AI Data Licensing Deals Tracker: Reddit, Spirit Airlines, Shutterstock & More

This tracker lists AI data licensing deals confirmed by a party's own announcement, securities filing or court docket. Disclosed prices are rare: Reddit's $203 million in aggregate contracts, Informa's $10 million-plus Microsoft fee and Wiley's AI revenue are the exceptions. The only operational-data deal with a public price, Google's $10 million Spirit Airlines bid, awaits court approval.

How we built this tracker

Three rules, applied to every row:

1. Primary sources only. A row needs a release from one of the parties, an SEC or exchange filing, or a court docket summary. News reports appear only as context in the text.
2. Values only when disclosed. If neither party published a price, the table says "Not disclosed," even when a figure circulates widely. Reported figures we left out are listed further down, with the reason.
3. Verified on Oct 29, 2026. We opened every source link on that date. Deals we could not confirm from a primary source are omitted, not guessed.

We are a paid referral partner of Mode, one of the programs listed below, and earn a fee if a company we introduce licenses data through it; see our disclosure. That relationship does not affect which deals appear here.

Which content and platform licensing deals are confirmed?

These are deals where a publisher, platform or content library licensed material to an AI developer. They are the best public evidence of how AI data licenses are structured, even though they involve published content rather than internal company records.

Which operational and workplace data deals are confirmed?

This is the market most relevant to ordinary companies: internal email, chat, documents, tickets and operational systems licensed to train workplace AI agents. Here, public records consist of one court-supervised auction, program terms and infrastructure deals.

Program rows are standing offers, not completed transactions. We include them because their published terms are the closest thing to a public price list for company data. For a side-by-side of buyers, see AI data brokers compared; for Mode specifically, see our BizBuyData review.

What does the Spirit Airlines auction show?

It is the only public price we found for an operating company's internal records, and the fight over it previews the issues every seller faces.

What was for sale. The asset schedule covers "all emails, OneDrive, SharePoint, and Teams data within the native Microsoft 365 environment." The flight attendants' union counted roughly 100 million emails across 80,000 accounts, 500 million Teams items, 175,658 employee records dating from 1986 and 5 million crew pairings (Stretto). Google told Business Insider: "We are buying the company's internal data and custom software, but we are not buying their customer or credit card information."

How the bidding went. Google opened at $5 million. Mercor offered more, up to $10 million, if it could use its own de-identification tools instead of a third party; the estate chose Google's $10 million bid with independent de-identification, citing "non-economic factors" that "may be outcome-determinative." micro1 later filed notice of a $12.5 million all-cash bid.

What changed under pressure. Before the ombudsman's report, Google agreed to drop passenger record, refund, onboard purchase, timecard and WiFi sales datasets, with no change to the price (Stretto). The ombudsman said privacy risks to consumers had been "eliminated or mitigated," but her review did not cover employees (Business Insider).

Where it stands. Unions, a software vendor and an engine maker objected. In an Oct 8 letter, 121 members of Congress warned that documents stripped of identifiers "can potentially be reassembled." The hearing was moved from Oct 14 to Nov 6, 2026 (Travel Weekly). Google has bid; it has not paid.

The lesson for a smaller company is about scope, not price. Each objection targeted a category that was in scope by default: employee records, safety reports, vendor-owned data. A $10 million bid for an airline's entire Microsoft 365 estate is also a useful sanity check on per-company figures you see in ads. Our logistics page covers the operational detail.

What patterns show up across the deals?

Prices are mostly private. Of the 15 content and platform rows, only five disclose money. Two are single-deal figures (Wiley's $23 million project and Informa's $10 million-plus Microsoft fee); the other three are aggregates across several licensees (Reddit's contracts, Informa's 2024 total and Wiley's fiscal 2026 total). For operational data, Business Insider reported that executives it interviewed "wouldn't share the exact amounts their companies received" (BI). Any site quoting precise per-employee prices for company data is extrapolating.

Structures repeat. Three patterns appear in the primary documents:

• Continuous access plus periodic transfers. Reddit delivers "through continuous access to our data API as well as quarterly transfers of Reddit data" (S-1).
• Upfront fee plus recurring payments. Informa's Microsoft deal pairs a "$10m+" initial fee with payments over three years and "limits on verbatim text extracts" (Informa).
• One-time project. Wiley booked a single $23 million "content rights project" in one quarter.

