Skip to content

AI data market

Shutterstock AI data licensing analysis: catalog scale is not revenue scale

By SourceX Editorial · Reviewed by Noah Loul ·

10 sections

Short answer

A Shutterstock AI data licensing analysis has to keep three numbers apart. The catalog count, 800 million images and 59 million clips at December 31, 2024, measures approved stills and footage, not the metadata licensed for training. The training right is carved out of the standard license and sold separately. Revenue sits in a blended segment, $203.3 million in 2025, with no dollar figure for the data offering.

Value drivers

Shutterstock AI data

This page covers shutterstock AI data licensing analysis: catalog scale is not revenue scale.Illustrative

Key takeaways

  • Shutterstock counted 800 million images and 59 million footage clips at December 31, 2024; its training product licenses metadata across images, footage, music and 3D.
  • Its standard license prohibits training use, so AI revenue depends on a separately sold dataset grant.
  • Data, Distribution, and Services revenue rose 16% to $203.3 million in 2025, but the segment also holds Giphy and Studios work.
  • Data, Distribution, and Services revenue fell 16% to $56.1 million in the second quarter of 2026; Shutterstock gives no dollar figure for the data offering.
  • The Getty merger agreement ended on July 7, 2026; the $173.7 million goodwill charge, triggered by Getty's June 30 announcement and booked in the quarter ended that day, values the company, not the catalog.

What does a Shutterstock AI data licensing analysis have to separate?

A Shutterstock AI data licensing analysis has to separate three numbers the company reports in different places: how many assets its catalog holds, what it actually licenses for model training, and the revenue line where that licensing lands. Each is measured in a different unit, none converts into another, and the gaps between them are where an investor's questions belong.

The investor question is narrower than whether stock content is useful to AI. It is whether a catalog built for one licensing business can be priced as a training-data asset from public evidence alone. Useful evidence would show the licensable scope in the same unit as the revenue, a rights chain that permits training use, and a dollar figure for the data offering over a stated period.

Shutterstock publishes enough to test the first two in part and the third not at all. This analysis uses only its filings, its license and contributor pages and the merger filings, each read on 2026-10-28, and leaves every undisclosed field blank rather than filling it with press estimates.

Catalog scale: what 800 million images counts, and what it leaves out

The catalog count measures approved stills and footage clips on Shutterstock's marketplace, not the material it licenses for training. Its annual report for 2024, as amended, says the collection had grown to 800 million images and 59 million footage clips as of December 31, 2024. Its first-quarter 2024 earnings release defines content in the collection as the approved images (photographs, vectors and illustrations) and footage, counted in clips, in its library at the end of the period.

The training product is described in a different unit. The annual report for 2025 says the Data, Distribution, and Services segment includes licenses of metadata tied to images, footage, music and 3D models, and that this metadata is used to train AI models. Music and 3D models sit outside the catalog count, and the metadata itself, the descriptions and labels attached to each asset, is not counted at all.

Two more adjustments separate the catalog from what a buyer can license. Shutterstock's contributor help center says an opt-out from data licensing was added to contributor account settings in January 2023, so assets that still appear in the marketplace count may be excluded from data deals. And no December 31, 2025 count appeared in the excerpts of the 2025 annual report read for this page, so the latest catalog size is treated here as unknown.

Licensing scope: the training right is carved out and sold on its own

Shutterstock's standard licenses withhold training use, so its AI revenue rests on a separate grant rather than on the ordinary customer license. Its license agreement states: "Using any Visual Content as training data for any artificial intelligence, machine learning, or generative AI system, tool, process, or dataset is prohibited." Its contributor help center describes the other side: companies purchasing Shutterstock datasets may only use them to train machine learning and computer vision models, and uses such as marketing or advertising are prohibited.

Supply comes in without exclusivity. The 2025 annual report says: "We do not typically require that contributors of content to our library provide their content to us on an exclusive basis, with the exception of custom content and certain editorial, music and other content to which we have exclusive distribution rights." Contributors, it adds, "earn royalties based on a tiered earnings rate schedule that is tied to annual licensing volume." The help center says dataset inclusion of existing library content is covered under Section 1a of the Contributor Terms of Service, which grants Shutterstock the right to develop new features and products.

