BREAKING: SpaceX just offered $60 billion for Cursor. The FTX ghost makes it stranger
A $200,000 check. Four MIT kids. The fastest-growing software company ever. Elon Musk locking up the asset OpenAI discovered first, six days before taking Sam Altman to trial.
In April 2022, Alameda Research wrote a $200,000 check into a four-person startup nobody had heard of.
The startup was building a CAD tool for mechanical engineers. The check bought roughly 5% of the company.
One year later, the FTX bankruptcy estate sold that stake for $200,000. Cost basis. Zero gain.
On April 21, 2026, SpaceX announced it had the right to acquire that same company for $60 billion.
The undiluted Alameda position, had it survived, would be worth approximately $3 billion today.
This is the story of Cursor.:
This issue is brought to you by DigitalOcean:
Deploy sold out. The livestream is still open and worth your Monday morning
The in-person event in SF filled up fast. But the content is the real reason to show up. Deploy is built around production inference, not roadmaps or keynotes. Real teams showing what is actually running at scale today.
▫️ Real customer stories on business-critical tradeoffs from Character, Workato, VAST Data, Arcee, and vLLM
▫️ NVIDIA’s Kari Briski on the future of agentic AI
▫️ First look at DigitalOcean’s next-gen inference products
April 28. Free to watch.
Four MIT kids and the fastest revenue curve in software history
Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger graduated from MIT in 2022 and started building immediately. Their path to Cursor:
A CAD tool for mechanical engineers
A brief detour through text-to-image
An AI-native code editor, shipped publicly in March 2023
The key decision was building as a full fork of VS Code rather than an extension on top of it. Extensions are constrained by what the API exposes. A full fork lets you rebuild the UI entirely. Features competitors could not ship, Cursor shipped.
What followed was a revenue curve with no real analog in software history:
Milestone Date $1M ARR October 2023 $100M ARR January 2025 $500M ARR June 2025 $1B ARR November 2025 $2B ARR February 2026 $6B ARR guided End of 2026
The $1M to $100M sprint took 12 months. Wiz took 18. Deel took 20. Ramp took 24.
The team has roughly 150-200 employees. At $2B ARR, revenue per employee approaches $10 million, among the highest ratios in software history.
A few other numbers worth knowing:
7 million+ monthly free users
1 million+ paying individual developers
50,000+ paying teams
~70% of Fortune 1000 as enterprise customers
Enterprise revenue went from 25% of the mix in late 2024 to 60% by February 2026
“My favorite enterprise AI service.”
Jensen Huang, CEO of Nvidia
In Stack Overflow’s 2025 Developer Survey across 49,000+ respondents, Cursor registered 18% usage in its first year of inclusion. The fastest IDE adoption in software history. Claude Code registered 10%.
The funding ladder
Round Date Amount Valuation Lead Pre-seed Apr 2022 $400K ~$4M Neo, Alameda, Heroic Seed Oct 2023 $8M — OpenAI Startup Fund Series A Aug 2024 ~$60M $400M a16z + Thrive Series B Dec 2024 $100M $2.6B Thrive Capital Series C Jun 2025 $900M $9.9B Thrive Capital Series D Nov 2025 $2.3B $29.3B Accel + Coatue Series E (in talks) Apr 2026 ~$2B ~$50B pre a16z + Thrive
Two things to notice in that table.
First: OpenAI was the first institutional investor. They tried to acquire Cursor twice. Got rebuffed both times. Settled for Windsurf at $3 billion.
Second: Thrive Capital appears in four consecutive rounds. More on that below.
Musk is offering $60 billion for the company OpenAI discovered first. Seven and a half times the Windsurf price. Six days before jury selection in Musk v. Altman.
What real conviction looks like: Thrive Capital
Most investors got one shot at Cursor. Thrive took three.
Series B at $2.6B valuation. Led the round.
Series C at $9.9B valuation. Led again.
Series D at $29.3B valuation. Co-led.
Three consecutive rounds. Increasing check sizes. Increasing conviction across 24 months.
If the SpaceX deal closes at $60 billion:
The Series B position is approaching a 19.3x MOIC
The Series C position is at roughly 5.6x
The Series D, closed six months ago, is already a clean 2x
Blended across all three rounds: 3.6x MOIC on ~$826 million invested
Total gains: over $2 billion
This is what differentiated investing looks like in practice. A fund that built its position across three years, increased it at each step, and is now sitting on one of the defining venture outcomes of the AI era.
The interesting question is what this means for Thrive’s new $10 billion fund. Deploying that scale in venture is genuinely hard. Cursor is the proof that concentrated conviction can still produce extraordinary returns. Whether the math works at $10 billion is a separate question entirely.
Quick note from this week’s sponsor:
Deploy sold out. The livestream is still open and worth your Monday morning.
The in-person event in SF filled up fast. Deploy is built around production inference, not roadmaps or keynotes. Real teams showing what is actually running at scale today.
▫️ Real customer stories from Character, Workato, VAST Data, Arcee, and vLLM
▫️ NVIDIA’s Kari Briski on the future of agentic AI
April 28. Free to watch.
The SpaceX deal: what it actually is
The April 21 announcement is a call option, not a closed acquisition.
Here is the structure:
SpaceX has the right to acquire Cursor for $60 billion later in 2026
If SpaceX walks away, it pays Cursor $10 billion for joint work on Composer
Cursor granted SpaceX 12 months of acquisition exclusivity
Michael Truell’s response on X was deliberate in its understatement:
“Excited to partner with the SpaceX team to scale up Composer. A meaningful step on our path to build the best place to code with AI.”
Cursor’s official blog post made no mention of the acquisition option or the $10 billion alternative. It framed the deal purely as a compute partnership. That editorial choice signals something about how seriously they are taking the independence narrative.
