The scarcest input in AI is $CULTURE.
Mather Economics just put a number on what AI should pay for the human work it learns from. The question now is whether the people who make culture are counted in that number, or left out of it again.
The market for human creativity now has a price tag.
This month Mather Economics released The Price of Intelligence, the first independent study of what collective licensing for AI is worth. Its answer: AI firms have an economic reason to pay 15 to 35 percent of their revenue to keep access to licensed content. Mather started the work on its own, and the Internet Licensing Collective neither commissioned nor paid for it.1
“Content is the third essential input to AI, alongside compute and power.”Matt Lindsay, PhD · President & Chief Economist, Mather Economics2
Compute and power trade on open markets with visible prices. Content does not, and culture least of all. This note sets out what the report proves, where culture still falls outside it, and the direction I believe we have to take: a sustainable culture economy where millions of Culture Makers and their communities are paid for the value of their work, and AI companies license the culture they need to innovate at scale.
Paying for content is good economics, not charity.
Mather measures what an AI firm would lose without licensed content, then treats that loss as the most it would rationally pay. Three channels drive the value: training data that improves the model, retrieval that improves every answer, and legal certainty that unlocks enterprise buyers. Enterprise is where most revenue sits, between 40 and 80 percent at the leading firms.1
On a conservative $100 billion baseline for OpenAI and Anthropic, the estimate works out to $15 to $35 billion in annual content value. If that revenue grows 50 percent a year, licensing passes the entire 2025 U.S. display advertising market, $81.6 billion, within three to five years.1
The report is careful to say those mature rates are not the right rate for AI. Its point is narrower and stronger: when a platform makes money on third-party work at scale, the market norm is never zero. AI firms have already conceded as much. More than 90 licensing agreements have been announced, with disclosed payments in the hundreds of millions a year.1
Mather also reaches back to the precedent MADE CX has always pointed to. A 1909 law gave songwriters a performance right, but nobody could collect on it until ASCAP introduced the blanket license in 1914. Then BMI, founded in 1939, widened the table to blues, country and jazz creators. The model reached nearly every working songwriter once it let in the people the first collective had left out.1
A price for the web is not a price for culture.
Mather builds on the Coase theorem: a market works when property rights are clear and the cost of making a deal is low. Its diagnosis of web content is that both conditions are missing. For culture, a third piece is missing too. Nobody has agreed on what a cultural work is worth before it is used.1
The study models a publisher-anchored collective, and it says plainly that such a collective will not cover everything AI trains on or retrieves. Visual art, photography, voice and likeness, oral history, family archives and community IP sit outside a publisher crawl and outside any price list. That is where a large share of Black cultural production lives.
The report's own logic says this is the content AI can least afford to lose. It finds willingness to pay is highest for work that is original, verified and hard to substitute, and it expects that value to rise as public human text runs out around 2028. A sound, a style or a lineage cannot be synthesized from scratch. Culture is the least substitutable input in the stack.1
A license only shields an AI firm if it can show where an answer came from. Research cited in the report found that 50 to 90 percent of LLM responses are not fully supported by the sources they cite.1 For culture, provenance is not a feature. It is the title itself.
ASCAP proved one license could pay a nation of songwriters. The question for this era is who gets in the door.
Five moves to build a sustainable culture economy.
A culture economy is infrastructure, not a settlement. Mather says the same of content licensing as a whole: a way to cut transaction costs, steady supply and keep people producing.1 Here is how it gets built for culture.
Title the work
Every Culture Maker, estate and archive registers what they hold, with identity and chain of custody on record. Title turns a contribution into a counterparty.
SolvesUnclear property rightsPrice it before it is used
Collectives can allocate money after use. Culture also needs a rating before use, so its makers negotiate from a valuation instead of a crawl count. That is the job of the Cultural Property Rights Standard (CPRS).
SolvesNo agreed valueLicense once, at scale
One Cultural Use License across a rated repertoire replaces thousands of one-off negotiations, with costs that scale with use.
SolvesHigh transaction costsPay the maker and the community
Originators receive 80 percent of every license. Four percent returns to their communities through the CX Foundation. The balance funds valuation, custody, royalty collection and enforcement.