On the customer side, the same report says content on its websites is offered under perpetual, royalty-free licenses, except certain custom, editorial, music and other content with specific licensing requirements, and that multiple customers can license the same image, footage clip or music track. Custom content is the exception, licensed on an exclusive basis to the customer that commissioned it.

The economics follow from those terms. A non-exclusive, perpetual catalog can be licensed for training to more than one developer, which is what makes an existing catalog attractive. The same non-exclusivity means the contributors behind it can supply other platforms, so the scarce element is the documented grant and the metadata Shutterstock attaches, not sole possession of the images.

Revenue: a blended segment with no figure for the data offering

Shutterstock reports training-data revenue inside a wider segment and does not disclose the data offering's size. The 2025 annual report says Data, Distribution, and Services revenue increased by 16%, to $203.3 million in 2025. The segment also holds Giphy distribution and advertising and Shutterstock Studios production work, so $203.3 million is a ceiling on data licensing revenue, not a measure of it.

The latest quarter shows the limit of that disclosure. The quarterly report for the period ended June 30, 2026 says Data, Distribution, and Services revenue fell 16% to $56.1 million in the quarter; Shutterstock gives no dollar figure for the data offering. A segment total cannot be split from outside: an investor learns that the blended segment shrank, not how much of the change came from training data.

Normalize before comparing. The annual figure is GAAP segment revenue for a calendar year; the quarterly figure is one quarter's segment revenue and its change from a year earlier. Neither is contract value, bookings or annual recurring revenue, and neither can be set beside a private company's run-rate or funding total as if they measured the same thing.

Shutterstock's disclosures, dated, with the blanks left blank

Every row comes from a Shutterstock or Getty Images document read on 2026-10-28. Not disclosed means the field did not appear in the filings and pages read, not that the figure does not exist. The figure after the table sets the catalog count, the training unit and the revenue line side by side.

The evidence table:

Shutterstock's disclosures, dated, with the blanks left blank
FieldWhat the document saysSource and dateMetric typeNot disclosed
Catalog scale800 million images and 59 million footage clipsForm 10-K/A for 2024, as of Dec 31, 2024Count of approved stills and clipsCount at Dec 31, 2025; music, 3D and metadata counts
Training unitLicenses of metadata tied to images, footage, music and 3D models, used to train AI modelsForm 10-K for 2025Product descriptionVolume of metadata licensed
Training rightStandard license prohibits training use; dataset buyers limited to training ML and computer vision modelsLicense page and contributor help center, read Oct 28, 2026Contract termsDuration and exclusivity of data deals
Contributor grantDataset inclusion under Section 1a of contributor terms; opt-out added January 2023Contributor help center, read Oct 28, 2026Contract termsShare of the catalog opted out
Segment revenueData, Distribution, and Services up 16% to $203.3 millionForm 10-K for 2025GAAP segment revenue, annualData offering revenue alone
Latest segment trendSegment down 16% to $56.1 million; no dollar figure for the data offeringForm 10-Q, quarter ended June 30, 2026GAAP segment revenue, quarterlyData offering revenue, customer count, renewals
Corporate statusMerger agreement with Getty Images terminated July 7, 2026; goodwill impairment of $173.7 million in the quarter ended June 30, 2026Form 8-K dated July 8, 2026; Form 10-Q, quarter ended June 30, 2026Event and non-cash chargeEffect on data licensing contracts
Three panels left to right: Shutterstock's catalog count of 800 million images and 59 million clips at December 31, 2024, the training unit of image, footage, music and 3D metadata with volume not disclosed, and the $203.3 million 2025 segment with data revenue alone not disclosed, separated by gaps marked different unit and different line.
Catalog count, training unit and revenue line, from Shutterstock's filings for 2024 and 2025, checked 2026-10-28.

The strongest alternative: catalog licensing is too small to matter

The strongest alternative reading is that the segment's revenue comes mainly from Studios production and Giphy distribution, leaving catalog licensing too small to move the investment case. The filings read here can neither confirm nor rule that out. They never state the data offering's revenue, and the second-quarter 2026 report gives the segment's 16% fall to $56.1 million without a dollar figure for the data offering. On the public record, the data offering could be most of the segment or a small part of it.

