Cloudflare CEO Matthew Prince is warning that the economic bargain behind the open web is under pressure. In his view, the long-standing exchange between websites and search engines has shifted from mutual benefit toward extraction, with publishers giving more content to crawlers while receiving fewer visitors in return.
The issue is not only lower traffic. Prince argues that shrinking referrals weaken advertising, subscriptions, and the less formal reward of knowing that people are actually reading the work. If those incentives fade, he says the reason to publish online becomes harder to defend.
The search bargain is changing
For years, search engines crawled websites, indexed their pages, and sent people back to the original source. That system gave search platforms useful answers while giving publishers visibility, readers, and potential revenue.
Prince says Cloudflare data shows that this balance has deteriorated sharply. Ten years ago, Google crawled two pages for every one visitor it sent back to the original site. That ratio later moved to 6:1, and now stands at 18:1, according to Prince.
The practical effect is simple: publishers are giving more to the indexing system while receiving less direct audience value. Even if a website remains visible inside a search experience, that visibility may no longer translate into a click, a subscription, or an advertising impression.
Prince points to Google's increasing tendency to answer questions on its own platform. Six months ago, Cloudflare observed that 75 percent of Google searches ended without a click to another website. With Google's "AI Overview" feature, Prince estimates that figure could now be as high as 90 percent.
That shift matters because the open web has depended on movement. A reader searches, finds a page, visits the source, and the publisher has a chance to build value from that visit. When the answer stays inside the search platform, the publisher may still supply the underlying information, but the business benefit is much weaker.
AI search makes the imbalance larger
Prince argues that AI-based search intensifies the problem because the crawler-to-referral ratios are even more extreme. Six months ago, OpenAI crawled 250 pages for every visitor it sent to a source. Now, according to Prince, that ratio is 1,500:1.
Anthropic's ratio is described as even more severe at 60,000:1. These numbers show a much wider gap between the amount of content accessed and the amount of traffic returned.
For publishers, the concern is not only that AI systems read the web. It is that users may accept AI-generated summaries and rarely check the original sources. The source article says early research backs up that trend.
That creates a direct challenge for content creators. If people receive a summarized answer without visiting the source, the publisher loses the chance to earn ad revenue, convert a reader into a subscriber, or deepen its relationship with an audience.
Prince frames the stakes in broad terms. "I don't know if you can't sell subscriptions and you can't sell ads and you don't get an ego hit from knowing that people are consuming your stuff why anyone's going to create content," he says.
Why licensing deals may not solve it
Some publishers have responded by signing content licensing agreements with AI companies such as OpenAI. Prince argues that these deals can be "naive" when they fail to create real scarcity.
His point is that a paid agreement with one AI company may not protect the same content from being scraped by others. If a publisher sells access while leaving the door open to unpaid crawlers, the paid buyer is not operating in a truly scarce market.
Prince summarizes the issue bluntly: "You can't have a market if you don't have scarcity." In his view, short-term payments can weaken the long-term value of content if other AI systems can still obtain similar access for free.
The renewal problem is central to his warning. If content becomes easy to obtain elsewhere, publishers may have less leverage when current agreements come up again. Prince says, "The renewal of the deal that you signed today will be worse tomorrow, I guarantee that."
The source article also notes that journalism professor Jeff Jarvis has called these licensing payments "pure lobbying." It further describes the situation as steering publishers toward a classic big tech prisoner's dilemma, where individual deals may make sense in the moment but leave the wider industry in a weaker position.
Cloudflare's proposed block
Prince says publishers need to prevent AI crawlers from taking their content if they want to create scarcity. Cloudflare plans to offer a free protection system that blocks websites from being scraped by AI models.
A symbolic launch is set for the end of June in New York. Publishers will gather to press a "red button" and activate the block. Prince says all the major media companies are on board, and even some AI providers have shown interest.
The proposal is not only defensive. Prince also describes a possible future market in which content is valued by how much it contributes to human knowledge, rather than by page views alone.
In that model, large language models could license specific content to fill knowledge gaps. The source gives one example: information relevant to a medical AI model would be worth more than cultural content aimed at teenagers.
Prince describes the idea this way: "Imagine that in the future instead of being compensated by how many page views you drove if you could instead be compensated by how much the information that you create advances human knowledge." His broader warning is even sharper: "If we don't fix this, the internet will die."
What is at stake for the open web
The debate matters because the open web depends on a large network of people and organizations choosing to publish useful material online. If the incentives decline, the supply of high-quality public content may also decline.
Cloudflare has its own stake in the outcome. The company runs a global network for hosting, optimizing, and securing websites and online services. That business depends on an ecosystem where publishers and other organizations still see value in making content and services available online.
Prince's argument is ultimately about leverage. Search engines and AI models need web content, but publishers need a way to turn that content into sustainable value. Without a new balance, the systems that summarize and retrieve information may weaken the same web they rely on.