When Spirit Airlines shut down in May 2026, it left behind $8.1 billion in debt and roughly two decades of operational data. On August 19, a federal bankruptcy court is expected to approve a deal that sends most of that data to Google for $10 million.

That number deserves a second look. Not because $10 million is a lot of money for Google, but because of what it reveals about the current moment in enterprise AI.

What Google Actually Bought

The deal includes a remarkable volume of real-world enterprise data:

• Around 100 million internal emails
• 500 million Microsoft Teams messages and collaboration records
• Pricing data from over 7 billion competitor flight records
• 7.5 billion passenger transaction records spanning nearly two decades
• HR, strategy, marketing, and project management files
• More than 30 million lines of code from Spirit’s operational systems

What’s explicitly excluded: the personal profiles of Spirit’s 97.5 million passengers, 52.4 million loyalty members, and 740,000 co-branded cardholders. Before Google receives anything, a third party will scrub all personally identifiable information. The court filing states the deidentified data “cannot be associated with, reasonably used to infer information about, or otherwise linked to, a particular consumer.”

Google beat out Mercor.io, an AI-focused recruitment firm, whose bid came in at $7.5 million. Google’s winning bid: $10 million.

Why This Matters for Business Leaders

The obvious takeaway is that enterprise data has value that most businesses have not yet quantified. Google didn’t pay $10 million for Spirit’s brand, its routes, or its customer relationships. It paid for the raw operational record of two decades of running a business. Email threads. Pricing strategy documents. Scheduling data. The kind of messy, real-world information that makes AI models actually useful in production environments.

This is a signal about where AI training is headed. Public internet data is increasingly scraped, contested legally, and of questionable quality for business-specific tasks. The next frontier is proprietary operational data from real companies running real operations. Airlines, logistics companies, banks, healthcare providers, manufacturers, all of them are sitting on training datasets that AI developers want.

If your business has been running for more than five years and uses enterprise software, you have data that is worth something.

The “De-identified” Question

The privacy protections in this deal will get tested. The deal excludes customer names and loyalty data, and a third party will scrub PII. But the definition of “de-identified” is doing a lot of work here.

The data includes billions of transaction records, pricing decisions, and operational patterns. Researchers have repeatedly demonstrated that large behavioral datasets can be re-identified even after anonymization attempts, particularly when the dataset is large enough and behaviorally rich enough to fingerprint individuals.

Flight attendant unions were vocal critics of this deal. Their concern: 176,000 employee records are part of the package, including what the court filing describes as human resources data. Even scrubbed of names, granular employee data tied to shift patterns, performance metrics, and communications is a different privacy calculus than aggregate pricing data.

For businesses and legal teams, this is a reminder that “de-identified” is a process, not a guarantee. The strength of anonymization depends entirely on what techniques were used and how rich the dataset is.

Precedent: Your Data Has a Liquidation Value

Here is the part that most business leaders are not yet thinking about. When a company goes bankrupt, its data is an asset. Creditors can sell it. Courts can approve that sale to buyers who will use it to train AI systems.

The Spirit Airlines deal is not unique in principle, but the scale and the buyer make it a landmark. This is Alphabet — one of the world’s largest AI developers — paying market rate for enterprise operational data via a bankruptcy court.

This creates real questions for every business:

What data policies govern your data if your company is acquired or restructured? Most vendor agreements and employment contracts were written before this market existed. They may not clearly address what happens to operational data in a liquidation event.

What is your data actually worth? The answer is not zero. A decade of real operational emails, pricing decisions, and workflow patterns represents training signal that AI companies will pay for.

How are you thinking about data as a strategic asset? Companies that understand their data’s structure, lineage, and potential value will be better positioned as AI integration deepens across industries.

What This Means for Business

The Google-Spirit deal is a useful frame for thinking about enterprise AI in 2026. The competition for capable AI is partly a competition for the data that makes those models useful in specific domains. Operational data from real businesses is becoming increasingly valuable precisely because it captures the complexity and messiness of actual work, not synthetic examples of it.

For EDNA’s community of data professionals and business leaders, this moment is a prompt to revisit two things: how your organization thinks about its data as a strategic asset, and whether your data governance policies are built for a world where that data has real market value.

The businesses that figure this out early will be better positioned to negotiate, protect, and where appropriate, monetize the data they are already generating every day.

---

If you want to explore how AI can help your organization get more value from its existing data, Enterprise DNA’s Omni Advisory service works with leadership teams to build practical AI data strategies. Our Learn platform also has courses on data governance and business intelligence for teams at every level.