A second alternative is that 2024 and 2025 reflected catch-up purchases by model developers rather than repeat demand. The public record cannot settle that either, since the documents read give no count of data licensing customers, no contract durations and no renewal rates. Both readings should stay open until Shutterstock, or any catalog licensor, reports the data offering in dollars with its customer count.

Press figures that put a dollar value on Shutterstock's AI licensing are deliberately left out. The filings read for this page do not break that revenue out, and a secondhand number cannot stand in for a disclosure.

What the ended Getty merger and the impairment do and do not show

Shutterstock remains a standalone listed company after the planned combination with Getty Images fell away. Getty Images' annual report for 2025 says that on January 6, 2025 it entered into an agreement to combine with Shutterstock in a merger of equals. Shutterstock's Form 8-K dated July 8, 2026 says the merger agreement was terminated on July 7, 2026.

The second-quarter 2026 report records a non-cash goodwill impairment charge of $173.7 million, which it says resulted from the decline in the company's fair value after the announcement of the terminated merger agreement. The charge sits in the quarter that ended June 30, 2026: it was triggered by Getty Images' announcement that day, before the agreement was terminated on July 7. That charge compares the whole company's fair value with its book value. It is not a valuation of the catalog or of the training-data rights, and no filing read here puts a value on either.

The same report says Content revenue decreased by 17%, to $165.7 million, in the quarter, driven primarily by weakness in new customer acquisition. That matters to the catalog thesis in one specific way: the marketplace is what draws contributors and their uploads, so a shrinking content business can slow growth in the asset the data offering licenses.

What could make the catalog-licensing thesis wrong

Five developments could make an existing catalog worth less as training data than its size suggests. Each is visible, or conspicuously absent, in the documents cited above.

At a glance

What could make the catalog-licensing thesis wrong

  • Opaque size: the data offering's revenue, customer count and contract terms are not disclosed, so any thesis about it rests on the blended segment rather than on the offering's own figures.
  • Demand reversal: Data, Distribution, and Services revenue fell 16% year over year in the second quarter of 2026, with no dollar figure for the data offering; if model developers buy less stock media, catalog scale stops mattering.
  • Supply erosion: contributors can opt out of data licensing and generally supply Shutterstock non-exclusively, so the licensable share of the catalog can shrink or become common to several platforms.
  • Marketplace weakness: Content revenue fell 17% in the same quarter, putting pressure on the business that grows the catalog in the first place.
  • Law and strategy: court rulings on training without a license, which this page does not predict, could cut the price of licensed catalogs for some uses, and no filing read here sets out a plan for the data offering after the merger ended.

Questions to ask Shutterstock, any catalog licensor and SourceX

Ask each company for its data offering in its own unit and period, then compare only like with like. The same list applies to SourceX, which works from a different supply model and should face the same questions.

SourceX does not license from a contributor library. Its investor page describes licensing data straight from businesses and passing licensed datasets to AI developers with the rights and permitted uses written down, and its buyer request page says datasets are sourced on request rather than kept in stock. Under its published terms the supplying company signs off on scope and terms before it delivers anything, and licenses are exclusive for AI training, with the term and any alternatives set in each deal's Dataset Deal Confirmation. SourceX publishes no commercial figures and shares company metrics with investors directly; it is raising under Rule 506(c), so only verified accredited investors can take part, through the investor page. Questions worth asking:

Checklist · 6 items

  1. What was data offering revenue, in dollars, for each of the last four quarters and the last two years?
  2. How many data licensing customers paid in each period, and what share came from the largest one?
  3. What share of the catalog is opted out of data licensing, by media type?
  4. Which unit is priced in a training license: assets, metadata records, hours of footage or files?
  5. What do training licenses say about duration, exclusivity, renewal and limits on model use?
  6. How is catalog size counted today, and has the definition changed since 2024?

0 of 6 ticked. Tick each item as you go.

