What Getty's Q2 earnings say about licensed ad creative

Getty Images’ Q2 2026 results, reported Aug. 10, 2026, are a useful benchmark for ad creative licensing because they put commercial creative, editorial demand, subscriptions, iStock, video, and AI pressure into the same dated record. The short read is not that licensed stock collapsed. Total revenue was $229.1 million, down 2.5% year over year and down 4.1% on a currency-neutral basis; Creative revenue was $127.4 million, down 2.6%; Editorial revenue was $96.5 million, up 9.2% and up 7.6% currency neutral.[1]

The more useful read is split demand. Getty said annual subscription revenue rose to 58.8% of total revenue from 53.5%, Premium Access represented more than 40% of revenue, Premium Access revenue grew 5.5%, and retention was near 100%.[2] At the same time, management described pressure on iStock from search-engine referral declines as engines implement AI-generated answers, said agency revenue fell 13%, and said microstock continues to be affected by generative AI.[2]

The weak part of the record looks like the low-differentiation long tail

The pressure points line up too neatly to treat them as one-off noise. Getty’s LTM purchasing customers fell 10.0% to 636,000, LTM active annual subscribers fell 25.2% to 240,000, and LTM paid downloads fell 2.8% to 90 million.[1] Those are usage and customer-count measures, not just accounting lines, so they matter for anyone buying licensed creative by habit.

But the subscriber number needs care. Management tied the active annual subscriber decline to both the June 2025 exit from the iStock free-trial program and ongoing search-related traffic headwinds.[2] That means the 25.2% decline should not be read as a clean measurement of buyers abandoning licensed stock. Part of it reflects Getty choosing to remove a trial mechanism; part of it reflects fewer buyers arriving through the discovery paths that historically fed iStock.

The more important operational signal is where the pressure appears. CEO Craig Peters said iStock is facing “search engine referral traffic declines and the knock-on impact to our affiliate traffic sources as the search engines implement AI generated answers.” He also said agency revenue was down 13% and that microstock continues to be impacted by generative AI.[2] For a media buyer, that describes a specific lane: inexpensive, broadly substitutable imagery bought because it is fast, good enough, and easy to find.

That is exactly the lane where AI generation competes hardest. A generic SaaS background, a loose lifestyle concept, a placeholder product mood board, or a low-risk social variation does not always need the same provenance as a campaign using recognizable people, editorial context, or regulated claims. When discovery happens through search and the search result itself starts answering with synthetic options, the buyer may never reach the microstock page. When the creative job is judged mainly on variation cost, the friction of licensing can start to look expensive even when the individual stock asset is cheap.

The paid-download decline is smaller than the customer and subscriber declines, which also matters. It suggests the market is not simply turning off licensing. Fewer or different buyers may be doing more of the remaining purchasing, and higher-value subscription relationships can carry more weight than casual one-off usage. Getty’s filings cannot prove what private brands, agencies, or performance teams are doing inside their creative workflows. They can show that the bottom end of the licensed funnel is leaking, and that management is explicitly connecting part of that leakage to AI search and generative tools.

The sticky part is enterprise, editorial, and Premium Access

The other side of the record is just as important. Premium Access represented more than 40% of Getty’s revenue, grew 5.5%, and held retention near 100%.[2] Subscription revenue rose to 58.8% of total revenue from 53.5%.[2] Getty also said corporate and media customers represented roughly 75% of revenue.[2] Those are not the usage patterns of a market where every buyer has decided that generation replaces licensing.

Premium Access is not just a prettier version of microstock. It is closer to procurement infrastructure: negotiated access, rights management, account servicing, predictable workflows, and a catalog that can satisfy teams that cannot afford a vague answer when legal, brand, or compliance asks where an image came from. That does not make every Premium Access asset irreplaceable. It does mean the buying reason is different from the reason someone downloads a cheap generic image for a low-stakes variant.

Editorial revenue makes the same point from another angle. Editorial revenue rose 9.2% year over year to $96.5 million, even while Creative declined.[1] Editorial assets carry a different kind of value: recognizable events, public figures, sports, entertainment, news context, and documentary credibility. Synthetic imagery can imitate a visual style, but it cannot become the actual photographed event. That distinction still matters in advertising when a campaign wants cultural adjacency, sponsorship context, public-event credibility, or a brand-safe path around real-world subjects.

Getty also reported custom content solutions growth of more than 350% and Unsplash+ growth of more than 15%.[2] Those are smaller lines than the core revenue categories, and they should not be inflated into proof that all licensed creative is healthy. They do show that demand is not only sitting in the old stock-library model. Buyers are still paying for managed creation, curated access, and cleaner commercial use cases when the job demands more than a quick substitute image.

The sourcing rule: locate the creative job before moving the budget

The buying decision should start with the category of the creative job, not with a universal position on stock versus AI. Getty’s Q2 record supports a practical split: test AI generation against stock where the asset is generic, low-risk, and valued mainly for cheap variation; keep licensed assets in the mix where provenance, rights clarity, editorial trust, premium brand environment, or enterprise workflow can become the expensive part of the campaign.

• More exposed to AI generation: generic backgrounds, abstract category imagery, simple lifestyle placeholders, low-budget social variations, early concepting, and creative where synthetic sameness will not damage the offer.
• More defensible for licensed sourcing: recognizable people, editorial adjacency, regulated products, premium brand campaigns, cultural or news context, claims-sensitive landing pages, and assets that legal or procurement must be able to trace.
• Worth testing both ways: product-ad variants, ecommerce support imagery, evergreen display concepts, and performance creative where the cost of production matters but brand perception still has measurable downside.

