Cloudflare CEO says AI bots are overtaking the open web — and the web’s business model must change

Cloudflare CEO Matthew Prince says automated traffic has already overtaken human traffic across the internet, a shift he argues is forcing the web to rebuild its business model around AI bots, agents and payments for access. His message is blunt: the old advertising-driven web was built for people, but the next version will need a way to charge machines.

In a wide-ranging conversation about the future of the internet, Prince described a world where AI systems increasingly scrape, summarize and act on web content at scale, creating a growing strain on publishers, platforms and the infrastructure that supports them. He said Cloudflare is trying to help websites control which bots can access their content, and under what terms, while also exploring payment systems that could let content creators charge AI companies and agents for use of their data.

The stakes are significant. Cloudflare, which sits in front of a large share of the web, believes the rise of bots could reshape how information is published, discovered and monetized online. The company’s chief executive also linked the trend to a broader shift inside Cloudflare itself, where he has already cut more than a thousand jobs while leaning harder on AI internally.

What Cloudflare says is changing on the internet

Cloudflare’s central thesis is that the web is no longer primarily a place where people browse pages one by one. Increasingly, it is becoming a machine-readable system where AI tools pull in huge amounts of content, parse it, and use it to answer questions or complete tasks on behalf of users.

Prince said Cloudflare’s network gives the company a rare vantage point: it can see traffic patterns across a significant portion of the internet, allowing it to estimate how quickly non-human activity is growing. According to his account, the company found in June that bots accounted for more than half of all internet traffic, and the mix has continued to tilt toward automation since then.

That trend matters because the internet’s dominant economic model was designed for human attention. Display advertising, search referrals and brand recognition all depend on people seeing pages, clicking links and spending time on sites. Bots, by contrast, often extract information without generating the kinds of visits that publishers and businesses rely on.

Why automated traffic matters more than it used to

Automated traffic used to be a nuisance. Today, Prince argues, it is becoming the core driver of web usage, with AI systems consuming content at a scale that could dwarf human browsing in the years ahead.

He said Cloudflare’s internal projections have moved quickly as the pace of automation accelerated. In the company’s view, the crossover point between human and non-human traffic arrived far earlier than expected. Looking further ahead, Cloudflare believes machine-driven activity could become many times larger than human activity if current trends continue.

The implication is not simply more load on servers. It is a structural change in how the web pays for itself. A site designed to serve millions of people can survive on ads, subscriptions or e-commerce conversions. A site that serves billions of machine requests without compensation may struggle to justify the cost of hosting, crawling and bandwidth.

How Cloudflare wants to make AI pay for the web

Cloudflare’s answer is to give website owners more control over bot access and, in some cases, more leverage to demand payment. The company says publishers and site operators should be able to decide whether they want to allow AI crawlers, block them entirely or let them in only if there is compensation attached.

Prince framed the effort as a return to an older internet idea: if information is valuable, maybe access to it should carry a direct cost instead of being funded indirectly through ads. That model, he said, could support content creation in an era when AI agents may consume far more pages than humans ever did.

Cloudflare is working with firms including Coinbase and Stripe to think through how a payment layer might function. The goal is to make it possible for a website to receive tiny amounts of money each time a bot or AI system accesses information, much like a streaming service pays rights holders per use.

Prince argued that the web may need a modern version of “payment required” for the AI era, so creators can be paid when machines use their work at scale.

Why the old ad model is breaking down

Advertising depends on human behavior. People see the ad, click the ad or at least view the page long enough for the impression to be valuable. AI systems do not behave that way. They can ingest enormous volumes of data without ever providing the attention that funds the publisher.

Cloudflare’s view is that the economics of the internet are entering a dangerous mismatch: the costs of production and distribution are rising because of machine traffic, but the traditional revenue streams are not rising with them. If that gap widens, more publishers may stop creating original material, and the open web could become less useful over time.

• Human traffic has long powered web advertising.
• AI bots can consume content without generating ad revenue.
• Micropayments or access fees could create a new funding mechanism.
• Website owners may want the ability to set different rules for humans and machines.

What happens if bots become the majority of the web?

If automated traffic keeps rising, the web could become less like a public library and more like a utility grid with toll gates. That is the future Cloudflare is preparing for, whether or not the rest of the industry is comfortable with it.

Prince said the company’s role is to help shape those rules before the current system collapses under its own contradictions. In his view, the problem is not just technical. It is also an incentive problem. If the internet makes it easier for machines to consume than for people to create, the supply of original information could shrink.

One concern is what happens to smaller publishers, niche experts and public-interest sites if their work is aggressively scraped but rarely visited. Another is whether answer engines and chatbots become the primary interface to the web, reducing the visibility of the sources behind their responses.

The issue extends beyond media companies. Any business that posts menus, product details, documentation, local information or research online may eventually have to decide whether machine access is free, limited or paid.

How AI agents could change search and browsing

AI agents are not just summarizing pages; they are beginning to act on behalf of users, which means they may visit many more pages than a person would ever open manually. That shift can multiply demand on the web without producing proportionate value for the sites being visited.

Prince illustrated the point with a simple example: if an AI agent is asked where to eat lunch, it may crawl every restaurant menu nearby even though the user will ultimately choose only one place. At scale, that behavior can create enormous waste unless there is some cost attached to access.

