The new gatekeepers of the AI economy

Less than four years after generative artificial intelligence entered the mainstream, a new market has emerged around one of the technology's most valuable resources: information. OpenAI alone has announced licensing agreements with more than 20 publishers spanning North America, Europe, Latin America and Asia, while companies including Google,  Anthropic and Perplexity have pursued similar partnerships with news organizations and content owners.

While much of the discussion has focused on the new revenue they could generate for publishers, recent analyses from antitrust researchers, competition regulators and media policy experts suggest these deals represent far more than commercial partnerships. They point to the emergence of a market in which access to trusted information is concentrated among a relatively small group of companies capable of developing frontier AI models.

The licensing agreements now being signed are about far more than copyright. They are laying the foundations of a new information marketplace that will help determine who becomes the gatekeepers of the AI economy. If the internet's first generation rewarded those who organized information, the AI era may reward those who control access to it.

A familiar pattern in a new market

The shift bears a striking resemblance to the internet's first great platform transition. During the rise of search engines and social media, companies such as Google and Meta became the primary gateway through which billions of people accessed online information. Publishers continued producing journalism, but readers increasingly reached their content through platforms rather than by visiting news websites directly. Advertising revenue followed user attention, allowing a handful of technology companies to capture an expanding share of the digital advertising market while many publishers became dependent on platforms they did not control to reach audiences.

The Open Markets Institute, a Washington-based research organization focused on competition and monopoly power, argues in an April 2026 analysis that the companies that became gatekeepers of the digital advertising economy are now positioning themselves to become gatekeepers of AI content licensing.

Instead of directing users to original sources through search results, AI assistants can answer questions directly using information obtained through licensing agreements or web scraping. The report argues this changes publishers' bargaining position because only a limited number of companies possess the financial resources, computing infrastructure and distribution networks required to develop frontier AI models. Today, that group includes OpenAI, Google, Anthropic, Meta, Microsoft, xAI and Chinese developers such as Alibaba and Baidu. With relatively few potential buyers for premium content, the report argues that publishers face a concentrated market in which negotiating power rests largely with AI developers.

Control extends across the AI stack

The licensing market cannot be viewed in isolation. It sits within a broader ecosystem in which data, computing power, cloud infrastructure and AI models reinforce one another.

The United Kingdom's 2024 AI Foundation Models Update Paper argues that market power in artificial intelligence is created through the interaction of multiple layers of the technology stack rather than any single component. Companies that control cloud infrastructure, proprietary datasets, frontier models and consumer-facing applications can strengthen their position across the entire ecosystem.

Many of the companies negotiating licensing agreements already span multiple layers of the AI ecosystem. Microsoft partners with OpenAI while operating Azure, Google combines Gemini with Google Cloud and Search, and Amazon backs Anthropic through Amazon Web Services.

The accompanying technical report notes that this degree of vertical integration can lower costs, accelerate innovation and improve efficiency. Those same advantages, however, may create barriers for competitors that cannot match the computing capacity, financial resources or exclusive commercial partnerships available to established firms, making early intervention critical before dominant positions become entrenched.

Publishers weigh opportunity against dependence

For publishers, licensing agreements offer an important new source of revenue after years of declining digital advertising income.

OpenAI has signed agreements with the Financial Times, News Corp, Axel Springer, Condé Nast, Vox Media and The Atlantic, while Getty Images has expanded its partnership with the company. Similar agreements are emerging worldwide as AI developers compete for trusted information.

The commercial terms behind those deals, however, remain highly fragmented. According to the Open Markets Institute, publishers retain anywhere from 50 percent to 100 percent of licensing revenue depending on the intermediary. TollBit allows publishers to keep all licensing revenues, Cloudflare's marketplace leaves publishers with roughly 70 percent, and ProRata allocates about half. With no established standard for dividing licensing revenues, the design of these marketplaces may ultimately determine who captures the greatest share of the value created by AI.

The opportunity also comes with new dependencies. The Open Markets Institute argues that if only a handful of companies dominate the market for premium content, publishers could become reliant on a small number of AI buyers. Nieman Lab, a journalism publication at Harvard University, notes that publishers face a difficult tradeoff. Refusing to license content may mean foregoing an important new source of income, while embracing long-term agreements could strengthen the same companies that already occupy influential positions across the digital information economy.

Disputes also continue over how AI developers obtain information outside formal licensing agreements. Allegations that some companies bypassed robots.txt, the voluntary internet standard publishers use to discourage automated scraping, underscore that the rules governing access to online information remain unsettled.

The next competition debate

Copyright has dominated much of the public discussion surrounding artificial intelligence, but competition policy is becoming equally important.

The first generation of the internet determined who controlled access to audiences. The AI era may determine who controls access to trusted information. If only a small number of companies possess the resources to license vast quantities of premium content while also controlling cloud infrastructure, computing power, foundation models and consumer platforms, they could occupy influential positions across nearly every layer of the AI economy.

That outcome is not inevitable. Competition authorities in the United Kingdom, Europe and elsewhere are examining these markets before commercial relationships become deeply entrenched. The emergence of AI licensing agreements therefore represents more than a new business model for publishers. It is an early test of how the information economy will function in the age of artificial intelligence.