The Death of Stock Photography in 2026: Why Getty's $4B Business Is Collapsing

On March 16, 2026, Getty Images reported full-year 2025 revenue of $981.3 million - the highest in the company's 30-year history. Six weeks later, on May 7, the same company guided 2026 revenue down to a range of $948-988 million, with Q1 2026 creative revenue already off 4.5% year-over-year and down 8.0% on a currency-neutral basis. Shutterstock, meanwhile, just posted a Q1 2026 net loss of $47.6 million on Content revenue that fell 12% year-over-year. (Getty Images Q4 2025 release, Seeking Alpha 2026 guidance, Shutterstock SEC 8-K Q1 2026)

In the same quarter, Adobe Firefly crossed 24 billion cumulative generations, hit 6 million monthly active users, and announced that AI-generated content now makes up 80% of Adobe Stock's illustration category. (Adobe Firefly 2026 stats)

That is what an industry collapse looks like when you take the camera off the press release and put it on the ledger. The $4 billion stock photography business is not dying because customers stopped needing pictures. It is dying because the marginal cost of a usable picture fell from $12 (Getty single-image license) to roughly $0.014 (one Flux 2 generation), and the choice between "search a library" and "type a sentence" stopped being a debate eighteen months ago. This piece walks through what happened, what's left of the licensing business, and where stock photography still wins.

Empty photo studio with a single softbox lit - the visual metaphor for an industry where the lights are going out

The $4B industry that ran the internet for 20 years

Stock photography is older than the web. Getty Images was founded in 1995 by Mark Getty and Jonathan Klein with a thesis that intellectual property would be the oil of the 21st century. They were right - just not about the form. Through the 2000s and 2010s, Getty, Corbis (which Getty acquired in 2016), Shutterstock, iStock, Adobe Stock, and a handful of European players (Alamy, Depositphotos, 123RF) carved up a market that, by 2024, multiple research firms pegged between $3.7 billion and $6.8 billion in annual revenue depending on how you counted editorial, creative, video, and microstock. (Business Research Insights stock photography)

The business model was simple. Photographers shot inventory on speculation. Agencies took 60-85% of every license. Customers - magazines, ad agencies, corporate marketing teams, bloggers, news organizations - paid anywhere from $10 per microstock download to $499 for a Getty Premium Access single use. The system worked because creating a usable commercial photograph required a camera, a lighting kit, a model release, a location scout, retouching time, and a keywording pass. That stack put the per-image cost floor somewhere around $80-200 to produce, which is why even a $12 license cleared margin once you sold the same image enough times.

The internet didn't kill stock photography. The internet scaled it. Shutterstock IPO'd in 2012 at a $1.5 billion valuation. Getty went public via SPAC in 2022 at $4.8 billion. As recently as January 2025, Getty and Shutterstock announced a "merger of equals" that valued the combined company at $3.7 billion in enterprise value - a deal explicitly pitched as defensive against generative AI. (Getty + Shutterstock merger announcement)

The merger has not yet closed as of June 2026, and the eighteen months of regulatory drift while the deal hangs has lined up almost perfectly with the period in which their core business stopped working.

Why AI killed stock photography in 18 months

There are three reasons generative image models displaced stock libraries faster than anyone predicted, and they compound on each other.

The first is cost. A Getty Creative license for commercial web use runs $175-499 depending on size and territory. A Shutterstock subscription gives you 10 images for $29/month - roughly $2.90 each, with a quality floor that drops the moment you scroll past the first page of results. Flux 2 via the Black Forest Labs API costs $0.014 per generation. Midjourney V8 unlimited Standard is $30/month. Imagen 4 Ultra through the Google Cloud API costs $0.04 per image. Versely bundles 50+ image models for less than the price of one Getty Premium license. The numerator hasn't moved. The denominator collapsed by three orders of magnitude. (Best AI Image Generators 2026)

The second is specificity. Stock libraries are search problems - you type "diverse team in modern office" and scroll through 4,000 hits trying to find the one where the laptop screen isn't a 2015 Macbook Pro and the conference room doesn't have a clock from 2008. AI image generation is a specification problem - you describe exactly the room, the laptop, the lighting, the demographics, and the model gives you that image, then four variations of it. Specificity is what the brief always wanted; libraries just couldn't deliver it without a custom shoot.

