Template and Asset Businesses in the AI Era: Which Digital Asset Categories Are Growing (And Which Are Dying)

TL;DR. Almost every claim about whether digital asset businesses still work is anecdote. It does not have to be. Two of the largest asset marketplaces on earth, Shutterstock and Getty Images, are public companies that file audited revenue by product line and publish operating metrics every quarter. We read the FY2025 annual reports of both, filed in February and March 2026, plus the second-quarter filings from August 2026, and built a ratio neither company reports: paid downloads per 1,000 images in the catalogue. At Getty it falls in every consecutive period we can measure, from 177.6 in FY2023 to 144.0 in the twelve months to June 2026, a decline of 18.9%, while the catalogue grew 16.8% over the same span. That is the whole thesis of the asset business, that more assets produce more income, failing in audited numbers. Then the obvious escape route, licensing your archive to AI companies, turns out to be a trap of its own: in the first half of 2026 that revenue line fell 27.8% at Shutterstock and 44.5% at Getty. Only one line grew, at 10.1%, and it is the one nobody can generate. Below: what both filings show, the four tests we derived from them, and what actually still sells.

This piece sits next to our work on pricing digital products in the AI era and the real maintenance cost of passive income. Those ask what to charge and what it costs to keep. This one asks the prior question: whether the category itself is still standing.

Why these two companies answer the question

If you sell templates, presets, stock photography, icon sets, WordPress themes, Notion dashboards, Figma kits, LUTs, sound packs or 3D models, you are in the same economic category as Shutterstock and Getty Images. You have a catalogue, buyers arrive through search, and you get paid per licence or per subscription. The difference is scale, not structure.

The relevance is more direct than that. In July 2024 Shutterstock acquired Envato, the company behind ThemeForest, CodeCanyon, Envato Elements, GraphicRiver, VideoHive and Placeit. Shutterstock’s own annual report describes the purpose of that acquisition in six words: Envato “enhances digital creative assets and templates.” The template marketplace where a large share of the world’s independent asset sellers actually list is now a line inside a public company’s audited accounts. That is unusual and it is worth using.

Getty Images is the counterweight. It sells the same commodity, licensable visual content, but with a different rights posture: exclusive contributors, staff photographers, event coverage, an archive of over 150 million images. Where the two companies agree, the finding is structural rather than a story about one management team.

The metric nobody publishes: yield per asset

Getty discloses two numbers in the same table every quarter: the size of its image collection and its paid download volume. It never divides one by the other. We did.

Original CEOtudent analysis. Asset yield at Getty Images, derived from disclosed collection size and paid download volume

Period	Paid downloads	Image collection	Downloads per 1,000 images	Change
FY2023	95 million	535 million	177.6	–
FY2024	93 million	572 million	162.6	-8.4%
FY2025	92 million	609 million	151.1	-7.1%
LTM to 30 Jun 2026	90 million	625 million	144.0	-4.7%

Four consecutive periods, four consecutive declines, total -18.9%. Over the same span the catalogue grew from 535 million to 625 million images, up 16.8%, and demand did not follow. Two caveats we state rather than hide: the denominator counts images while the numerator includes video downloads, so the true per-asset yield is lower than shown, not higher; and Getty’s download figure excludes editorial subscriptions, editorial feeds and certain bulk API deals, which is why we treat this as a trend line rather than an absolute. The direction is not ambiguous in either case.

The customer side of the same table confirms it. Getty’s last-twelve-month purchasing customers fell from 799,000 in FY2023 to 636,000 by June 2026, down 20.4%. Its annual subscriber revenue retention rate, which was 92.4% in FY2023, was 88.4% in the twelve months to June 2026. Shutterstock’s half-year filing shows the same shape: paid downloads down 13.1% year over year, subscribers down from 1,073,000 to 951,000.

If you have felt your own asset listings going quieter while your catalogue got bigger, this is what that feels like from inside two billion-dollar companies.

Which lines grew, and which did not

Both companies break revenue into product lines, and the lines behave completely differently. We indexed each to 2023 = 100 so they can be read together. Every figure below is as reported in the FY2025 annual report of each company; the line totals reconcile exactly to each company’s reported total revenue, which is our check that we transcribed them correctly.

