Your Company Data Is the Most Undervalued Asset in Your Business
Spirit Airlines is shut down. Google still agreed to pay $10 million for its data.[1]
That is the headline, and it should rearrange how you think about the CRM, billing, and usage history sitting in your own systems. In Bad Data = Bad Results, we argued that a model is only as good as the records it was trained on. This is the other side of that coin. Once the records are true enough to learn from, they are not exhaust. They are the asset.
Google agreed to pay $10 million for a shutdown airline's operating history
Spirit halted operations in May 2026 and has been selling remaining assets in bankruptcy. CNN's Chris Isidore noted it was the first significant U.S. airline in 25 years forced to halt operations entirely rather than be sold to another airline.[1] Planes, slots, and real estate are the obvious inventory. The data sale is the tell.
Google won the bankruptcy auction and agreed to pay $10 million for part of Spirit's enterprise dataset. A Google spokesperson told CNN: "We acquired part of an enterprise dataset from Spirit Airlines, which can be helpful in improving our products and AI models." The same spokesperson confirmed Google will not receive personal information as part of the purchase.[1] Backup bidder Mercor.io bid about $7.5 million.[1]
As of August 28, 2026, this is an agreed $10 million deal, not a closed sale. After the Association of Flight Attendants-CWA objected, a U.S. bankruptcy court delayed the approval hearing to September 9, 2026.[2]
Reporting on what sits in the dataset has some tension, and that matters for how you read the story. CNN described emails and internal communications, spreadsheets, and transactions with the public, including bookings and frequent flyer information, plus HR records.[1] Google has said it will not receive personal information.[1] Spirit has said the records would be de-identified, with no customer PII.[2] Skift pointed at finance, operations, revenue management, pricing models, booking curves, and flight behavior.[3] We are not going to pretend this is a sale of every passenger's personal file. The useful fact for operators is narrower, and it is the one that should stick: a shutdown airline's operational and enterprise data drew a $10 million bid from Google.
Even Google, with more public-web data than anyone, is willing to pay a shutdown entity to get a company's internal history.
Rivals can copy a model architecture. They cannot copy your ten years of win/loss, churn, and usage.
You cannot buy someone else's operating history
In the AI era, the generic layer is getting cheap. Anyone can prompt a public model. Anyone can buy a scoring widget. Architecture is not the moat. Your books are.
A competitor can hire the same vendors you hire. They cannot replay your last decade of which leads became customers, which customers expanded, which ones left, what they were doing in the product the month before they left, and what your reps actually did about it. That sequence is unique to your company. It is also the only sequence a commercial model can learn from if you want scores that match your market, your sales motion, and your churn pattern.
Public-web AI is trained on language, pages, and whatever else leaked onto the internet. It is good at sounding fluent. It is not good at telling your sales team which inbound to call at 9 a.m. on Tuesday, because it has never seen your win/loss table. It has never seen your billing file. It has never seen that your "enterprise" segment churns for a different reason than your mid-market book.
Proprietary operational data is the opposite. It is narrow, dated, and specific. That is the point. Lead source, stage history, contract value, usage, tickets, collection status, tenure. Those fields are how the business actually ran. A model fitted to them is a model of your operation, not a model of the internet.
This is why a shutdown airline's internal records can clear eight figures at auction while a mid-market company still treats the same class of data as a storage bill. Google cannot scrape your Salesforce. Your rival cannot download your billing history. You already have the thing they would have to pay for.
Most companies treat this like exhaust
Walk through a normal week. Marketing dumps leads into the CRM. Sales works the ones that feel hot. CS lives in the accounts that are loud. Finance exports invoices for the board deck. Product looks at usage in a different tool. Nobody is trying to be wasteful. The data is a byproduct of doing the job.
Then someone buys an AI tool trained on someone else's world and wonders why the scores feel generic. Of course they do. The generic model never saw your operation. Your operation never got fitted.
The Spirit auction is a loud version of a quiet fact. Internal business data (how a company priced, booked, operated, forecasted, and communicated) has become something Google will pay real money for.[1] You do not need to be an airline. You do not need 20 years of history. You need the trail you already leave: CRM, billing, product usage, support, and the outcomes attached to them.
That trail is sitting there whether you use it or not. If you do not, it is just log files. If you do, it is the only input that can produce lead scores, churn lists, and forecasts that match your business instead of a demo.
The asset is already on your servers
Owning the data is the starting position, not the trophy. The last post in this series was about making the records honest enough to learn from. The next one, Using Your Data, is about putting that history to work on the problems that actually move revenue: churn, lead optimization, forecasting, and customer targeting.
You do not need Google's check to prove the point. You need to stop treating your operating history like exhaust, and start treating it like the one asset a competitor cannot copy.
If you want models fitted to the CRM, billing, and usage data you already have, talk to Gamify Data. We start with your history, not a new lake, and we build scores for the operating decisions your team already makes.
Sources
- Chris Isidore, "Google is buying all of Spirit Airlines' data to feed its AI models," CNN Business, August 18, 2026. https://www.cnn.com/2026/08/18/business/google-spirit-airlines-data
- Natalia Bueno Rebolledo, "US court delays hearing on Google's purchase of Spirit Airlines data as union objects," Reuters, August 18, 2026. https://www.reuters.com/legal/litigation/us-court-delays-hearing-googles-purchase-spirit-airlines-data-union-objects-2026-08-19/
- "Google Scoops Up Spirit's Data in Bankruptcy Sale to Train AI," Skift, August 17, 2026. https://skift.com/2026/08/17/google-scoops-up-spirits-data-in-bankruptcy-sale-to-train-ai/