Google pays $10M for Spirit Airlines' business data for AI training

This deal reflects a fundamental shift in how AI companies source training material. For years, large language models like Google's and OpenAI's relied on vast amounts of text harvested freely from the internet—everything from books to web pages to academic papers. However, those companies have now exhausted most of that accessible public data, creating a bottleneck: without fresh data, the pace of AI model improvement slows. The $10 million acquisition of Spirit Airlines' operational data signals that proprietary business datasets—emails, chat logs, code, and process documentation—are now valuable commodities in a new marketplace. This trend emerged visibly in April, when reporting revealed companies like Mercor offering cash to professionals for materials from their former employers, seeking industry-specific knowledge for AI training. The competitive bidding for Spirit's data shows that multiple AI firms recognize this constraint and are willing to pay significantly to solve it. At the same time, the public commitments around data scrubbing and exclusion of passenger records suggest growing awareness of privacy risks—a tension that may shape regulation and acquisition terms in this emerging market.