Spirit Airlines collapsed and auctioned its assets, drawing headlines when Google offered $10M for its operational data — 100 million emails, 500 million Teams messages, and 30 million lines of code. Mercor, an AI hiring platform, bid $7.5M. The data represents day-to-day decisions and conversations, not customer or invoice records. Airline staff have petitioned against the sale over privacy concerns.

Why did Google want this data? Traditional structured data is easy to analyze, but Gen AI now extracts structure from massive unstructured datasets. More critically, this data enables reinforcement learning for complex operational decisions — like whether to fly through a storm or divert — which physical simulations cannot capture. Decades of expert conversations reveal how the airline navigated the pandemic, demand spikes, and supply chain crises.

Valuing operational data is difficult because it lacks direct market value. The author, who helps PE firms with commercial due diligence, outlines techniques to estimate worth, focusing on operational data's unique role in training AI agents for high-stakes, judgment-heavy environments.