Operational programs so far look like the third pattern, a one-time export priced by scope, though Mode's FAQ notes the agreement can define exclusivity. We walk through these clauses in how AI data licensing works.

Uses are broadening. Early deals focused on training. Later ones, such as Condé Nast and NYT–Amazon, also cover display of summaries inside AI products. Workplace data is wanted for a third use: training agents on how multi-step work gets done, which is why buyers ask for email, chat and tickets rather than published content (Business Insider).

OpenAI is the most frequent licensee in the record. It appears in eight of the 15 content rows. Google, Microsoft and Amazon each appear in at least one. In the operational market, the counterparties are intermediaries (Mode, micro1, Mercor) and their downstream buyers are generally not named.

Demand is growing, by the sellers' own account. Wiley's AI revenue rose 23% in fiscal 2026. micro1 plans to spend $1 billion on company data over the next year, per Business Insider, and the same outlet cited a Gartner projection that spending on AI data will reach about $6.5 billion in 2027 (BI).

How did the market move from content to company data?

The record above falls into three phases.

2023: archives and image libraries. The first wave licensed large, already-published collections. Shutterstock's six-year agreement and the AP archive deal, both announced in July 2023, gave OpenAI training material whose rights holders were easy to identify.

2024: platforms and publishers, at scale. Reddit's IPO filing put a hard number on the market: $203.0 million in contracts signed in a single month. Google and OpenAI both took Reddit Data API access within three months. Publishers followed (the FT, News Corp, Vox Media, Condé Nast), and academic publishers reported AI revenue in their results: Wiley's $23 million project, Informa's $10 million-plus Microsoft fee and $75 million-plus in 2024. Several of these deals added display rights, so licensed content could appear inside AI products, not just train them.

2026: internal company records. As AI developers turned to agents that do office work, demand shifted toward material that is not on the public web: email, chat, tickets and project histories. That data is spread across millions of companies, so intermediaries stepped in. micro1, Mercor and Mode publish programs aimed at ordinary businesses; marketplaces such as Protege raised capital; and the Spirit Airlines bankruptcy put an operating company's Microsoft 365 estate up for a court-supervised auction. Mode's release describes the gap these programs fill: the examples labs need "typically remain inside organizations, rather than on the public internet" (release).

The shift changes who carries the risk. A publisher licensing its articles owns the rights and knows what is in them. A company licensing its Slack history is licensing content written by employees, clients and vendors, which is why the Spirit objections came from unions, a software vendor and an engine maker rather than from the airline.

What should you watch next?

• The Nov 6, 2026 Spirit hearing. Approval, conditions or denial will be an early judicial signal on how employee and vendor data must be handled in a sale of workplace records. micro1's $12.5 million notice also raises the question of whether the auction result holds.
• Published program terms. Mode changed its eligibility thresholds and timeline within two weeks of launch, so buyers' ranges, minimums and payment terms may keep moving; we re-check them on every update.
• Disclosed revenue. Public licensors disclose AI licensing money in filings and results, as Reddit's S-1, Wiley and Informa did. Those numbers are the most reliable indicator of whether this market is growing or plateauing.
• Regulatory attention. The 121-member congressional letter on Spirit and the ombudsman's comment that "technology has far outpaced the law" (Stretto) suggest de-identification standards for workplace data could tighten.

Which widely repeated deal claims did we leave out?

If a number you have seen elsewhere is missing here, this table is likely why.

How should you use this tracker if you're thinking of licensing your company's data?

Use it for three things:

1. Calibrate expectations. Published program ranges ($100K–$5M from Mode, $150K to over $2M from micro1) are estimates. No completed small-company deal has a public price.
2. Borrow structure. The clauses in public deals, such as term length, exclusivity, recurring payments and use limits, are the clauses you should negotiate. See our agreement terms guide.
3. Anticipate objections. The Spirit fight shows that employee data, third-party and vendor data, and regulated records draw challenges. Our safety checklist covers what to exclude, and our valuation guide covers what remains worth licensing.

This page is information, not legal or financial advice; we are not lawyers or a CPA firm. Talk to your counsel before you license anything.

Update log. Oct 29, 2026: tracker published; all rows verified. Next update after the Nov 6, 2026 Spirit Airlines sale hearing.