Original CEOtudent analysis. Revenue by product line indexed to FY2023 = 100

Company and line	What it is	FY2023	FY2024	FY2025	2-year change
Shutterstock, Content	Images, footage, music, 3D, templates (incl. Envato)	100	103.1	106.7	+6.7%
Shutterstock, Data / Distribution / Services	Metadata for AI training, Giphy, Studios	100	127.6	148.0	+48.0%
Getty, Creative	Commercial stock stills, video, generative AI	100	95.5	96.2	-3.8%
Getty, Editorial	News, sport, entertainment, archive	100	107.9	115.3	+15.3%
Getty, Other	Data licensing, music, asset management, print	100	236.0	319.0	+219.0%

Group the two companies by what the asset actually is rather than by who sells it, and the split is stark. Commodity creative supply, meaning Shutterstock Content plus Getty Creative, grew 2.1% in two years. Everything rights-based, data-based or service-based grew 32.1%. That is a 15.4x difference in growth rate between the two halves of the same industry.

And Shutterstock’s headline Content growth of 6.7% is not what it looks like. The company’s own risk disclosure says it plainly: over the last several years, revenue from its organic Content business has declined, and total revenue grew because of Content business acquisitions and growth in the data and services business. In other words the template and stock line grew by buying Envato, not by selling more assets. Strip acquisitions out and the largest asset marketplace in the world is shrinking in its core product.

The complication that changes the advice

Here is where the easy story breaks, and it is the reason this piece exists.

The obvious lesson from the table above is: stop selling assets one at a time and license your archive to AI companies instead. Getty’s “Other” line tripled. Shutterstock’s data offering, described in its filing as metadata “used to train AI models,” is inside the line that grew 48%. Every commentary piece written in 2025 reached that conclusion.

Then the first half of 2026 arrived.

Verified data. First half 2026 versus first half 2025, as reported

Line	H1 2025	H1 2026	Change
Shutterstock, Content	$402.7m	$343.8m	-14.6%
Shutterstock, Data / Distribution / Services	$106.9m	$77.2m	-27.8%
Getty, Creative	$263.0m	$253.6m	-3.6%
Getty, Other (incl. data licensing)	$25.0m	$13.9m	-44.5%
Getty, Editorial	$171.0m	$188.2m	+10.1%

The AI data-licensing line fell harder than anything else at both companies, independently, in the same six months. Both filings explain why in almost identical language: data access and licensing agreements typically result in a greater portion of revenue being recognised in an accelerated manner. Translated: these are large one-off deals, booked in a lump, and a period without a signing looks like a collapse. It is deal revenue wearing the costume of a business line.

One line grew. Getty Editorial, up 10.1%, on top of 6.9% the year before and 7.9% the year before that. Editorial is news, sport, entertainment and archive: content produced by staff photographers at 160,000 events a year, from 75 exclusive editorial partners, that legally cannot be generated because the event happened once and someone with credentials was standing there. Over half of it, 53.5% in FY2025, is sold through annual subscriptions rather than per download.

That is the finding. Across two companies, four years and two independent reporting streams, the durable line is not the one AI cannot make well. It is the one AI cannot make at all, sold on a recurring relationship rather than a transaction.

The four-test asset durability framework

We built this from the pattern in the filings rather than from opinion. Score your own category out of four.

CEOtudent editorial framework. The asset durability test

Test	The question	Passes	Fails
1. Generation gap	Can a competent model produce an acceptable substitute in under five minutes?	Event photography, verified datasets, licensed brand assets, anything requiring access or credentials	Generic stock imagery, icon sets, basic templates, background music
2. Rights position	Do you hold something a buyer cannot obtain elsewhere at any price?	Exclusivity, model and property releases, indemnification, a licence a legal team will accept	Non-exclusive listings on a marketplace that hosts hundreds of thousands of other contributors
3. Relationship shape	Does revenue arrive on a renewal or per unit?	Annual access, memberships, update commitments	One-off downloads with no return visit
4. Yield direction	Is revenue per asset in your catalogue rising or falling year over year?	Rising, or flat with a growing catalogue	Falling while you add more, which is the Getty pattern above

Two or fewer passes means you are in the commodity half of the industry, the half that grew 2.1% in two years while its customer count fell by a fifth. Three or four means you are in the half that grew 32%.

The uncomfortable implication for most independent sellers is that test 1 and test 2 are usually decided before you start work. You cannot add a rights position to a Notion template after the fact. You choose it at the category level, which is exactly why this is a business-model question and not a marketing question.

