As shutdowns among tech startups rise, an end-of-life ecosystem has emerged to handle operational wind-downs, with firms now selling digital archives to AI labs. According to MIT Technology Review, companies such as SimpleClosure and Sunset broker the sale of dying startups' code and workspace data—including emails, chats, documents, and Figma files—to foundation AI labs and mature tech companies. SimpleClosure's Asset Hub platform expects to return over $10 million to startups by the end of 2026, with nearly half of its deals now involving data sales. This demand is driven by AI labs hitting the limits of public and synthetic datasets. Companies like Surge AI use these archives to build reinforcement-learning environments where AI agents practice complex tasks. Larger deals are also occurring; Google outbid Mercor to purchase Spirit Airlines' decades of data, including millions of emails and messages, for $10 million during its bankruptcy. Despite the financial recovery for founders and creditors, the practice raises significant legal and privacy questions. Attorneys and law professors note that workspace archives often contain sensitive personal details, and employees typically do not receive proceeds from the sales or provide express consent for their data to be monetized by third parties. While suppliers and buyers attempt to scrub personally identifiable information, the market currently operates in a largely unregulated space.
AI labs are buying data from dead startups to train their models
A growing ecosystem of tech startups is helping failing companies wind down by selling their digital archives to AI labs. As AI training labs face limits on public and synthetic data, they are increasingly buying code and workspace logs from dead startups. However, these sales raise legal and privacy concerns regarding worker data.
Why it matters
This emerging market provides AI labs with new sources of real-world training data as they exhaust public and synthetic options, though it introduces significant legal and privacy challenges surrounding employee data.