Startups shuttering their operations now have a new revenue stream: selling their accumulated data to AI training labs. A company called SimpleClosure brokered a deal this year for crypto entrepreneur Keelan Miskell to sell his defunct startup's codebase and workspace data (emails, chats, documents from Google, Notion, Figma, and Slack) for approximately $40,000 combined.
SimpleClosure added data sales to its services in April through its Asset Hub platform. The company has processed over 150 such deals and projects returning more than $10 million to startups by end of 2026. CEO Dori Yona reports that 47% of SimpleClosure's wind-down deals now include code or data sales, a practice he describes as once "somewhat taboo" but now "increasingly routine." Competitor Sunset runs a similarly growing resale business, with founder Brendan Mahony comparing the data to "particle board" or wood shavings once considered waste but now valuable.
Demand surged last year as AI training labs exhausted public and synthetic datasets. Surge AI, which builds "reinforcement-learning gyms" for client AI agents to practice tasks, buys real startup archives because they capture authentic scenarios, disagreements, and judgment calls that synthetic data misses. The value is unpredictable, however. Pricing varies widely because buyers base valuations on "research theses" estimating how many useful training tasks emerge from a given data volume. No company has published comparisons of real versus synthetic data training results.
High-profile acquisitions demonstrate the market's scale. Google paid $10 million in August to acquire Spirit Airlines' data during bankruptcy, including 100 million emails and 500 million Teams messages. SpaceX's AI unit is reportedly considering similar purchases from struggling startups.
Legal and ethical concerns cloud the transactions. Spirit's flight attendants' union objected to the sale on privacy grounds. Attorney Greg Ewing characterizes such data as "legally radioactive." Suppliers and buyers scrub personally identifiable information and exclude HR records, yet emails and calendars often contain private details like doctor appointments, school pickups, and personal plans. Ifeoma Ajunwa, law professor at Emory University, calls this "captured capital" and argues that "employers should have express consent before selling workers' data." However, no federal law prohibits such sales.
Proceeds typically flow to senior creditors and investors, not employees. Founders also lack visibility into who purchases their data or how it will be used, and have little assurance they receive fair offers. Data prices will likely stabilize in coming months, though an AI market downturn could shift the trajectory.