Google’s next big AI model could end up using troves of old data from Spirit Airlines.
Google successfully bid $10 million for troves of corporate data from Spirit Airlines, the budget airline that went under earlier this year, according to Chapter 11 bankruptcy court filings. A spokesperson from Google told Business Insider that this data from Spirit Airlines “can be helpful in improving our products and AI models.”
“We are buying the company’s internal data and custom software, but we are not buying their customer or credit card information,” the spokesperson said.
AI training startup Mercor made a rival bid at $7.5 million.
“Companies are sitting on decades of records that show how real work gets done, and that data is now some of the most valuable material for training and evaluating AI,” a Mercor spokesperson said in a statement. “We partner with leading companies to license their operational data to the labs building the next generation of models. Spirit was that same process applied to a bankruptcy estate.”
A representative for Spirit Airlines declined to comment beyond the court filings.
The move comes as tech companies, having largely exhausted the open internet for training data, race to grab as much data as they can in other forms.
Corporate data is especially valuable because it can help train bots to handle customer complaints more effectively or debug websites. AI training startup Micro1, for example, announced a new program earlier this year where it pays 50 midsize companies $100,000 to $2 million for access to their anonymized data.
The auction for Spirit Airlines’ data was competitive, according to the filings. Google bid $5 million, only for Mercor to counter with $5.2 million. Mercor also said it was willing to pay more — $7 million — if it could obtain the raw data first and then anonymize it. Ultimately, Google bid $10 million and won the deal.
The white-hot race to build the best AI models has led to unconventional tactics for obtaining corporate data.
Meta, for example, announced an initiative to track its workers’ keystrokes and mouse movements so it could improve how well its AI uses computers. Meta suspended the program after an internal data exposure and a strong backlash from employees.
