Spirit Airlines has parked its budget flight business in the big airline hangar in the sky, but it will continue to live on in a Google data center. Google has emerged victorious in a bankruptcy auction to purchase a trove of data collected by the now-defunct airline and reportedly plans to use it to help train its AI models.
Google’s winning bid in the auction was priced at $10 million, outbidding Mercor, an AI training data giant. According to a notice published on the US Bankruptcy Court for the Southern District of New York’s docket, as the winning bidder, Google is now the proud owner of more than 100 million company emails, 500 million Microsoft Teams chats, 30 million lines of code, development metadata, software models and algorithms, as well as revenue, aircraft operations and employee productivity data.
Most of the information Google picked up comes from the company’s internal operations. Details like Spirit Airlines’ 97.5 million passenger profiles and more than 50 million records on Spirit loyalty program members were not included in the purchase. So if you used to fly Spirit, rest easy knowing the company didn’t totally sell you out (at least, not in this particular batch). To be sure, Google reportedly said that it will ensure data is “rigorously scrubbed of any personally identifiable information by a third party before receipt,” per Bloomberg.
So what exactly does Google want with all this information? For AI, of course. “We acquired part of an enterprise dataset from Spirit Airlines, which can help improve our products and AI models,” Google told Bloomberg.
Proprietary business data has become a big get for companies working on AI models. It’s become a micro-market for failed startups to sell off their emails, Slack messages, and other communications to AI companies to squeeze a little more cash out of their dead businesses. Earlier this year, the Wall Street Journal reported that Mercor—the same company involved in the bidding war for Spirit’s trove—was offering people money for materials from their previous jobs, hoping to extract industry-specific knowledge to train its AI.
AI models definitely need some help if they’re ever going to step into an executive role. Anthropic experimented by putting Claude in charge of a vending machine, only for it to lose money hand over fist. Andon Labs has been running an experiment that puts different AI models in charge of a cafe, and so far they’ve all managed to lose money at a baffling rate. Whether stuffing a model with data from the now-bankrupt Spirit Airlines will help with that is anyone’s guess, but hey, data is data.