Methods note: sources and rules behind the Shutterstock table

The evidence table, the figure and the risk list were built on 2026-10-28 from Shutterstock's annual report for 2025, its amended annual report for 2024, its first-quarter 2024 earnings release, its quarterly report for the period ended June 30, 2026, its Form 8-K dated July 8, 2026, Getty Images' annual report for 2025, Shutterstock's license agreement page and its contributor help article on data licensing. Direct fetches of sec.gov and shutterstock.com were blocked, so each document was verified through search excerpts limited to its own domain; quoted wording appears in those excerpts.

Rules: figures stay in the unit and period the filer used and are never converted; a growth rate without a base is shown only as a rate; fields not found in the documents read are marked not disclosed; press estimates are excluded; no company is ranked. Not measured: the data offering's revenue, customer count and contract terms, the opted-out share of the catalog and the December 31, 2025 catalog size.

Frequently asked questions

Does Shutterstock report AI data licensing revenue separately?

Not in the filings read for this page. Training licenses sit in the Data, Distribution, and Services segment, which reported $203.3 million for 2025 alongside Giphy distribution and advertising and Shutterstock Studios. The quarterly report for June 30, 2026 gives the segment's revenue, down 16% to $56.1 million, and no dollar figure for the data offering.

Can a Shutterstock customer train an AI model on licensed images?

Not under the standard license, which prohibits using visual content as training data for any artificial intelligence, machine learning or generative AI system. Training use comes through separately sold datasets, which Shutterstock's contributor help center says buyers may use only to train machine learning and computer vision models.

What rights do contributors give Shutterstock for data licensing?

Contributors generally supply content non-exclusively and earn tiered royalties tied to annual licensing volume, according to the 2025 annual report. Shutterstock's help center says dataset inclusion falls under Section 1a of its Contributor Terms of Service, and contributors have had an opt-out from data licensing since January 2023.

Did Getty Images and Shutterstock merge?

No. The companies agreed in January 2025 to combine, but Shutterstock's Form 8-K dated July 8, 2026 says the merger agreement was terminated on July 7, 2026. Shutterstock recorded a $173.7 million non-cash goodwill impairment in the second quarter of 2026, triggered by Getty's June 30 announcement, before the termination.

Why is catalog size a weak measure of training-data value?

The count covers approved stills and footage clips, while the training product licenses metadata across images, footage, music and 3D models, and opted-out content can stay in the count. Revenue arrives in a third unit, a blended segment, so the three numbers cannot be converted into one another.

Sources

  • Shutterstock's annual report for 2025 says Data, Distribution, and Services revenue rose 16% to $203.3 million; the segment includes metadata licenses tied to images, footage, music and 3D models used to train AI models, plus Giphy and Shutterstock Studios; contributors supply content non-exclusively with stated exceptions and earn tiered royalties; customer licenses are perpetual, royalty-free and non-exclusive except custom content. Source
  • Shutterstock's amended annual report for 2024 says its collection had grown to 800 million images and 59 million footage clips as of December 31, 2024. Source
  • Shutterstock's first-quarter 2024 earnings release defines content in its collection as approved images (photographs, vectors and illustrations) and footage, in number of clips, in its library at the end of the period. Source
  • Shutterstock's quarterly report for the period ended June 30, 2026 says Data, Distribution, and Services revenue fell 16% to $56.1 million, with no dollar figure for the data offering; Content revenue fell 17% to $165.7 million on weak new customer acquisition; and a $173.7 million non-cash goodwill impairment was recorded in the quarter after the announcement of the terminated merger agreement. Source
  • Shutterstock's Form 8-K dated July 8, 2026 reports that the merger agreement with Getty Images was terminated on July 7, 2026. Source
  • Getty Images' annual report for 2025 says it entered into an agreement and plan of merger with Shutterstock on January 6, 2025, in a merger-of-equals transaction. Source
  • Shutterstock's license agreement prohibits using any visual content as training data for any artificial intelligence, machine learning or generative AI system, tool, process or dataset. Source
  • Shutterstock's contributor help article says dataset inclusion of existing library content is covered under Section 1a of its Contributor Terms of Service, that an opt-out was added in January 2023, and that dataset buyers may only use datasets to train machine learning and computer vision models. Source

Request SourceX investor materials

SourceX shares company and commercial metrics with investors directly. Any offering is made only to verified accredited investors.

See if your company qualifies