The timing is pointed. Jury selection in Musk v. Altman begins April 27. Six days after the announcement. Musk is locking up the asset OpenAI found first, at 20x the price of OpenAI’s comparable acquisition, in the same week he takes Altman to trial.
Why this deal makes sense for both sides
For SpaceX:
SpaceX filed confidentially for IPO on April 1, targeting roughly $1.75 trillion. The problem: xAI lost $6.4 billion in 2025. Cursor brings $2B in real ARR and 70% of the Fortune 1000 into the IPO story. Commercial ballast for a vehicle built around more speculative hardware bets.
xAI also publicly admitted Grok “is currently behind in coding” and was “not built right first time around.” In March 2026, xAI had already hired two senior Cursor product-engineering leads, both reporting directly to Musk. The April announcement was visible from miles away.
For Cursor:
The Windsurf saga showed the existential risk of frontier-lab model dependence. When OpenAI tried to acquire Windsurf, Anthropic cut off Claude API access immediately. Cursor’s core models are Claude, GPT, and Gemini, all supplied by companies now competing with Cursor directly via Claude Code, Codex, and Jules.
The SpaceX option gives Cursor three things:
A $10 billion floor even if the deal breaks
Guaranteed compute for Composer training at a scale unavailable from frontier labs
Leverage against the API cutoff risk that destroyed Windsurf’s independence
Cursor is also running a $2 billion round at $50 billion pre-money in parallel. You do that when you have options, not when you have already decided.
The Windsurf saga: why this matters
April 2025: OpenAI agrees to acquire Windsurf for $3 billion.
During exclusivity: A Microsoft IP dispute stalls the deal. Anthropic cuts off Claude API access.
July 11, 2025: The exclusivity expires.
Within hours: Google announces a $2.4 billion reverse-acquihire of the Windsurf founders into DeepMind.
Three days later: Cognition acquires the remaining Windsurf company, IP, brand, $82M ARR, and 350 enterprise customers, for roughly $250 million.
The whole episode proved one thing: any coding tool dependent on frontier-lab APIs is permanently exposed to the founders of those labs using API access as a weapon. Cursor saw it clearly. Composer, their proprietary inference model, was the strategic response.
The FTX ghost
The Alameda Research investment came through Clifton Bay Investments LLC, the FTX Group’s for-profit venture arm. A $200,000 check in April 2022. Roughly 5% of the company at a ~$4 million post-money valuation.
When FTX collapsed in November 2022, this became an asset of the bankruptcy estate under John Ray.
In April 2023, Ray sold it for $200,000. Cost basis.
The buyer has never been publicly identified.
The counterfactual at each subsequent valuation:
~$450 million at the May 2025 $9 billion valuation
~$1.47 billion at the November 2025 Series D
~$3 billion at SpaceX’s $60 billion option price
These are undiluted estimates. Real dilution would reduce them materially.
The numbers circulating on X after the SpaceX announcement are almost impossible to reconcile with the story the world told about FTX:
“So you’re telling me SBF did Anthropic, Cursor, Solana and Robinhood? His NAV would’ve been about $52-86 billion today. 9.8-16.2x TVPI.” @ JasonrShuman
The Anthropic position sits alongside Cursor as the defining monument to how fast the AI era outran the bankruptcy timeline. Ray sold the Anthropic stake for approximately $1.3 billion across 2024. At Anthropic’s February 2026 valuation of $380 billion, that position would be worth roughly $30 billion today.
Bankman-Fried has made this argument from prison. The math does not vindicate him. But it does complicate the story in ways that are genuinely uncomfortable to sit with.
Anysphere has never publicly acknowledged the Alameda backing in any press release, funding announcement, or blog post.
What happens next
Two scenarios. One tells the rest of the AI market something significant.
If SpaceX exercises at $60 billion:
Cursor becomes Musk infrastructure
Anthropic and OpenAI accelerate their own IDE plays
Fortune 500 companies with data sovereignty concerns start re-evaluating
The Musk empire consolidation arc completes: X into xAI, xAI into SpaceX, Cursor into SpaceX, then IPO
If the option lapses:
Cursor pockets $10 billion in compute resources
Closes its $50 billion independent round
Becomes a serious IPO candidate, potentially the first standalone AI-native software company to reach public markets at scale
The company’s behavior right now reads like the second scenario. A $2 billion parallel fundraising process is not how you act when you have already decided to sell.
The FTX angle gives this story its emotional core. A $200,000 check from a company synonymous with fraud bought what would have been one of the most valuable venture positions of the decade. The estate sold it at cost to move fast. Three years later, Elon Musk is offering $60 billion for the same company, six days before taking Sam Altman to trial.
The bill for that haste, if SpaceX exercises, reads at roughly $3 billion undiluted.
For the complete breakdown of the AI coding landscape in 2026, including every tool, benchmark, and workflow: The complete guide to AI coding in 2026
For the infrastructure layer Cursor is now plugging into: Anthropic just solved the hardest part of building AI agents
The Cursor valuation story tells you everything about where enterprise software budgets are heading. When AI coding tools can 10x developer productivity, the ROI is undeniable and the pricing power reflects it. At SourceMind AI, we help ops and IT leaders apply the same evaluation lens to AI tools in their domain: finance, procurement, and operations. Developer tools are just the first wave.
The most interesting part of this story is that the “acquisition” may be less important than the option structure. A normal acquisition transfers ownership. This kind of deal transfers leverage before ownership. SpaceX gets exclusivity, strategic proximity, and a way to keep Cursor out of OpenAI/Anthropic/Google’s hands. Cursor gets compute, a valuation anchor, and a credible escape route from frontier-lab dependency.