USA Today Co. Made More From AI Licensing Than Display Ads in Q1

USA Today Co. reported $33.75 million in "other" digital revenue for Q1 2026, a bucket that includes AI licensing deals with Meta, Microsoft, and Perplexity. That figure grew 125.6% year over year and now sits ahead of the company's display ad revenue, which fell on softer page views and weaker programmatic pricing. The implication for buyers is the part that matters: licensed pages get pulled from open exchanges, and CPMs tighten on what's left.

The bucket that just outran display

Sit with the full quarter print for a second. According to PPC Land's writeup, total revenue was $548.5 million, down 4.0% year over year. Digital revenue: $261.9 million, an all-time high at 47.8% of the mix. Inside that, the "Digital Other" bucket (AI licensing, syndication, affiliate, content partnerships) hit roughly $33.75 million and grew 125.6%. Display ads, the line item that has historically anchored the digital story for USA Today, were down about 3% on the same comparison.

The company doesn't break out AI licensing as its own line. So I want to be careful here. AI dollars aren't isolated in that figure, and affiliate plus syndication revenue are real numbers in the same bucket. But Digiday's writeup of the call quotes CEO Mike Reed saying these deals had a "notable impact" on the quarter. Against a stagnant affiliate market and a 125.6% growth rate on a roughly $15M base, the AI deals are doing most of the lifting.

Adjusted EBITDA grew 44.7% to $73.1 million. EBITDA margin moved from 8.8% to 13.3%. Net income flipped from a $7.3 million loss to $19.9 million in profit. This isn't a marginal accounting story.

For the first time, an American newspaper company made more from licensing pages to AI buyers than from selling display ads against them in the same quarter.

Meta, Microsoft, Perplexity, in that order

The deal stack matters because each one prices a different surface.

The Perplexity announcement landed July 30, 2025. Gannett joined the Perplexity Publisher Program with revenue share from advertising, plus access to Perplexity's data analytics and Sonar API. Financial terms weren't disclosed. Microsoft followed with a deal announced alongside Q3 2025 earnings on October 30. Meta has been licensing Gannett content for longer.

In February 2026, Microsoft launched its Publisher Content Marketplace with USA Today, Hearst Magazines, Vox Media, AP, Business Insider, and Condé Nast as launch partners. Yahoo signed on as a buy-side participant on a CPM-share model. Translation: when Copilot grounds an answer in a USA Today article, the publisher gets paid per impression.

The thing none of these deals do well, from a media buyer's perspective, is sit cleanly inside your media plan.

What it does to your open exchange buys

Here's the part most planners I talk to are still under-weighting. When a top-30 US news publisher routes premium content into AI licensing channels, the practical effect on programmatic looks like this:

• Page views on the publisher's owned properties stay flat or decline (Reed cited "softness in page views" on the Q1 call).
• Sessions that would have served display impressions don't happen, because the user got the answer in Copilot or Comet first.
• The publisher's incentive to chase open exchange CPMs softens, since AI licensing carries higher margin.
• Premium URLs increasingly get reserved for direct deals and PMPs, not the open auction.

The Guardian US already showed the pattern. ExchangeWire's reporting on 2026 publisher dynamics cited a 44% YoY increase in programmatic revenue at the Guardian US driven mostly by higher CPMs in private marketplace deals, not traffic. Same direction: the money moves toward gated channels.

If you're buying scaled news inventory through open exchanges, you're buying a shrinking pool. From what I've seen, the contraction shows up first as fewer impressions on premium domains in your post-buy reports, and then as a 10-15% CPM uptick across the publisher properties that haven't licensed yet, because the supply curve just shifted. Whether your DSP flags any of this depends on the verification stack you're using, and honestly, most of them are still reporting it as a CPM win rather than a supply contraction.

There's a second-order issue most planners miss. AI licensing deals are typically priced on training value or grounding rights, not on the same audience characteristics that drive programmatic CPMs. So the publisher gets paid more for the same article whether or not the user behind the AI query would have ever clicked an ad. That decoupling is what makes the licensing curve so dangerous to model around. The relationship between premium content supply and ad-supported demand has been roughly stable for two decades. It isn't anymore.

Worth noting: USA Today Co. also reported digital-only subscription growth of 6.2%, with ARPU at a record $10.30, even as A Media Operator covered the company's earlier 15% drop in digital subscribers tied to a price-and-margin pivot. Translation: the company is willing to trade audience scale for unit economics, which is exactly the posture you'd expect from a business shifting weight onto licensing. The audience the AI buyers want is the editorial output, not the engaged session.

This is a different shape of problem than the Mondelez story we covered, where brands paid to unblock crawlers. Publishers and brands are both paying to participate in the AI traffic loop. Just on different sides of the meter.

The Q2 caveat buyers should mark

CFO Trisha Gosser told the call that Q2 EBITDA growth would come at "a notably more moderate pace" than Q1's 45%, and that AI licensing would contribute less to the print. Reed himself called the licensing curve "lumpy or unpredictable" in the near term.

That's the honest read. One quarter doesn't make a category. AI licensing revenue arrives in chunks tied to deal cycles, contract structure, and milestone payments, not a steady run-rate. If you're modeling open-exchange supply contraction off this single print, you're probably overshooting.

But the direction is real. USA Today Co. is one of seven launch partners in the Microsoft Publisher Content Marketplace. Hearst, Vox, AP, Condé Nast, and Business Insider are all on the same revenue curve. Stack them up over the next four quarters and the 125.6% growth rate isn't an outlier, it's a market.

Three moves before your next IO renewal

Three things worth running this week:

1. Pull your top 20 publisher domains from the last 90 days of programmatic spend. Cross-reference against Digiday's tracker of publisher AI licensing deals. Mark the ones with active AI agreements.
2. For those domains, request an inventory roadmap from the rep. Ask specifically what percentage of premium URLs they expect to route through PMP versus open auction in H2 2026. If the rep doesn't know, that's its own answer.
3. Reset your programmatic CPM benchmarks for the long-tail news inventory you'll get pushed into. I'd budget for a 10-15% CPM uptick on what remains, plus a 20-30% reduction in scaled premium impressions on the licensed properties. Adjust your reach goals accordingly.

The publishers who built audiences for the open web are now selling the audience signal directly to AI buyers. The ad-supported version of those same audiences is shrinking on the back end. Two doors, same building, and one of them is starting to close.

Notice Me Senpai Editorial