OpenAI is expanding its ChatGPT ad pilot into France, Germany and Ireland, putting it on pace to cover ten markets in under a year. The same week, its own VP of media partnerships confirmed on record that publishers with content-licensing deals will not see a cut of that ad revenue. AI Search advertising is scaling faster than any ad platform in recent memory — the payment rail for the publishers underneath it is not scaling at all.

  • OpenAI's ad pilot reportedly reached $100 million in annualized revenue within six weeks and is now expanding to France, Germany, Ireland and Singapore, bringing it to roughly ten markets in under a year.
  • ChatGPT has more than 900 million weekly active users, a distribution base no comparable ad platform has matched at this stage of rollout.
  • OpenAI's VP of media partnerships has confirmed that publishers with licensing deals, including Le Monde and Axel Springer, get zero share of ChatGPT ad revenue.
  • The gap between fast-scaling AI Search demand and a missing publisher payment rail is exactly the opening a native, two-sided ad model needs to fill.

What Generative Engine Advertising actually is

Generative Engine Advertising, or GEA, is the practice of placing paid brand messages inside or beneath answers generated by AI systems like ChatGPT, rather than on a ranked results page. It replaces the auctioned blue link with a placement attached to a synthesized response, and replaces cost-per-click with formats closer to cost-per-view. OpenAI's expansion is the clearest evidence yet that GEA has moved from experiment to scaled category.

For two decades, the open web ran on an exchange: publishers produced content, search engines sent traffic, advertisers paid for clicks on that traffic, and a slice of that spend flowed back to publishers through referral volume and direct placements. OpenAI's rollout keeps the auction and the advertiser but removes the publisher's seat at the table: ChatGPT sells placements against answers built from licensed and crawled content, while paying the sources of that content a flat fee negotiated up front, not a share of what the resulting ads earn.

What Digiday reported, and what we verified

Digiday reported OpenAI is hiring a regional EMEA ads manager plus client-partner and customer-success roles across Paris, Munich and Dublin, extending the ad pilot to France, Germany and Ireland weeks after its U.K. launch. We verified the scale: OpenAI confirmed 900 million weekly active users in February 2026, and industry reporting puts the ad pilot's annualized revenue at $100 million within six weeks of its February launch, with CPMs already compressing from roughly $60 to $25 as self-serve access opens. The U.K., France and Germany are Europe's three largest digital ad markets, per IAB Europe's AdEx Benchmark 2025 report, which put total European digital ad spend at €131 billion in 2025.

Why a licensing fee isn't the same as a revenue share

The case that licensing deals already compensate publishers

OpenAI's position, voiced by VP of media partnerships Varun Shetty at WAN-IFRA's World News Media Congress, is that publishers already captured their value up front: multi-year deals reported in the hundreds of millions with News Corp, Axel Springer, the Financial Times and others paid for training and retrieval rights before a single ad ever ran. Shetty has also argued raw traffic was never the right measure, citing publisher reports that ChatGPT referrals convert and retain better than average search traffic.

Why that argument runs out once ads are in the mix

That argument held while ChatGPT was a cost center. It stops holding once OpenAI sells ads against the same licensed content at CPMs compressing toward self-serve scale. Perplexity's Publishers' Program and Prorata AI's 50%-revenue-share model both concede the point OpenAI won't: once a platform monetizes an answer built on someone else's reporting, the publisher is owed a piece of that specific ad dollar, not just the flat fee negotiated before it existed.

What this means for brands and for publishers

For CMOs, media buyers and agencies in France, Germany and Ireland: the window is open, not permanent

  • Test budgets now, while CPMs are still compressing toward self-serve pricing and before category competition arrives.
  • Treat AI Search placements as their own line item with a native format, not a repurposed search or social budget.
  • Expect thin measurement: OpenAI is still accepting only test budgets in every market it has launched, this one included.

For publishers in the licensed markets: price the next deal on ad revenue, not training access

Le Monde and Axel Springer are licensing partners headquartered in two of the exact markets OpenAI is now selling ads into, a reminder that the next negotiation isn't about training rights, it's about who gets paid when a licensed article underwrites a paid placement. Publishers renegotiating now, while OpenAI is still proving out the ad model, hold more leverage than they will once the format is standardized.

Three signals that show where this goes next

Watch whether any OpenAI publisher licensing renewal adds an ad-revenue clause where the last one didn't, whether Perplexity's or Prorata's revenue-share models gain ground as publishers route around a platform with no payout, and whether CPMs keep falling as self-serve access broadens across the new EMEA markets. Whichever platform standardizes publisher payment first sets the pricing benchmark the rest of the category has to match.

Conclusion

Hold on to this: AI Search advertising is no longer a pilot, it is a scaling ad category, but OpenAI's own confirmation that licensed publishers get no share of that revenue proves the category still has no payment rail on the supply side. Smalk AI's answer is to build the native ad layer for AI agents as a two-sided rail from the start, letting brands buy visibility inside AI answers while opening the revenue stream OpenAI just declined to build for the publishers feeding those answers. Watch whether a major licensing renewal adds a revenue-share clause in the next two quarters — that's the moment the category's pricing model actually gets set.