Google Purchased Defunct Startup Data in August 2026
AI laboratories are acquiring corporate archives to expand training sets as public data sources hit performance ceilings.
Updated on Oct. 13, 2026 in Artificial Intelligence
Live Poll
Do you believe defunct companies should have the right to sell employee data for AI training?
In August 2026, Google paid $10 million for the digital archives of Spirit Airlines, including 100 million emails and 500 million Teams messages. This acquisition is part of a growing trend where AI labs purchase data from failed startups to bolster model training.
Why it matters
AI developers are hitting performance plateaus with existing public and synthetic datasets, making historical corporate archives a high-value commodity. Startups are selling these assets during the wind-down process to recover capital for investors and creditors.
Google acquired 100 million emails, 500 million Teams messages, and 176,000 employee records, comprising a $10 million data asset sale.
The players
A global leader in search, cloud infrastructure, and large language model development that relies on massive-scale data ingestion for model training.
SimpleClosure
A platform specialized in facilitating the wind-down of technology startups and the liquidation of their digital and physical assets.
The details
Data suppliers utilize teams of scientists to scrub personally identifiable information and filter out sensitive human resources records before transferring datasets to buyers. Buyers typically assess the value of these archives based on the original startup's headcount, operational lifespan, and total data volume. No federal law in the United States currently prohibits the sale of employee-generated data by defunct entities.
Timeline
April 2026: SimpleClosure integrated data sales into its startup asset liquidation services.
August 2026: Google completed the purchase of Spirit Airlines data for $10 million.
September 2026: Reports emerged that SpaceX is considering similar startup data acquisitions.
End of 2026: SimpleClosure expects to return over $10 million in total proceeds to startups.
The Tech Race
This acquisition follows the accelerating trend of rising startup shutdowns, which provide a growing supply of historical corporate data. AI labs are now outbidding traditional buyers to secure these unique, non-public archives for model training.
Users of defunct digital services should be aware that their historical internal communications may be sold to tech giants during liquidation proceedings. Currently, no federal regulations prevent these companies from including employee emails or chat logs in such transactions.
The takeaway
The scarcity of high-quality, novel training data is turning the liquidation of failed startups into a primary engine for AI model development. Observers should track upcoming disclosures from firms like SimpleClosure to see if data pricing stabilizes as more assets enter the market.
What happens next
SimpleClosure is scheduled to finalize its year-end distributions to startup creditors and investors, with a goal of returning over $10 million in combined liquidity by December 31, 2026.
Further reading
Learn more about the latest developments in Artificial Intelligence as companies seek new sources of training data.
Source note: This article includes information reported by MIT Technology Review.
Live Poll
Do you believe defunct companies should have the right to sell employee data for AI training?
