Dead Startups' Data Becomes Fuel for AI Labs

When Keelan Miskell shut down his crypto startup, he expected little return. Instead, he sold its code base for roughly $25,000 and its workspace data in Google, Notion, Figma and Slack for a pending amount likely around $15,000. The buyer? An AI lab hungry for real-world training data.

Miskell's sale was brokered by SimpleClosure, a firm that helps dying startups wind down. SimpleClosure added data sales to its services in April through its Asset Hub platform. The platform facilitates the sale of code and workspace data, including emails, chats and documents. Buyers include foundation AI labs, vendors that build simulated workspaces for AI training and mature tech companies developing their own models.

A Growing Market for Dead Startups' Data

SimpleClosure is not alone. Rival Sunset also runs a rapidly growing data resale business. Founder and CEO Brendan Mahony says Sunset values a dying client's assets by head count, years in operation, volume of data and how the team used its tools. He compares operational exhaust data to wood shavings that everyone thought had no value but actually has a ton.

Demand for defunct companies' data emerged suddenly last year as AI training labs hit the limits of public and synthetic datasets. Surge AI, for example, buys data to build reinforcement-learning gyms where clients' AI agents practice tasks like debugging complex IT architecture or drafting appellate briefs. Nick Heiner, who leads Surge AI's work on these environments, says he seeks data archives reflecting a range of situations, from disagreements to tricky judgment calls. Heiner explains that one company's Slack history, if it captures conversations with variety and depth, can be worth more than another's Slack, email and calendars combined.

Legal and Ethical Concerns

Legally, these sales could be problematic. Spirit Airlines' flight attendants' union objected to the sale of the carrier's data to Google on privacy grounds. Greg Ewing, an attorney who advises AI startups, calls such data legally radioactive. Suppliers employ data scientists to scrub personally identifiable information and exclude HR records, and buyers say they scrub too. Ewing points out that private details lurk in emails and calendars, such as doctor appointments, school pickup arrangements and date-night plans. He describes a Wild West of AI training that is only starting to give way to proper licensing agreements and vetted use.

There is also the question of who profits. Proceeds typically flow to senior creditors and investors, not employees. Ifeoma Ajunwa, a law professor at Emory University and author of The Quantified Worker, sees this as captured capital. She argues that employers should have express consent before selling workers' data, but notes there is no federal law banning such sales.

Why This Matters

The rise of data resale from failed startups signals a fundamental shift in how AI models are trained. As public and synthetic datasets become insufficient, AI labs will increasingly turn to unconventional sources. This could accelerate AI development but also spark regulatory battles over privacy and consent.

For founders, the opportunity to recoup some investment is appealing, but the lack of transparency about data buyers and usage remains a concern. Employees, whose personal communications are often included, have little say. The market's growth may pressure lawmakers to address gaps in data privacy laws.

Key Numbers

These figures underscore the scale of this emerging market. SimpleClosure's Asset Hub has already facilitated over 150 deals, with nearly half involving data sales. The projected $10 million return to startups by 2026 highlights the potential value locked in defunct companies' digital remains.

As AI labs continue to seek real-world data, the market for dead startups' data is likely to expand. However, without clear regulations, the ethical and legal questions will persist. Stakeholders from founders to employees to regulators will need to navigate this new frontier carefully.