The third is freshness. A stock library's value depends on whether its inventory matches what brands need to ship today. In 2024, customers wanted photos that looked like 2024 - the right phones, the right cars, the right office aesthetic. In 2026, the same customers want photos that look like 2026, with current devices, current fashion, current ethnic distributions in their team shots. Stock libraries cannot re-shoot their entire inventory annually. AI models that generate fresh imagery from current training data have no such constraint. Every brand that ships a deck or a landing page in 2026 is choosing between "library photo that looks two years old" and "generated image that matches the brief exactly." The choice is not subtle.

The combined effect shows up in Getty's own numbers. Q1 2026 creative revenue (the part of the business that sells stock photos for marketing use) is down 8.0% on a currency-neutral basis. Editorial revenue (news, sports, celebrity photography) is up 11.0%. The split is the whole story: the part of the business that AI replaced is shrinking; the part it can't replace yet is growing. (Getty Q1 2026 results)

Photographer reviewing images on a laptop in a coffee shop - the contributor experience that is rapidly disappearing

The Getty pivot: licensing the library back to the machines

Getty has not stood still. The pivot, visible across its 2025 earnings calls, is to monetize the library as training data rather than as end-customer inventory. The most visible expression of that strategy is the company's investment in Bria, an Israeli startup that builds enterprise generative AI image models trained exclusively on licensed data. Getty is both a minority investor in Bria and one of its 30+ data partners (alongside Alamy, Envato, Freepik, and Depositphotos). Bria pays Getty for image rights, trains models on the licensed corpus, and revenue-shares back to contributors on a "Spotify-for-AI" model that allocates payouts based on each image's contribution to a generated output. (TechCrunch on Bria funding, Digiday on Bria training data)

Bria reported 400%+ year-over-year ARR growth in 2025 and raised $40 million in Series B at a valuation that suggests the licensing-clean enterprise generative model is the genuinely scarce asset in the AI image market.

The second leg of Getty's pivot is the copyright settlement business - or more precisely, the lawsuits that didn't produce settlements. On November 4, 2025, the UK High Court delivered judgment in Getty Images v. Stability AI, the most-watched generative AI copyright case in the world. The court rejected Getty's central copyright claim, ruling that AI model weights are not a "copy" of training images under the Copyright, Designs and Patents Act. The court found limited trademark liability for some Stable Diffusion outputs that reproduced Getty watermarks, but the structural claim - that training a generative model on copyrighted images is itself infringement - failed on jurisdictional grounds because the training did not occur in the UK. (Mayer Brown on Getty v Stability AI, Bird & Bird on the ruling)

That ruling matters because it foreclosed the path most stock agencies had assumed would generate a parallel revenue stream: suing AI companies into licensing deals. With the UK precedent in place, and U.S. cases still grinding through the Northern District of California, the bargaining power of stock agencies against AI labs is materially weaker than it was a year ago. Licensing deals will happen - they already are, with the New York Times, News Corp, and the Associated Press all signing OpenAI agreements - but on terms set by the labs, not the libraries.

The third leg is the in-house AI tool. Getty launched its own generative AI service in 2023, trained exclusively on Getty's licensed library with full indemnification. The merger filings with Shutterstock heavily emphasize the combined entity's plans to "invest in opportunities inclusive of AI" and offer "AI services alongside pre-shot and custom offerings." Translation: the combined company expects to be in the model-licensing business as much as the image-licensing business by 2028.

