Getty Images is going private. A $533m equity commitment from the Getty family and Koch Equity Development will retire $270m of senior unsecured notes, and the company will reorganise through a prepackaged Chapter 11 limited to certain US entities, expected to complete by year end. The two investors emerge owning 100%. The stated purpose includes addressing Getty's non-operating legal liabilities — which, for a company in active copyright litigation with Stability AI in both the US and the UK, is doing a lot of quiet work in a short phrase.

The restructuring isn't the story. The lender presentation is.

In it, Getty discloses AI Data Licensing Revenue as approximately 2.6% of total revenue in FY2024, 4.4% in FY2025, and 3.4% for the twelve months to Q2 2026. Against total revenue of roughly $978m, that's about $33m. It peaked last year. It is now declining as a share of the business.

This is one of the first primary, filed figures for what training-data licensing is actually worth to a major rights holder — not a press release, not an undisclosed "multi-year strategic partnership," but a number in a document shown to lenders. And it is small.

Why this matters more than the bankruptcy

For two years the prevailing narrative in media, publishing and stock imagery has been that licensing content to AI developers is the replacement revenue line. Traffic collapses, search referrals evaporate, so you sell the corpus instead.

Getty was the archetype of that thesis — a deep, rights-cleared, metadata-rich library of exactly what frontier labs need, run by a team willing to litigate and license at the same time. It executed about as well as the strategy allows. The result was under four points of revenue, trending down, while the same deck lists the Stability AI litigation as a liquidity item and a drag on 2024 EBITDA.

The licensing line and the legal line are now the same conversation. And the legal line is being routed through a bankruptcy court.

Three structural reasons the number is this small

Training data is a one-time purchase, not a subscription. Once a lab has ingested a corpus, it never needs to buy it again. Revenue is front-loaded and lumpy by nature — exactly the shape Getty's three-year trend shows.
There is no scarcity premium on general imagery. Buyers have substitutes: synthetic data, competing archives, open corpora, and the practical reality that enforcement against an already-trained model is slow and expensive. Scarcity sets price, not volume.
Licensing cannibalises the product. Every model trained on stock photography becomes a competitor to stock photography. Getty was selling the input to its own displacement, and $33m a year doesn't fund the transition.

So is data monetisation low-value in the age of AI?

Maybe - but one specific version of the thesis should now be considered dead. Selling a static archive as training data is a low-value, declining, one-off trade. That is evidenced, not argued.

What retains value is data a model cannot substitute away from:

First-party behavioural and transaction data, generated continuously by your own operations
Live operational telemetry tied to a system of record
Compliance-grade data with contractual provenance and audit trails that an enterprise buyer legally requires
Data with a feedback loop, where usage improves the asset

The distinction is between selling a corpus once and operating a data product that cannot be cloned.

What to do this quarter

Reunderwrite any data monetisation business case against 3.4% and falling, not against headline deal announcements.
Separate your data into two buckets: historical archive (treat as one-time, low value) and regenerating proprietary stream (treat as the actual asset).
Price your legal exposure and your licensing revenue in the same model. Getty's example shows them converging, not offsetting.
Test every data asset with one question: if a model trained on this once, would the buyer ever need to come back? If no, it's a sale, not a business.
If you're the buyer, note that licensing leverage has shifted toward you — and that the sellers' legal strategy is increasingly a distressed-balance-sheet strategy.

Sources: Getty Images lender presentation (SEC EX-99.2) · Getty Images press release (SEC EX-99.1)