Google wins bankruptcy auction for Spirit Airlines' internal data with a $10 million bid

When a company dies, its internal correspondence becomes an asset. This week that asset had a buyer and a price: Google, at $10 million.

What Google bought

Google won a bankruptcy auction for the internal data of Spirit Airlines with a $10 million bid, beating $7.5 million from AI data startup Mercor. Forbes, TechSpot and Tom’s Hardware reported the result on August 18.

The package is large and unusually detailed for a data sale. It includes roughly 100 million employee emails, 500 million Microsoft Teams messages, 17 million OneDrive files, more than 20 million SharePoint items, 516 code repositories holding about 30 million lines of custom software, and billions of flight pricing records.

Google says the enterprise dataset should help improve its products and AI models. A third party will strip personally identifiable information before the transfer, with Google paying for the anonymization. The airline’s 97.5 million passenger profiles and 50.2 million loyalty records are excluded from the sale. A federal bankruptcy judge still has to approve the deal.

Why a dead airline’s Teams chats are worth $10 million

The value is in the mundane part. Hundreds of millions of messages between employees coordinating real operations across email, chat and documents are a record of how work actually happens inside a large organization, which is close to the behavior AI agents are being trained to imitate. Public web text does not contain much of that; internal corporate correspondence does, and it almost never reaches the market intact.

A bankruptcy auction is one of the few situations where it can, and the correspondence of Spirit’s former staff turned out to have two bidders. That Mercor, a company whose business is assembling training data, was the underbidder is a signal of what kind of asset this is understood to be.

What remains open

Two things are unsettled. The first is the judge’s approval, which is required before any transfer. The second is what the anonymization actually removes. Stripping names and identifiers from 100 million emails is a well-defined task; removing everything that makes a message traceable to a specific person is a harder one, and the announced process names a third party rather than describing a standard.

For employees anywhere, the broader point is simple. Work messages are usually assumed to be too boring to matter. Their boring quality is exactly what makes them valuable as training material, and the moment an employer’s assets go up for sale, the inbox is among them.