Shutterstock's contributor problem

Shutterstock's situation is structurally worse than Getty's because its core model depends on a creator marketplace that is quietly emptying. Q1 2026 Content revenue at $178.1 million was down 12% year-over-year. The Data, Distribution, and Services line (which includes AI training data licensing - the supposed bright spot) fell 47% to $21.0 million. The combined effect was a $47.6 million net loss in a quarter the previous year had been an $18.7 million profit. (Shutterstock 8-K Q1 2026)

The contributor side of Shutterstock is also under stress. Adobe Stock, which historically competed for the same contributor pool, now has 80% AI-generated content in its illustration category and 25%+ of all new submissions incorporating Firefly-generated elements. Adobe's automated review system has rejection rates that some contributors report at 40%, flagging "similars" and "quality" issues that often trace back to over-saturated AI-generated content categories where every prompt produces a near-duplicate of every other prompt. (Adobe Firefly stats)

The result is a flywheel running in reverse. Contributors who once supplied human photography see falling per-image royalties as AI-generated content floods marketplaces. Fewer contributors means less fresh human inventory. Less fresh human inventory means less reason for customers to keep subscriptions. Fewer subscriptions means lower contributor royalties. The cycle has been visible in Shutterstock's contributor count for two consecutive years, though the company has stopped breaking it out in earnings releases.

Empty conference room with AI-generated business stock imagery on the wall display - the future of corporate visual content

Five industries that replaced stock with AI in eighteen months

The collapse is not theoretical. Talk to anyone running a marketing budget in 2026 and the shift has already happened in five categories:

E-commerce product backgrounds and lifestyle shots. Shopify merchants who previously paid Shutterstock $29/month for hero imagery now generate every product backdrop in Versely or directly through Flux. Lifestyle shots - "person using product in natural setting" - are the canonical AI-generation use case and have been since mid-2024.

SaaS marketing and B2B web design. The diverse-team-in-modern-office archetype that funded a quarter of Getty's creative business is now generated wholesale. Mid-market B2B companies have moved their entire hero-image budget from stock libraries to AI image tools, with monthly spend dropping from $400-800 in stock subscriptions to $20-50 in API credits.

Content marketing and blog illustrations. Editorial illustrations - the visual sidebars that fill HubSpot blogs and Substack posts - moved to AI in 2025 and never moved back. The combination of cost ($0.014 per generation) and brand-voice consistency (fine-tuned models trained on a brand's existing aesthetic) closes the door for stock libraries here permanently.

Real estate and architectural visualization. Listings that previously used stock interior shots for staging now use AI-generated rooms specified to the property's footprint. The accuracy gap closed in late 2025 with Imagen 4's improved spatial reasoning.

Social media content for SMBs. This was always the contested middle of the stock photo market - small businesses that couldn't afford custom shoots but couldn't make do with the visible-watermark "free" tier. AI generation killed this category fastest because the customer was most price-sensitive and the quality requirement was most forgiving.

For creators running multi-platform content, the math is even more lopsided. A single Versely subscription covers AI image generation across 50+ models, plus text-to-video, music generation, and the AI slideshow maker - all for less than the cost of one Getty Premium license. We've written about how this restructures content economics in our AI content creation cost breakdown.

Where stock photography still wins (and probably always will)

Stock isn't dead in every category. There are at least four use cases where licensed libraries are not just surviving but growing, and Getty's 11% YoY editorial revenue growth is the strongest signal.

Breaking news photography. Reuters, AP, and Getty wire photographers shoot the war, the election, the disaster, the press conference. AI cannot generate documentary imagery of events that actually happened, and the legal and editorial standards for news photography prohibit synthetic substitution. This category is structurally protected.

Sports. Getty's exclusive contracts with leagues, federations, and major events (the IOC, FIFA, Formula 1, the major U.S. leagues) generate inventory that AI cannot replicate. A photo of Messi at the moment of a specific goal in a specific match has no AI substitute and never will.

Celebrity and editorial portraits. Likeness rights and editorial standards make synthetic celebrity imagery a legal minefield. Magazines, news outlets, and PR teams that need real photos of real people at real events continue to license heavily.

Historical and archival imagery. Getty Archive holds millions of images stretching back to the 19th century. Documentary filmmakers, museums, and educational publishers depend on this inventory, and it has no AI analog.

The pattern across the four categories is the same: stock photography wins where the requirement is that the image depicts something that actually happened or someone who actually exists. AI wins where the requirement is that the image illustrates a concept. The first market is a few hundred million dollars a year and stable. The second market is what built the $4 billion industry, and it is the part that has collapsed.

