Micro1 Bids $12.5M to Outbid Google for Spirit Airlines' Data

Micro1 has filed a $12.5 million counterbid for the internal corporate records of Spirit Airlines, the budget carrier that ceased operations in May 2026 and is now liquidating its assets in bankruptcy court. The offer arrived after Spirit's bankruptcy auction had already closed with a winning $10 million bid from Google, beating a competing $7.5 million bid from Mercor. Micro1's higher, later offer is a procedural long shot: courts rarely reopen a closed auction to accept a late bid over a deal that's already been agreed to. A bankruptcy judge is scheduled to review the sale on September 9, 2026.

Mercor's losing bid wasn't simply outbid on price. Court filings describe Mercor offering the same $10 million Google ultimately paid, on the condition that Mercor would handle de-identification itself using its own tools. Spirit's advisers went with Google instead because Google agreed to pay a neutral third party to scrub the data before Google ever touched it, a structure the advisers judged safer.

Neither company is bidding on Spirit's planes or ground equipment. The core of the archive is the airline's internal communications, built up by 17,000 employees over several decades: roughly 500 million Microsoft Teams messages, 100 million emails, 17 million OneDrive files, 20.5 million SharePoint items, more than 30 million recorded customer service calls, and roughly 97.5 million customer profiles, loyalty records, and service chats. Alongside the messages sits structured operational data: flight movements, crew pairings, booking curves, competitor-pricing observations, maintenance tasks, and fuel logs. The pairing of the two is what makes the archive distinct. A model trained on it can see a flight delay register in the operational systems, then read the Teams messages where employees diagnosed and worked around it, a link that's difficult to construct from scraped or synthetic data. The fierce bidding over this kind of material reflects a broader shift: AI labs are increasingly willing to pay a premium for proprietary, real-world operational data rather than relying on scraped or synthetic sources.

Beyond the messages, the sale also includes Spirit's software itself: an estimated 516 repositories totaling about 30 million lines of code across the airline's web, mobile, and backend systems, along with the commit histories, pull requests, and bug reports that go with them. For AI labs training coding models, that's a rarer asset than the messages archive. Most public code used for training comes from open-source repositories; a large enterprise codebase with years of real bug reports and review history behind it is not something that shows up on GitHub.

Google's agreed deal excludes customer chat sessions, loyalty records, and call recordings entirely. Micro1's counterbid reaches further: it proposes acquiring the roughly 97.5 million customer profiles, loyalty records, and service chats that Google's deal left out, while explicitly excluding disciplinary files, investigatory materials, and anything tied to union collective bargaining. Whether Micro1 would be allowed to acquire unredacted consumer records is a separate question; sales of that kind draw heavy regulatory scrutiny in bankruptcy, and Micro1 has said its interest is in de-identified business operations rather than identifying individual customers. Micro1 has also proposed that an ombudsman selected by Spirit's own advisers, rather than by Micro1 itself, oversee how the data is handled.

Micro1 founder Ali Ansari framed the bid around a concept he calls realism, in a LinkedIn post laying out the company's reasoning: the idea that the more a model's training data reflects the genuine messiness of real work, the closer it gets to the conditions it faces once deployed. "The limit is the real world itself," he wrote. "But the real world is also dynamic. It grows with a company and shifts as operations evolve." Ansari says Micro1 has made binding commitments tied to the deal: the company will prohibit re-associating the data with named individuals, won't profile any former Spirit employee, and is paying for an independent data ombudsman to review how the data is used. He also committed the company to spending at least $1 million providing former Spirit employees preferential access to paid human data roles on Micro1's platform, along with free access to its AI training and reskilling curriculum, calling that $1 million "just the minimum."

Unions Are Fighting the Sale

Beyond the procedural hurdle of reopening a closed auction, Micro1 faces organized opposition from the people whose records are in the archive. Former Spirit employees and their unions, including the Association of Flight Attendants-CWA, have taken legal action to block the sale entirely, arguing that decades of workplace communications shouldn't be monetized to train AI models regardless of who buys them or what privacy commitments come attached. The objection carries more weight given that many former employees still haven't been paid their accrued vacation time by the bankrupt airline.