People Inc.’s Jon Roberts on the AI licensing boom – and the revenue lag

By Sara Guaglione • March 25, 2026 •

Ivy Liu

This article is part of Digiday’s coverage of its Digiday Publishing Summit. More from the series →

The surge in AI content licensing marketplaces has created a new wave of opportunities for publishers — just not much money yet.

“I think there’s a boom in companies who claim they will sell your content for money, and not yet a boom in the money from those deals. So there seems to be more of these marketplaces every minute. But not all of them are going to be meaningful,” Jon Roberts, chief innovation officer at People Inc., said onstage on Monday as the keynote speaker at the Digiday Publishing Summit in Vail, Colorado.

As tech companies develop AI licensing marketplaces for publishers’ content, the number of deals is climbing but the economics remain largely unproven. Roberts broke down the opportunities he’s seeing that could generate real revenue for People Inc. and publishers at scale – and how it could work in their favor.

Different AI licensing deals for different purposes

People Inc. has AI licensing deals with OpenAI, Meta and Microsoft. Roberts put the deals with OpenAI and Meta in one bucket, and the one with Microsoft in another.

“With OpenAI and with Meta, they’re the consumer. They’re the end user. We’re working with their AI labs. They know that the better the inputs they have, the better the outputs they’re going to have. There’s some amount of training encompassed in that. They’re multi-year deals and it’s all you can eat,” he said. “They have full access to all the content that falls within the scope of the contract. So for them, we’re providing the information in a way that’s more direct, more useful, [and] faster to get at than crawling.”

But the Microsoft marketplace is a pay-per-use model, surfacing content (and paying for it) when it’s requested.

AI bot blocking is key to developing a marketplace

A strong AI bot-block strategy is the backbone of developing a marketplace for content licensing, according to Roberts. People Inc. blocks all bots, unless they’re on a “permissioned list,” he said.

“That has resulted in us blocking [tens] of millions of unauthorized attempts to crawl our content every day. So there is a large and growing appetite for access to content in real time that is exploding. Now, when you’ve got that much appetite, there’s a market.”

People Inc. is now allowing some bots through to give these marketplaces access to its content and get a better understanding of the economics. “The big question, of course, is what is the economics of this going to be? Will it be meaningful? How will it evolve?” he said.

A few publishers solving the problems of many?

Execs at publishers smaller than People Inc. have told Digiday they are trying to figure out how to get into the AI licensing space, but are struggling to get through the door and get the attention of large AI and tech companies.

Roberts said this is one of the reasons People Inc – and other large publishers – are participating in the development of new standards for AI content licensing between publishers and AI systems, such as IAB Tech Lab’s Content Monetization Protocols and Really Simple Licensing.

“The goal of [AI licensing] marketplaces is to be able to tackle this problem at web scale. If we haven’t solved for the whole web, then we haven’t done anything. So that’s what we’re attempting to build with them,” Roberts said.

Amazon, Google next to develop AI licensing marketplaces?

Roberts declined to confirm People Inc. was in talks with Amazon about an AI licensing marketplace it is reportedly developing, first repoted by The Information. But he did say that Microsoft, Amazon and Google’s cloud platforms would be the “obvious” places to build these systems.

“The joy of not talking about non-public deals is we don’t talk about them,” Roberts said. “There are a lot of things in progress, because there are new marketplaces popping up all the time… [Microsoft’s cloud platform] Azure has a play that says they will pay for the quality of IP, or they believe people should be paid for the quality of their IP. [Amazon’s cloud platform] AWS may or may not be launching something that would structure the same thing. And then the third large place where AI workloads are run is [Google Cloud Platform], who have been very silent on that kind of structure.”

The complicated economics of AI licensing marketplaces

Publishers and tech companies are working on creating new compensation structures to match the value of the content publishers are allowing AI systems to access, called “pay by demonstrated value” or “pay per value.”

It’s a complicated process that involves many factors that can influence pricing, such as whether the content is being used to train an LLM or how often a publisher’s content is cited or referenced in AI-generated answers.

But Roberts argued there is already a “rate card” on which these transactions could be modeled.

“OpenAI will charge 10 bucks per 1,000 searches, or 25 for premium [use]. If you say the amount of AI need for your content is going to be 10x the human need – [which could be] a potential future – and you were getting a $15 CPM for every time it read your content, you can all do the math and say, ‘That’s interesting.’” Roberts said.

Microsoft’s marketplace won’t charge flat pricing for content licensing because that would encourage clickbait, or volume over quantity, Roberts said.

He stressed that if publishers can provide clean, structured, rights-cleared content in a fast and efficient system, AI outputs would be cheaper to run and higher quality – an incentive for tech companies to participate in these marketplaces.

“That’s why we’re trying to support the premium markets that work. But if you’re just trying to add a tax to a system, nobody will buy it. If you build something better, then it’s dumb if they don’t buy it. And I think that this year, we’re on the verge of making that switch,” Roberts said.