Why creators don't need stock libraries anymore

For independent creators, freelancers, agencies, and SMBs, the stock library subscription is now a line item to cut, not a line item to negotiate. The replacement stack in 2026 is straightforward: an AI image generator with multiple models for variety, an image-to-video tool for animating those stills, and a captions or overlay tool for finishing.

Versely bundles that stack. The AI image generator gives access to Flux 2, Midjourney V8, Imagen 4, GPT Image 2, Nano Banana Pro, and 45+ other current models from a single interface. The image-to-video tool animates any generated still using Sora 2, Kling, Veo 3.1, or Hailuo. The AI slideshow maker sequences multiple generations into ready-to-post short-form video. For the customer who used to pay Getty $499 for a single use license, that whole stack is less per month - and produces unlimited fresh imagery rather than one frozen photograph.

We've covered the broader model landscape in our latest AI image models roundup for creators trying to figure out which generator fits which use case.

Creator at a laptop generating images, the lightweight modern alternative to a studio shoot

FAQ

Is stock photography dead? The creative-imagery side of stock photography - lifestyle, B2B, marketing concepts, illustrations - is in terminal decline, and the numbers from Getty (creative revenue down 8.0% currency-neutral in Q1 2026) and Shutterstock (Content revenue down 12% YoY in the same quarter) make that clear. Editorial, news, sports, and archival photography remain healthy and are actually growing. So stock isn't dead; the parts of stock that AI can replace are dying very fast, and the parts it can't are fine.

Should I cancel my Getty or Shutterstock subscription? If you use stock primarily for marketing, blog illustrations, social media, lifestyle imagery, or concept visuals, yes. The math has stopped working - a single AI image tool subscription gives you more usable output at 5-10% of the cost. If you use stock for editorial publishing, news content, or imagery that must depict real events or real people, keep your subscription but renegotiate it. Pricing power is shifting toward customers as agency revenue declines.

Will AI image generation be sued out of existence? The November 2025 UK ruling in Getty v. Stability AI - which rejected the central copyright claim against generative training - significantly reduced this risk. Pending US cases could still produce different outcomes, but the trend is toward "training on publicly available imagery is permissible if the model weights don't constitute a derivative work." Models that train exclusively on licensed data (Adobe Firefly, Bria, Getty's own AI tool) offer additional safety for enterprise customers who need indemnification. We covered the legal landscape in our AI copyright and safety guide.

What about Adobe Firefly - isn't that just stock with AI features? Adobe Firefly is the closest thing to a "winner" of the stock-vs-AI transition because Adobe owned the contributor marketplace, the photo editing software, and the AI model simultaneously. Firefly is trained on Adobe Stock's licensed library, generates indemnified output, and integrates directly into Photoshop and Premiere. It's also why Adobe is one of very few companies whose stock photo business is growing rather than shrinking - because they pivoted the contributor model to AI submissions years ahead of competitors.

Are stock photographers losing their jobs? Yes, in significant numbers, though it's distributed and slow-moving. Microstock contributors who shot lifestyle, B2B, and concept imagery for $0.25-3 per download have seen volume collapse over 2024-2025. Editorial photographers, sports photographers, and news photographers are largely unaffected and in some cases better paid as the value of authenticated, real-world imagery rises. The mid-tier - commercial photographers who shot agency campaigns and inventory for stock libraries - is the hardest hit. Many have pivoted to becoming AI image directors, consulting on prompt engineering and brand-voice fine-tuning rather than shooting.

The bottom line

Stock photography is not collapsing because anyone changed their mind about wanting visual content. It is collapsing because the unit economics of generating a usable image fell by three orders of magnitude in eighteen months, and an industry built on charging $12-499 for a license cannot survive a competitor charging $0.014 per generation. Getty's editorial business will be fine because nothing AI can do replaces a photo of an event that actually happened. Getty's creative business - and the entire microstock industry behind Shutterstock, iStock, Depositphotos, and the rest - is in structural decline that the pending Shutterstock merger cannot reverse.

For creators, the practical action in 2026 is to stop paying for inventory you don't need. Open the AI image generator, describe exactly the image your brief calls for, and ship in minutes for cents. The stock photo era is ending. The generated-image era - specific, fresh, branded, cheap - has already replaced it for everyone who actually had to pay for the old one.

Sources: