# AI Content Pricing Power Comes From Bidders, Not Coalitions *Publishers finally have AI buyers but no say on price. The open web solved this once, and it wasn't with a coalition. It was with more bidders.* Published: 2026-10-03 | Read time: 5 min read | Author: Smalk AI Research | Source: https://www.smalk.ai/blog/ai-content-pricing-power-bidders-not-coalitions --- Publishers finally have buyers for their content in AI search. What they lack is AI content pricing power, and Fast Company's October 2, 2026 analysis shows why: every buyer at the table names its own price. The open web faced this exact problem a decade ago. It did not solve it with a coalition. It solved it with more bidders. - AI content is now bought by AI platforms, data brokers and new exchanges, but in every channel the buyer sets the price. - Publisher collective action against a dominant buyer has a poor record: Germany and Spain tried it in 2014 and lost traffic instead of gaining revenue. - Header bidding raised publisher yields by putting several buyers on the same impression at once, not by uniting sellers. - Every current AI content buyer treats a page as an input cost to minimize, while brands value the same page as media, which makes them the missing second bidder. - Publishers should use coalitions such as SPUR for standards and measurement, and competing demand for price. ## What AI content pricing power is AI content pricing power is a publisher's ability to influence what an AI system pays when its content is fetched, used or cited to build an answer. In the link economy, pricing power came from audience: advertisers competed for the reader a page attracted. In the agentic web, that reader often never arrives, so the page gets priced as raw material instead of media. ## The ad exchange auction vs the AI content exchange In 2014, most publishers sold programmatic inventory through a waterfall, offering each impression to buyers one at a time, with Google's AdX holding last look. Header bidding broke that sequence by letting several exchanges bid on the same impression simultaneously. The Telegraph told Digiday its programmatic revenue rose 70% within nine months of moving all inventory to header bidding. Google spent years containing the change, launching its own exchange bidding product in 2016. Internal documents surfaced in the DOJ ad tech case later acknowledged how much last look had been worth to AdX. The lesson is simple: publisher prices rose when demand competed, not when supply organized. The AI content market today looks like the waterfall. Cloudflare's Pay Per Use beta, launched September 30, 2026, lets buyers make offers that publishers accept or decline. Google's pilot pays what Google judges a contribution to be worth. The rails are new, but the sequence is old, and the buyer always holds last look. ## What Fast Company reported, and what we verified Pete Pachal's October 2, 2026 Fast Company column argues that a market for AI content now exists, but publishers have no say in the price. Data brokers such as Exa, Parallel and Tavily resell scraped content to AI firms, agencies and investors, a market that one estimate cited in Matthew Scott Goldstein's report puts at about $1 billion. We could not reproduce that figure independently, so treat it as directional. The bot numbers hold up. DataDome's 2026 State of Bot & Agent Security Report, released September 22, found bad bot traffic grew 124% between July 2025 and June 2026, with scraping up 185.2%, and 65.3% of 21,491 tested sites failed to stop any of its ten bot types. One nuance: Meta's 46.3% share refers to identified AI agent and crawler requests across DataDome's customer base, and the report covers websites broadly, not publishers specifically. ## Why publisher coalitions will not set the price of AI content ### The collective-leverage case at its strongest Pachal's argument deserves its full weight. Spotify did not pay artists just because it built a reliable exchange; rights holders had the collective muscle to insist. People Inc.'s Jonathan Roberts summed up the current state this summer as 30 Napsters for content and no Spotify. With the Department of Justice backing fair use for training, coalitions such as SPUR, which the Associated Press recently joined, look like the only lever left. ### What Germany and Spain taught publishers in 2014 Europe has already run this experiment. When Germany's ancillary copyright law let publishers demand fees, Google dropped their snippets in October 2014, and the VG Media publishers soon granted Google a free license. Axel Springer reported Google search traffic down 40% and Google News traffic down 80% within two weeks. Spain made fees mandatory, and Google closed Google News Spain in December 2014. Collective leverage fails against a buyer that can walk away, and