What this means if you are the one-person version

Stop optimising catalogue size. The Getty numbers are the clearest available evidence that the volume strategy has stopped working at scale. Adding your 200th template into a falling-yield category is not a growth plan, and it is a real cost in maintenance, which we cost out in the passive income maintenance piece.

Move the money to renewal. Getty’s growing line is majority subscription. Shutterstock’s average revenue per customer rose 9.8% in the year to June 2026 even as subscriber count fell 11.4%, which is what it looks like when fewer, deeper relationships replace many shallow ones. Fewer buyers paying more on a schedule beats more buyers paying once.

Do not build a business on AI licensing deals. A revenue line that falls 28% and 45% in one half-year at two different companies is not an income stream you can plan around. Treat any such deal as a windfall against a business that already works.

Compete on access, not on output quality. The generation gap test is about what you can reach, not how well you execute. Proprietary data, permission, credentials, an installed base, a legal position. Our framework for turning that into products is in licensing your knowledge, and the first-sale problem is handled in the cold start problem.

Read your own yield curve every quarter. Divide revenue by the number of live assets in your catalogue. If that number falls three periods running while you keep publishing, you have Getty’s problem, and no amount of additional output fixes it.

Where a CEO and a student read this differently

A CEO reads the table and asks which line to fund. That is the right first question, and the answer is the rights-and-renewal half.

A student reads it and asks what changed underneath, because the same shift will happen again in another category. What changed is that the cost of producing an acceptable asset fell toward zero while the cost of holding a rights position did not move. Every asset category is now sorted by that single test, and the sorting is not finished. The discipline is to run test 1 against your own category every few months rather than to assume today’s answer holds.

Frequently asked questions

Are template and asset businesses dead?
No. Half of the industry, measured by revenue line, is growing at 32% over two years. The commodity half grew 2.1% with a customer count falling 20%. The category is splitting, not dying, and which half you are in is decided by rights position and relationship shape rather than by effort.

Does the Getty download decline just mean people buy subscriptions instead?
Partly, and that is why we show customer counts alongside. Purchasing customers fell 20.4% and annual subscribers fell 23.6% from their FY2024 peak, so the buyer base itself contracted rather than simply changing how it pays. Subscriber revenue retention also fell four percentage points to 88.4%.

Should I license my work to AI companies?
Take the money if offered, plan nothing on it. Both public companies show that line falling 28% and 45% year over year in the same six-month period, because it is lumpy one-off deals recognised in a lump, not recurring revenue. Neither filing describes it as predictable.

Is generative AI actually the cause?
The filings do not make a causal claim and neither will we. What they show is timing and shape: falling yield per asset, falling customer counts, a collapsing commodity line and a growing exclusive line, all in the years generation costs fell. Both companies list AI among their principal risks, and both now sell generative content themselves.

What is the single most useful number here?
Downloads per 1,000 catalogue images: 177.6, 162.6, 151.1, 144.0. It is the volume strategy failing in audited data, four periods in a row, and you can compute the equivalent for your own catalogue in about ten minutes.

Does this apply to code and software templates too?
Envato’s ThemeForest and CodeCanyon sit inside Shutterstock’s Content line, the line whose organic revenue the company says has been declining. That is the only audited evidence available on code templates specifically, and it points the same way as the imagery evidence.

Sources

• Shutterstock, Inc., Annual Report on Form 10-K for the fiscal year ended 31 December 2025, filed with the United States Securities and Exchange Commission on 17 February 2026. Revenue by product offering; key operating metrics; risk factors on organic Content revenue; description of the Envato and Giphy acquisitions and the data offering.
• Shutterstock, Inc., Quarterly Report on Form 10-Q for the quarter ended 30 June 2026, filed 7 August 2026. Revenues by product offering and key operating metrics for the six months ended 30 June 2026 and 2025.
• Getty Images Holdings, Inc., Annual Report on Form 10-K for the fiscal year ended 31 December 2025, filed with the United States Securities and Exchange Commission on 16 March 2026. Revenue by product; key performance indicators including image collection, paid download volume, purchasing customers and annual subscriber revenue retention rate; product line descriptions.
• Getty Images Holdings, Inc., Quarterly Report on Form 10-Q for the quarter ended 30 June 2026, filed 10 August 2026. Revenue by major product and last-twelve-month key performance indicators.
• World Economic Forum, Future of Jobs Report 2025, on the reallocation of task value as generation costs fall.

---

This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.

This post is also available in: