The FTX estate just turned a $3 billion AI windfall into a $200,000 yard sale, and it tells you everything about how bankruptcy courts value the future.
The Summary
- SpaceX agreed to acquire AI coding startup Cursor at a $60 billion valuation, making FTX's 5% stake worth roughly $3 billion today
- FTX bankruptcy liquidators sold that same stake for $200,000 in 2023, a 15,000x miss in under three years
- This represents one of the largest missed recoveries in crypto bankruptcy history, and a case study in how traditional bankruptcy frameworks fail to price optionality in tech assets
The Signal
When FTX collapsed in November 2022, its estate held a grab bag of crypto tokens, real estate, and venture bets. Among them was a 5% stake in Cursor, an AI coding assistant that was interesting but not yet essential. Bankruptcy trustees had a job to do: convert everything to cash, fast, and get creditors paid. In 2023, they sold that Cursor stake for $200,000.
SpaceX's $60 billion acquisition offer values that same stake at $3 billion today. That's a 15,000x return in under three years, money that would have made FTX creditors nearly whole on their own.
"This retroactively turned a routine bankruptcy asset sale into one of the largest missed recoveries in crypto history."
The math is brutal, but the lesson is structural. Bankruptcy courts operate on a mark-to-market logic built for industrial assets, not venture bets on pre-exponential companies. Trustees are incentivized to liquidate, not hold. Their fiduciary duty is to convert uncertain future value into certain present cash. In 2023, $200,000 for an AI coding tool was defensible. Today it looks like malpractice.
Key dynamics at play:
- Cursor went from "interesting dev tool" to "essential AI agent infrastructure" in 24 months
- SpaceX's entry signals Cursor is now a cornerstone bet in the agent economy buildout
- FTX creditors will get paid in full regardless, but they're leaving billions on the table because the legal system can't price optionality
This isn't just about one bad sale. It's about how legacy institutions are structurally blind to the asymmetric upside in agent-era companies. The same courts that forced FTX to dump Cursor at $200K would have held a stake in ExxonMobil forever, waiting for commodity cycles to turn. They understand oil. They don't understand the compounding value of tools that make developers 10x more productive.
The irony is thick. FTX was a casino dressed as an exchange, run by people who believed rules didn't apply to them. The estate's liquidation has been a model of sober, by-the-book asset recovery. And yet the system's own conservative bias just cost creditors $3 billion in upside. Sam Bankman-Fried made catastrophic decisions. The bankruptcy trustees made prudent ones. Both groups lost billions.
The Implication
If you're holding venture-stage equity in AI infrastructure companies, this is your warning that traditional bankruptcy processes will never value what you're holding correctly. The gap between "what this is worth in liquidation" and "what this becomes in 36 months" is now measured in orders of magnitude, not percentages.
For creditors in future crypto blowups, push for structures that let the estate hold high-conviction tech bets instead of fire-selling them. For builders, remember that the companies winning in AI right now are tools that make other builders faster. Cursor went from $200K to $60B because it became essential infrastructure for the agent economy. Build something people can't work without.