That middle category is where budget waste often hides. A team may keep licensing because it is familiar, even though AI could produce enough variants for a low-risk test. Another team may generate because the unit cost looks attractive, then spend the savings on review cycles, retouching, legal hesitation, or lower trust when the creative looks synthetic. The site’s earlier records on the generative AI ad perception gap and the AI creative trust gap are useful checks here, especially when a campaign relies on the audience believing the image reflects a real product, person, place, or moment.

A workable sourcing review can be short. Before shifting spend away from licensing, ask what would actually fail if the asset were synthetic: performance, legal approval, platform review, brand fit, customer trust, or internal sign-off. If the honest answer is “almost nothing,” the asset belongs in the test lane. If the answer involves rights, real-world subjects, compliance, or brand credibility, the licensed option still has a job to do.

This is also where creative cost comparisons need context. A generated image can be cheap per unit, and that matters; the site’s GPT API ad creative pricing comparison is relevant for that part of the decision. But the right comparison is not only file cost versus license cost. It is usable approved asset cost: prompts, editing, rejection, legal review, platform suitability, and the cost of replacing an asset if someone challenges it late.

Video gives a warning, not a price conclusion

Getty’s video numbers point in the same direction, but they should not be overstated. The company’s video collection grew 11.7% year over year to 39 million assets, while the LTM video attachment rate fell 150 basis points to 15.2%.[1] That combination suggests supply is expanding faster than licensed video attachment demand.

It does not prove video pricing pressure. Getty does not disclose enough price detail to make that claim from these figures alone. For buyers, the safer inference is narrower: there may be more available licensed video inventory than there is incremental attachment behavior, and AI video tools are likely to intensify the same generic-asset pressure already visible in microstock. Premium, rights-sensitive, or editorial video remains a separate buying problem.

Do not turn the investor story into the licensing story

The surrounding market coverage is noisy because Getty’s Q2 release also contained real financial strain. Adjusted EBITDA was $62.3 million, down 8.4%, with adjusted EBITDA margin at 27.2% compared with 28.9% a year earlier.[1] Free cash flow was negative $122.6 million, driven largely by a $110.9 million warrant-litigation payment; the CFO said free cash flow excluding that and other specified items would have been approximately negative $4.5 million.[2] Getty also withdrew guidance and disclosed going-concern risk language.[3]

That financial context explains why many search results around the earnings event focus on misses, liquidity, and the stock reaction. StockStory reported that Getty missed Q2 CY2026 revenue estimates and that the stock dropped 15.2%.[4] Those facts matter if the question is investor risk. They are not the cleanest answer to whether a media buyer should license a campaign image, generate it, or test both.

One line in particular should be kept out of the licensing-demand argument: “Other” revenue. Getty’s Other revenue fell 66.7% year over year, but that decline is largely about the timing of OpenAI-deal revenue, with the deal signed in Q3 2025 and recognized mostly in 2025.[1][2] It is not evidence that ordinary ad creative licensing demand suddenly evaporated.

What Getty’s Q2 record can and cannot prove

Getty is unusually useful because it touches several parts of the creative supply chain in public: premium commercial licensing, iStock, subscriptions, editorial, video, and AI licensing. It is still one company. Its mix, pricing, sales execution, customer base, and debt position are Getty-specific. A private agency’s internal production data, a platform’s generated-ad adoption, or a brand’s legal rejection rate could all move differently.

So the earnings record should be used as a benchmark, not a law. It supports the conclusion that the cheap end of licensed stock is vulnerable where AI generation and AI-generated search answers reduce the need to visit, browse, and license a generic asset. It also supports the conclusion that enterprise, editorial, and premium-access licensing still has defensible demand where rights, provenance, workflow, and trust are part of the asset’s value.

Peters also argued that consumer sentiment is slowing AI use in ad creative.[2] That is an executive view, not an independent measurement inside this earnings record. It is directionally relevant, but it should be checked against buyer and consumer data rather than accepted as a reason to avoid AI creative outright. The better use is tactical: if consumers can tell an ad is synthetic and trust falls, that is a reason to reserve generation for categories where perceived artificiality has limited downside, not a reason to keep licensing every generic image.

The media-buyer read

Getty’s Q2 2026 licensing record says the cheap, price-sensitive end of stock is leaking toward AI, especially where discovery used to depend on search referrals and the asset itself is easy to substitute. It does not say licensed creative is dead. Premium Access, editorial, corporate, and media demand are behaving more like defensible procurement categories than like abandoned libraries.

Before reallocating budget, classify the creative job by risk, distinctiveness, rights needs, and tolerance for synthetic-looking output. If the campaign only needs inexpensive variation, AI should compete directly with stock. If the campaign needs provenance, recognizable subjects, editorial credibility, or enterprise approval, licensed assets still have a reason to earn their cost.

References

1. Getty Images Reports Second Quarter 2026 Results — Getty Images, Aug. 10, 2026
2. Getty Images (GETY) Q2 2026 Earnings Call Transcript — The Motley Fool, Aug. 17, 2026
3. Getty Images Holdings, Inc. Reports Material Event 8-K — StockTitan
4. Getty Images (NYSE:GETY) Misses Q2 CY2026 Revenue Estimates, Stock Drops 15.2% — StockStory