Published May 12, 2025 · By CampaignsLive · Insights
The death-of-stock-photography conversation has been running in earnest since roughly mid-2023. The thesis is that AI image generation has made the entire stock photography category structurally obsolete: any image a stock library can sell, an AI tool can generate cheaper, more specifically, and more flexibly. The thesis has been overstated, the timeline has been wrong in most predictions, and the actual collapse has been more uneven than the headlines have suggested. But the collapse is happening.
This is a working assessment from the midpoint of 2025, eighteen months after the version of the thesis that turned out to be roughly correct started circulating in earnest.
What has actually happened to the major players
Three patterns are visible.
Getty Images. Getty has held up better than its competitors, partly by virtue of its editorial and news business (which AI does not threaten) and partly by virtue of an aggressive IP-protection posture (the litigation against Stability AI being the most visible piece). Getty’s commercial stock business has contracted; the editorial side has not. The company has launched its own AI image generation product, trained on its own licensed library, which has positioned it as one of the few stock providers with a defensible AI play. The financial impact through 2024 was negative but not catastrophic.
Shutterstock / Pond5. Shutterstock’s response to the threat was, by 2024, the most aggressive of the major players. The company licensed its content library to OpenAI for training in late 2022 and built its own generative product on top of multiple model providers. The commercial-stock revenue continued to decline. The licensing revenue partially offset the decline. The company’s net financial position through 2024 was worse than 2022, in line with the trend the death-of-stock thesis predicted.
Adobe Stock. Adobe’s approach was structurally different. The company built its Firefly generative AI on its own ethically-sourced content corpus (Adobe Stock contributors who opted in for training compensation) and integrated the result into Adobe’s creative software. The strategic positioning was less about defending stock photography as a category and more about absorbing the generative use case into Adobe’s broader creative-tools business. The stock business has declined; the broader Adobe business has held up.
The mid-tier libraries. The libraries below the top three — Depositphotos, iStockphoto, Alamy, and the longer tail — have been the worst affected. The death-of-stock thesis has been most accurate at this tier. Several have run material layoffs, contributor compensation has reduced, and the volume of new contributions to these libraries has dropped significantly through 2024.
What categories of stock have collapsed
Not all of stock has collapsed at the same rate. The categories where the collapse has been most complete:
Conceptual stock photography. “Businessman shaking hands,” “team brainstorming around a table,” “woman drinking coffee thoughtfully.” The generic conceptual imagery that has historically padded corporate websites, slide decks, and middle-tier brand creative is the cleanest case for AI replacement. The category was already largely synthetic in spirit; making it actually synthetic was a small step.
Background and environmental imagery. Skies, weather, water, foliage, architectural exteriors, urban scenes. The atmospheric material that historically existed as a category in stock libraries because it was uneconomic to shoot bespoke for each use. AI generation handles this category cleanly and cheaply.
Generic product shots. The on-white, single-product e-commerce-style photography that was the long tail of stock libraries. Most of this has moved into generative pipelines or into bespoke in-house production, both of which have eroded the stock-library use case.
What categories of stock have not collapsed
Three categories have held up:
Editorial photography. News, sports, current events, real people in real situations. The category that depends on what actually happened rather than what looks plausible. AI generation cannot, by definition, replace this. Getty’s editorial business is the clearest demonstration that this category remains structurally protected.
Talent and likeness-rights imagery. Stock of specific recognizable people — actors, athletes, politicians — that requires rights clearances. AI generation has not solved the rights side of this. Brand teams that need this material still source it through traditional licensing.
Specialized industry photography. Medical, scientific, technical, industrial imagery where the audience is expert enough to detect synthetic markers and where accuracy is consequential. The trade press in these verticals has continued to use traditionally-sourced photography because the synthetic alternatives are not reliably accurate.
What stock contributors have done
The longer human story of the stock-photo collapse is what the photographers, models, and stylists who depended on the category have done.
Many have moved out of stock contribution entirely. The economics of contributing to stock libraries — already thin in the late 2010s — collapsed below sustainable levels for most contributors through 2024. The volume of new submissions to the major libraries dropped materially.
Some have moved into AI-assisted production, contributing to libraries’ generative products in new arrangements (Adobe Firefly’s contributor program being the most visible example). These programs offer compensation tied to model training rather than to per-license sales, which is a structurally different relationship than the prior one. The compensation is currently lower; the long-term shape is unclear.
A smaller group has moved into bespoke production work — direct relationships with brands and agencies, replacing the stock-license-and-adapt workflow with a commission-and-produce workflow. This has been the most economically sustainable of the responses for the photographers who have been able to make the transition.
What this means for brand teams
Two things.
The stock library is no longer the default for the categories where it used to be the default. Brand teams that have historically reached for stock as the cheap, fast, good-enough option for conceptual, environmental, and generic imagery should not be doing that in 2025. The AI generative path produces better, more specific, more brand-appropriate work at lower cost. The stock library remains the default for editorial, likeness, and specialized industry imagery — and for those categories, it will continue to be.
The licensing and rights story has not gotten simpler. A brand using AI-generated imagery in place of stock is taking on the rights complexity that the stock license historically resolved cleanly. The shift from “licensed stock with clear rights documentation” to “generated imagery with platform-specific rights terms” is a real complexity addition, and brand teams that have not paid attention to it have run into the kinds of issues the Why Your Brand Should Own Its AI Creative piece covers in detail.
The longer trajectory
By the midpoint of 2025, the stock photography category has bifurcated cleanly. The high-end of the market — editorial, talent, specialized — is structurally protected and will be there in five years. The low-end of the market — conceptual, environmental, generic — has effectively been replaced by AI generation. The middle is being squeezed in both directions, and the libraries that have done well are the ones that built their own AI play to capture the displaced volume rather than trying to defend the old category from the outside.
The collapse the original thesis predicted is happening. It is happening unevenly, slower in some places than predicted and faster in others, with more secondary effects than the simple version of the story suggested. The category will not disappear, but it will not look anything like its 2022 self in the second half of the decade.
For the related discussion of how the broader production stack reorganized around the shift, see How Agencies Restructured Around AI Creative in 2024.