SolvesValue that never flows backGovern it together
Membership is free and non-exclusive, makers keep their copyright, and the people whose work is licensed help set the payment rules. One member, one vote.
SolvesTrust
Licensed culture is a better input, not just a fairer one.
Mather lists what a working license market gives AI firms. Each of those benefits has a cultural counterpart, and each one depends on title.
| What AI firms need | What the report finds | What a culture economy supplies |
|---|---|---|
| Lower deal costs | Thousands of bilateral talks block valuable agreements | One Cultural Use License across a rated repertoire |
| Legal certainty | Unresolved rights suppress enterprise adoption | Registered title and a documented chain of custody |
| Quality supply | Verified, structured content cuts cleaning and filtering costs | Rated, verified works with known origin |
| Provenance | Citation errors weaken the license as a shield | An identity record on every licensed work |
| Future supply | Unpaid creators stop producing as human data runs out | Culture Makers who are paid keep making culture |
This infrastructure is already running. MADE CX has rated more than $138 million in assessed cultural value and executed 32 Cultural Use Licenses, including one with Microsoft MADE CX platform data.
Introducing the CPRS Collective.
On November 12, the nonprofit Internet Licensing Collective (ILC, formerly the RSL Collective) launched with more than 3,000 media brands, content owners and publishing platforms, and signed Yahoo Scout as its first commercial licensee.4 It is the clearest proof yet that AI will pay a collective for content, and that work deserves support.
| Term | ILC at launch | What culture still needs |
|---|---|---|
| License | One blanket license across billions of web pages; costs scale with use and growth | A repertoire beyond the web: art, archives, likeness, oral history, community IP |
| Uses | Licensees choose training, RAG, grounding, agentic applications or AI overviews | Use terms for style, voice and likeness |
| Membership | Free and non-exclusive; publishers pick their licenses and keep direct deals | The same, for Culture Makers, estates and institutions |
| Payment | Allocated by value contributed: how content is used, the value of the AI product and how often it is relied on | A rating that sets value before use, and a share for the community |
| Cost | Nonprofit; operating costs projected at or below 10%, the rest to rights holders | Valuation, custody, collection and enforcement for works with no publisher behind them |
Table 2 · ILC terms as announced November 12, 2026. Financial terms of the Yahoo Scout agreement were not disclosed.4
Culture needs a table of its own beside it, where the people who make culture set its terms.
To create a sustainable culture economy for the AI era, where millions of Culture Makers and their communities are paid for the value of their work, and AI companies license the culture they need to innovate at scale.
The CPRS Collective is the public, member-led home for the standard. It brings Culture Makers, museums and archives, family estates, institutions and brands into one conversation about how culture is priced in the AI era, and it runs on infrastructure that is already live at made.cx.
Founding Member
Culture Makers, estates and professionals who sign the CPRS pledge.
Founding Institution
Museums, archives, HBCUs and foundations ready to bring collections to the table.
AI & Brands
One Cultural Use License across the rated repertoire, with clear provenance.
Founding seats come first. Joining is free and non-exclusive, and you keep every right to your work. Until then, the most useful thing any Culture Maker can do is title what they hold.
Sources.
- Lindsay, M., Hart, N., and Stalvey, B. The Price of Intelligence: Sizing the AI Content Licensing Opportunity and How Collective Licensing Can Unlock It. Mather Economics, November 2026. Figures cited: willingness to pay of 15–35% of revenue; $100B baseline; display advertising comparison; platform revenue shares; 90+ announced agreements; data exhaustion 2026–2032, median ~2028; enterprise share 40–80%; citation accuracy; ASCAP and BMI history; authors' note on independence.
- Mather Economics, announcement of The Price of Intelligence, November 2026.
- MADE CX platform data: assessed value and executed Cultural Use Licenses as of November 2026. CPRS figures are modeled valuation bands, not appraisals, prices or investment advice.
- Internet Licensing Collective, “The Internet Licensing Collective (ILC) Launches with Yahoo Scout as First Commercial Partner,” GlobeNewswire, November 12, 2026. internetcollective.org.