The New York Times expects AI content marketplaces to need years before they can value publisher content in real time for AI training or inference, according to Adam Greenberg, the company's vice president of strategic partnerships.

Speaking onstage at the Digiday Publishing Summit in Miami last week, Greenberg said the AI content licensing ecosystem has improved for publishers in recent years. AI companies now increasingly expect to compensate publishers for their content, or at least give them the ability to opt out of its use, he said.

But Greenberg was blunt about the current state of the market. He called the emerging AI marketplaces "underdeveloped."

His skepticism is grounded in experience. Greenberg spent 14 years at Google building ecosystems including AMP and Web Stories. That background shaped his view of how long these marketplaces take to mature.

"I think where we end up is very unclear," Greenberg said. "It's very hard to get these things going. It takes years, and so a marketplace that is valuing content in real time for use in AI training or inference is just going to take a long time."

His comments point to a structural problem for publishers negotiating with AI firms today. Without a functioning marketplace that prices content dynamically, publishers and AI companies must strike individual licensing deals in a market where the long-term value of content in AI applications remains uncertain.

Greenberg's remarks also signal how the balance of expectations has shifted since the first wave of AI scraping controversies. Compensation and opt-out mechanisms have moved from publisher demands toward baseline expectations among AI companies, at least at leading publishers like The Times.

The gap between improved negotiating conditions and a mature marketplace leaves publishers in a transitional period. Greenberg's forecast suggests publishers should plan for years of bilateral dealmaking before real-time content valuation becomes practical.