AI platforms can walk further than Google News ever could. When two-thirds of tested sites cannot block a single test bot, the walk-away threat runs in one direction only. ### Price discovery needs a buyer with a different motive Brian Morrissey of The Rebooting argues most content is too commoditized to command a price, since an AI needs only one good enchilada recipe. He is right about every buyer currently in the market. AI platforms and data brokers all treat a page as an input cost to push toward zero. A brand does not want any recipe; it wants presence on the page the engine actually reads and cites, and it values that page as media. That is the structural point the Fast Company piece leaves out. The second bidder the open web needed in 2014 was another exchange. The second bidder AI content needs in 2026 is the advertiser. ## What a second bidder means for brands and for publishers ### For CMOs, media buyers and agencies: you are already paying, just not the source Brands already spend to shape AI answers through GEO retainers, PR and sponsored content, yet almost none of that money reaches the pages engines cite, and none of it is priced per citation. Media buyers should ask for that inventory directly: labeled, measurable placements on the source pages AI agents read. Agencies that build this buying desk now will own the rate card later. ### For publishers: sell to two buyer classes, never one Take the licensing money, but never let a buyer-set payout become your reference price. Instrument which pages AI agents fetch and cite, demand page-level usage data in every deal, and put brand demand on the same inventory so licensing offers meet a floor set by competition. Use SPUR the way the open web used the IAB: for identity, measurement and standards, not for price. ## Three signals that will decide who sets the price of AI content First, whether Cloudflare's Pay Per Use admits more than one buyer type and shows publishers bid density, not just accepted offers. Second, whether Google's AI contribution panel goes page-level; participants currently describe it to Digiday as 'quite black box.' Third, how courts treat outputs in The New York Times v. OpenAI, after the DOJ conceded that outputs reconstructing original works may not be transformative. Inference, not training, is where pricing power will be won. ## Conclusion Hold on to this: AI content pricing power will come from the number and variety of bidders, not from the size of the sellers' coalition. Generative Engine Advertising adds the missing bidder. Smalk AI is the AI Search ad network that places native ads for AI agents on the publisher pages engines read and cite, giving brands presence where answers form and giving publishers a second buyer class whose demand puts a floor under every licensing offer. What to watch next: the first exchange that lets AI licensing buyers and brand advertisers bid on the same page. That is the moment the content waterfall becomes an auction, and AI content pricing power starts shifting back toward publishers. ## FAQ ### Who sets the price of publisher content used by AI search engines? Today the buyer does. Google's pay-per-value pilot pays what Google judges a contribution is worth, Cloudflare's Pay Per Use beta lets AI companies make offers that publishers accept or decline, and data brokers resell scraped content at commodity prices. Publishers can refuse, but they rarely set the number. ### Why can't publishers band together to set AI licensing prices? They can coordinate, but collective action against a dominant buyer has a weak record. In 2014, German publishers granted Google a free license within weeks of losing their snippets, and Spain's mandatory fee led Google to close Google News there. Coalitions are better suited to standards than to price. ### What is the difference between pay-per-crawl, pay-per-use and pay-per-value? Pay-per-crawl charges each time a bot fetches a page. Pay-per-use, Cloudflare's 2026 model, pays when content shapes an AI answer. Pay-per-value, Google's pilot, pays only when Google judges the content contributed significantly, at a rate Google sets. ### What does header bidding have to do with AI content licensing? Header bidding raised publisher ad revenue by letting several buyers bid on the same impression at once instead of in sequence. AI content is still sold buyer by buyer. Publisher prices rise when demand competes, so AI content needs more competing bidders, not just better payment rails. ### Why would brands pay publishers for visibility in AI answers? Because AI engines build answers from publisher pages, and brands need presence where those answers form. A brand values a cited page as media rather than as raw input, so it pays a different price than an AI platform, which is what gives publishers a real market.