In Cloudflare’s framing, the danger is a tragedy of the commons. Everyone benefits from a web that remains rich in current information, but individual publishers may not keep contributing if they bear the costs and receive too little in return.

Why Google is still central to the story

Google remains essential to understanding the transition because it shaped the first major economic era of the web. Prince described Google as the company that did more than any other to build the advertising machinery that funded the modern internet.

Search, analytics and ad infrastructure helped create an online ecosystem where publishers could make money by drawing human audiences. That system powered the growth of the open web for decades. But AI is now challenging the assumptions that made it work.

As more users get answers directly from chatbots and other AI interfaces, they may not visit the original source at all. That means the relationship between discoverability and monetization is weakening. Cloudflare’s push is, in part, an attempt to restore a financial link between information and value.

Prince described the advertising model as the engine of the first generation of the web, but said AI is forcing a search for the next one.

How much of the web is still human?

Cloudflare says the answer is changing fast. In the company’s view, what was once a mostly human ecosystem is now entering an era in which machine traffic can outweigh people, and that shift is happening faster than many expected.

The company’s data led it to revise its estimates repeatedly over a short period. First, Cloudflare believed automated traffic would surpass human traffic around late 2027. Then the estimate moved earlier. By spring, the company said the crossover had arrived sooner than that.

That speed matters because it suggests the transformation is not a distant possibility. It is already underway, affecting web infrastructure, publisher economics and the design choices that website owners face today.

Inside Cloudflare’s own AI transformation

The internet is not the only thing changing. Cloudflare itself is also reshaping how it works with AI. Earlier this year, the company laid off more than 1,000 employees, about one-fifth of its workforce, and Prince later wrote publicly about how he decided which roles could be replaced by artificial intelligence.

That makes Cloudflare both a participant in and a test case for the broader AI transition. The company is not simply guarding the web from machines; it is trying to use machines to make its own business leaner and more scalable.

Prince’s approach suggests two simultaneous beliefs: first, that AI will permanently change how internet businesses operate, and second, that companies that fail to adapt quickly may get left behind. That includes Cloudflare itself.

What Prince’s layoff strategy signals

Prince’s public comments about staffing changes were notable because they made the trade-offs explicit. Instead of treating AI as a vague productivity tool, he described it as something that can replace parts of the workforce and alter organizational design.

That view is controversial, but it fits Cloudflare’s broader thesis. If AI can change how content is created, consumed and monetized, it can also change how the companies serving that internet are structured.

The company’s message to investors and customers appears to be that AI is not a side project. It is the defining force reorganizing both the web and the businesses built on top of it.

Can micropayments really save the open web?

Maybe, but not automatically. Cloudflare’s proposal is compelling because it acknowledges the economics of machine access, yet the practical challenges are enormous.

A successful payment system would need frictionless transactions, broad adoption, predictable rules and enough trust that publishers believe they will actually get paid. It would also need major AI companies and platform providers to participate voluntarily or because web infrastructure makes participation the default.

The idea is not new. Web micropayments have been discussed for decades, but they never gained traction because the user experience was clunky and the incentives were misaligned. AI may finally create the need, but need alone is not enough. The system still has to work at internet scale.

Even so, the direction of travel is clear: more site owners want leverage over machine access, and more infrastructure providers are building tools to give it to them.

Why Cloudflare is positioned to shape the outcome

Cloudflare is unusually powerful in this debate because it sits in front of millions of websites and handles traffic between users, bots and origin servers. That gives it the technical ability to identify, block or route machine requests in ways many other companies cannot.

That position also makes Cloudflare a gatekeeper. If it succeeds, it could help establish the rules of the next web economy. If it fails, the company may still profit from the rising demand for bot management, even as publishers continue to struggle with the fallout.

Prince seems aware of both possibilities. His pitch is not just that Cloudflare can defend websites from AI pressure, but that it can help rebuild a healthier set of incentives before the current system degrades further.

• Cloudflare has visibility into a large slice of global web traffic.
• The company can offer publishers tools to block or charge bots.
• It is already working with payments companies on a possible access model.
• Its own business is being reshaped by AI internally.

What comes next for the web economy?

The next phase of the internet may be defined by a simple question: who pays when software, not people, consumes the web?

That question touches search, publishing, app development, e-commerce, local information, coding, documentation and every other category where AI systems can gather value from public pages. The answer will determine whether the open web remains a rich source of original material or becomes a resource slowly depleted by extraction.

Prince’s view is that the industry cannot keep pretending the old model will survive unchanged. If traffic keeps shifting toward AI systems, then publishers, infrastructure providers and payment companies will need to invent a new bargain.

Cloudflare is betting that the bargain will include rules, controls and payments at the edge of the network. The company is also betting that the businesses that adapt first will have the best chance of thriving in an internet where machines may outnumber people by a wide margin.

The larger question is whether the rest of the web will agree before the incentives collapse. For now, Cloudflare is making the case that the answer has to be yes.

Timeline of Cloudflare’s AI-era thesis

Cloudflare’s argument is ultimately about survival: if the web is becoming a machine market, then it needs machine economics. Whether those economics arrive through micropayments, licensing, platform deals or something else, the company is betting the era of free, unlimited bot access is ending.