The end of a startup doesn't necessarily mean the end of its data. As AI training faces limits with public information, a quiet market has emerged where AI labs acquire digital remnants from failed companies. From code repositories to chat logs, defunct ventures are finding a second life as training material, and founders are unexpectedly cashing out.

What Happened

SimpleClosure, a company built to help founders wind down operations, now runs an Asset Hub where code and workspace data can be sold. Since launching data sales in April, the platform has handled over 150 deals and expects to return more than $10 million to startups by the end of 2026. Founders like Keelan Miskell have netted thousands for code bases and workspace data across tools like Google Docs, Notion, Figma, and Slack. The market extends beyond SimpleClosure; rivals like Sunset also assess startup assets by headcount, years in operation, and data volume, comparing operational exhaust to wood shavings—once dismissed, now recognized as valuable. Deals often include code repositories, chat histories, email archives, and document collections, with 47% of SimpleClosure transactions now featuring code or data sales.

Why This Matters

Demand for defunct companies' data surged last year as AI training faces limits with public and synthetic datasets. Real-world archives offer nuances synthetic datasets lack, such as complex conversational patterns and contextual judgment calls. A single Slack history, for instance, can carry more training value than a combined archive of emails and calendars. However, the trade raises privacy flags: Spirit Airlines' pilot union objected when Google secured decades of employee communications during bankruptcy proceedings, and legal experts warn of 'legally radioactive' details tucked into emails and calendars—doctor appointments, school pickups, personal plans. While buyers and suppliers claim to scrub personally identifiable information, the lack of federal consent requirements means worker data can change hands without explicit approval. Proceeds from these windfalls typically flow to senior creditors and investors—not employees, a dynamic law professor Ifeoma Ajunwa describes as 'captured capital.'

Key Takeaways

  • AI labs are actively purchasing defunct startup data to bridge gaps in training datasets.
  • Brokerage platforms like SimpleClosure and Sunset are structuring the sale of code, chats, and documents.
  • Valuations remain uncertain, often based on research theses about useful task extraction rather than standardized metrics.
  • Large-scale deals, such as Google's acquisition of Spirit Airlines' data during bankruptcy, highlight the scale of transactions.
  • Privacy and legal risks persist, with no federal mandate requiring founder or worker consent.
  • Prices may stabilize as the market matures, but the ethical and regulatory landscape remains uncertain.

Conclusion

As AI development pushes forward, the afterlife of startup data illustrates how digital footprints can outlive their companies. Whether this becomes a sustainable revenue stream for failing founders or a privacy minefield depends on how regulators, buyers, and founders navigate the uncharted waters of post-shutdown data commerce.