Google paid $10 million in a bankruptcy auction for Spirit Airlines’ internal business data, including roughly 100 million emails and 500 million Microsoft Teams chats, to use for training its AI models.
What Google is actually getting
The dataset includes emails, Teams chats, spreadsheets, calendars, marketing materials, HR records, financial databases, and Spirit’s internally developed software, including source code, plugins, and documentation, according to reporting from Tom’s Hardware and Forbes. Google says the data will be cleared of personally identifiable information and customer records by a third party before the handoff.
The competing bid
Google’s $10 million offer beat a rival $7.5 million bid from Mercor, an AI hiring platform, according to the same reporting. A federal judge still needed to sign off on the sale.
Why it matters
Buying a defunct company’s internal communications for AI training, rather than licensing published content or scraping the open web, is a comparatively new category of data acquisition, and the fact that a rival AI company also bid on the same dataset suggests this kind of real, messy, internal-business-operations data has genuine training value that public web text doesn’t fully replicate. The privacy-scrubbing process is doing a lot of work here; whether that scrubbing actually holds up to scrutiny once, or if, any of Google’s resulting products become traceable back to Spirit-specific patterns is worth watching, not something confirmed one way or the other yet.
What to watch: whether the bankruptcy court approves the sale as expected, and how Google’s third-party scrubbing process gets verified.
More Big Tech & Industry Moves coverage, or everything tagged google.
