motion,
Justin Wright

If a 25-person startup built the most valuable AI coding product on earth using someone else's model, what does that tell you about where the real competitive advantage in AI actually lives, and what does it mean for the companies spending billions to build the models underneath?

Food for Thought

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!