Getty Images Is Heading for a Prepackaged Chapter 11. Interest Was Already Outrunning Operating Income, and the AI Pressure Comes Through Search

On 12 October 2026, Getty Images Holdings announced a restructuring deal that would hand 100% of the company to the Getty family and affiliates of Koch Equity Development, in exchange for up to $533 million of new equity and the retirement of $270 million of senior unsecured notes. Existing common shareholders are not expected to receive anything. The company said the process will use a prepackaged Chapter 11 limited to certain US entities, with a filing expected within weeks and completion by year end. The easy headline is that artificial intelligence killed the stock-photo business. The filings tell a more specific story, in which a debt load that was already too heavy met a revenue line under pressure from AI answering searches directly, and in which a hyped licensing deal with OpenAI could not close the gap.

The capital structure came first

As of 30 June 2026, Getty reported about $1.47 billion of face-value debt against $51.6 million of cash. According to PPC Land's reading of the company's filings, the stack was $1.17 billion of senior secured notes, $510.6 million of term loans (mostly euro-denominated), $270 million of senior unsecured notes and $120 million drawn on a $150 million revolver. Moody's cut Getty's corporate family rating from Caa1 to Caa3 on 2 September, and S&P moved to CCC. The company's 10 August filing warned of substantial doubt about its ability to continue as a going concern. On 31 August Getty elected to use 30-day grace periods on 1 September interest payments on its 9.75% notes due 2027 and 14% notes due 2028. The NYSE suspended trading on 29 September after shares settled at 12 cents on 28 September, down about 93% in twelve months, and the stock now trades over the counter.

The sharpest single number is interest cover. In the second quarter, Getty's operating income was $32.4 million and its interest expense was $57.3 million. Dividing the first by the second gives about 0.57, our arithmetic: for every dollar of interest the business generated 57 cents of operating profit. A year earlier interest expense was $36.6 million, so it rose by roughly 57%. Octus reported that Getty also burned $123 million of cash in the quarter, mainly on warrant litigation payments and merger-financing interest, and ended it with the $51.6 million noted above. A company in that position cannot fund its own debt service from operations, whatever its revenue trend.

Much of the pressure was self-created by the failed Shutterstock deal. Getty and Shutterstock agreed a merger in January 2025 valued at about $3.7 billion. The US Department of Justice cleared it on 23 February 2026, but the UK Competition and Markets Authority required the divestiture of Shutterstock's editorial business in its 15 May report. Getty's board rejected that condition on 30 June and sent termination notice on 7 July. Getty's chief executive, Craig Peters, said the company had spent more than $100 million in professional fees over 18 months, according to Octus. This matters for the AI argument: part of the interest bill and the cash drain came from financing a deal that never closed.

Where AI actually hits the revenue line

Second-quarter revenue was $229.1 million, down 2.5% (down 4.1% in constant currency). That is a decline, but not a collapse, and the split inside it is revealing. Creative revenue, the stock-image side where generative tools compete most directly, was $127.4 million, down 2.6%. Editorial revenue, which sells news, sports and entertainment photography that an image generator cannot fabricate, was $96.5 million, up 9.2%. Revenue in the 'other' category fell from $15.7 million to $5.2 million. If generative image tools were the main driver, one would expect the creative line to be falling far faster than 2.6%.

The customer metrics are weaker. Purchasing customers over the trailing twelve months fell 10.0% to 636,000, active annual subscribers fell 25.2% to 240,000, and annual subscriber revenue retention dropped to 88.4% from 93.4%. The mechanism Getty itself points to runs through search. Peters has said iStock's search-engine referral traffic has declined as search engines began answering queries directly, which cuts the flow of low-intent buyers who found images through a query and became subscribers. Getty also ended iStock's free-trial acquisition programme in June 2025, which complicates a clean read of the subscriber fall: some of the decline is a deliberate change in how customers are acquired. Separately, Peters attributed a 13% drop in revenue from advertising agencies to secular pressure, buyer consolidation and in-house production, including with generative tools. That figure is a decline in Getty's revenue from agencies, not a measure of total agency spending, and the three causes are not apportioned.

This is a distinction the common narrative misses. AI image generators substitute for a stock photo when a buyer would otherwise have licensed one. AI-powered search answers substitute for the click that used to bring that buyer to iStock in the first place. The second effect hits volume at the top of the funnel and is visible in subscriber counts, while the first shows up in agency budgets. The filings support both effects, but the sources do not quantify how much of the 25.2% subscriber decline each one explains.

The OpenAI deal that could not help

On 21 June 2026 Getty announced a multi-year display agreement putting licensed images into ChatGPT search and discovery. Shares rose by as much as 145% on the news, then reversed fully. The reversal is the instructive part. Licensing revenue from an AI partner, if it scales, adds to the top line over years; Getty's interest bill was due that quarter. Nothing in the reported terms suggests the deal was large enough to change the interest-cover arithmetic, and no payment figures were disclosed. The deal also arrived nine days before the board terminated the Shutterstock merger, when the capital structure was the immediate problem. A licensing deal is a revenue story, and Getty's problem on 21 June was a debt story.

What the plan does

According to the company announcement relayed by StreetInsider, holders of 83.5% of the senior unsecured notes agreed to the retirement of the $270 million of those notes, about 89% of secured bondholders and term loan holders and all revolver lenders agreed to lower interest obligations and extend maturities, and Koch Equity Development will provide up to $143 million of financing to fund the process. The announcement gave no total for the interest saved. Vendors are to be paid in full, and the company anticipates no impact on employees, customers, content contributors or partners. Those are the company's statements at the signing of an agreement, not outcomes. Debtwire reported on 11 September that Getty was preparing for a possible filing, and Bloomberg reported debtor-in-possession financing talks on 28 September; the 12 October announcement is the company's own account of the plan.

In plain terms the plan deletes the unsecured layer entirely, reduces the cost of the secured layer and replaces the owners. Whether that is enough depends on figures the announcement does not give: the post-restructuring debt total and its interest rate. Without those, the interest-cover ratio after emergence cannot be calculated, and anyone offering one is estimating.

The court rulings and the 'AI companies pay' thesis

A second narrative holds that Getty's decline is the cost of AI companies training on its images without paying. The UK litigation shows why that thesis is unsettled. In November 2025 the High Court (Joanna Smith J) rejected Getty's argument that Stable Diffusion was itself an infringing copy of its works. Getty had dropped its primary infringement claim because the training took place outside the UK. The court found that an early version of Stable Diffusion technically infringed Getty's trade marks but that the infringements were historic and limited, and most of the trade mark claim failed. Getty has been granted permission to appeal the secondary copyright point, which turns on whether a model must store an actual copy of a work to count as an infringing copy. The sources reviewed do not give a hearing date.

For readers weighing the thesis, the practical point is that the one major ruling so far went against Getty on the central copyright question and that, even if Getty wins on appeal, a legal victory would not repair a capital structure. The opposite view is that licensing deals such as the one with OpenAI show a market for permissioned data forming regardless of litigation outcomes. Both can be true. The Getty case shows a licensor can be commercially correct about the value of its archive and still fail financially, because the bill that comes due is interest, not royalties.

What would change this reading

Three disclosures would sharpen it: the post-emergence debt and interest cost, a split of the subscriber decline between the end of free trials and lost search traffic, and the size of the OpenAI payments. Until then, the evidence supports a narrower claim than the usual headline: Getty's balance sheet would have been a problem on its own, AI-driven substitution is visible in search referrals, subscribers and agency revenue, and the sources do not show that either one alone caused the filing.