Apple in Talks With Publishers on AI Siri Content Licensing Deals

Full analysis

Your draft

Apple is reportedly negotiating multiyear deals to pay news publishers when their content feeds answers from a revamped, AI-powered Siri, on a usage basis: you get paid based on how often Siri draws on your stuff. For ad-tech and media operators, the question is whether this is a new revenue line worth chasing, a distraction, or the first crack in a two-tier publisher market.

This is a Type 2 decision for most operators. Nobody has to commit anything yet. What's actually being decided, for a publisher exec, is whether to spend legal and BD bandwidth pursuing a pilot slot now, or wait for the reference deal to set terms. Forcing function: the WSJ report, one source, no named deal, no dollar figure. Slow burn, not a fire drill.

The Market Analyst. Strip the AI glamour and this is a content-licensing deal, not an ad-tech deal. No inventory, no auction, no DSP or SSP (the buy-side and sell-side software that runs programmatic ads) in the loop. For a generalist: Apple wants to rent news text to feed Siri, and pay per use. The exposed cohort is pure-play digital publishers whose revenue depends on people landing on their pages and seeing ads. If Siri answers the question, the pageview never happens, and the CPM (the price per thousand ad impressions) never gets earned. Retail media and CTV don't care. The named publisher here is the New York Times, which is exactly the kind of premium brand Apple wants and the kind least desperate to sign. That tells you something about pricing power.

The Skeptic. Siri has been a punchline for years, and a revamp announcement is not a shipped product. For this licensing money to matter, Siri has to become a high-frequency query surface, which means closing a capability gap with ChatGPT, Gemini, and Perplexity all at once. The usage-based payout sounds publisher-friendly and will almost certainly clear at single-digit CPM equivalents. Ask why OpenAI and Anthropic aren't doing this at scale. Because the economics only work if you're buying goodwill and legal cover, not content. For a generalist: Apple may be paying for permission to avoid lawsuits, not paying because the news is worth it.

The Operator. The usage-based model is a data reconciliation nightmare. Publishers have zero independent audit of Siri's query logs, so "how often your content was drawn upon" is whatever Apple's dashboard says. You are being asked to trust the counterparty's meter. That is the whole game. Any publisher consortium worth its retainer demands third-party measurement before signing, and that demand stalls every deal for quarters. The near-term cost is bandwidth: the BD and legal hours pulled into an Apple pilot are hours not spent on yield optimization during the upfront. For a generalist: you're negotiating to get paid by a machine whose scorecard only Apple can read.

The CFO. Run the math before the romance. A usage-based royalty on an assistant that today handles a rounding error of informational queries is a rounding error of revenue. The real cost is the traffic you cede to get it. Every query Siri answers with your licensed text is a pageview, and its ad dollars, gone. So you're potentially paying to accelerate your own traffic decline in exchange for a royalty that doesn't cover the lost CPMs. This only pencils if you believe the traffic is leaving anyway and the royalty is found money. For a generalist: you might be trading a dollar of ad revenue for a dime of licensing.

The Strategist. Apple's play is a legally clean content moat. Buy a licensed corpus that Google and OpenAI can't cheaply replicate without litigation, and use it to power privacy-first, on-device AI. Content is the fuel, not the product. The two-to-three year consequence is a split publisher market: a licensed tier that trades traffic upside for stable royalties, and an unlicensed tier that gets crawled anyway and earns nothing. For a generalist: the publishers who sign early set the price, and everyone else lives with it.

Where they part ways

The real disagreements are three.

First, is Siri ever a real query surface? The Strategist's whole thesis needs it to be. The Skeptic says five years of evidence say no. If Siri stays a punchline, the licensing revenue is noise and the moat is a press release.

Second, royalty versus cannibalization. The CFO says the royalty may not cover the ad dollars you lose when Siri answers instead of sending a click. The Strategist counters that the traffic is leaving regardless, so any royalty is upside. These can't both be right for the same publisher, and which one is true depends entirely on your current traffic mix.

Third, who holds the meter. The Operator's point is the one that decides deal terms: usage-based pay only works if usage is auditable, and right now it isn't. Apple counts, Apple pays, Apple wins the disputes.

What it hinges on

Two beliefs. One, whether Apple ships a Siri that people actually query at volume. Two, whether publishers extract independent measurement of that usage before signing. If both go the publisher's way, this is a modest but real new revenue line. If either fails, it's goodwill money and legal cover, priced accordingly.

The council leans skeptical on near-term dollars and slightly favorable on long-term structure. The licensed-versus-unlicensed split is the durable idea here, and it doesn't need Siri to succeed to start forming. Before any operator commits bandwidth: pin down the audit mechanism, model the traffic you'd cede, and treat the first published deal terms as the ceiling, not the floor.

The call

Prediction: The first Apple-Siri publisher licensing deal to become public will carry usage-based terms with no independently audited usage metric, and it will land at royalty levels that do not offset the pageview and ad revenue those publishers lose to Siri answers, keeping most large publishers on the sidelines through the 2027 upfront (May to June 2027).

Confidence: Medium. The meter problem is structural, but Apple could concede an audit to close the marquee NYT deal.

Why: Usage-based pay requires a usage count, and only Apple can see Siri's query logs, so the metric is Apple's word unless publishers force a third-party auditor into the contract, which stalls deals for quarters. Apple's own precedent and Google's prior publisher licensing deals have consistently delivered underwhelming dollars, and there is no reason a per-query royalty on a still-unproven assistant clears higher. The opposite outcome, a fat audited royalty that beats lost ad revenue, would require Apple to both concede its data advantage and overpay, neither of which fits a company buying legal cover rather than content.

Revisit by 2027-06-30: We're right if the first public Apple-Siri publisher deal uses Apple-reported usage with no third-party audit, and no top-10 US news publisher outside the pilot cohort has signed. We're wrong if a signed deal includes an independent usage auditor or a majority of top-10 publishers join by the 2027 upfront.