AI Agents Are Rewriting the Web’s Business Model

Introduction

The web is entering a new phase in which traffic volume and human attention are no longer the same thing. In an interview with The Verge, Cloudflare CEO Matthew Prince said automated traffic has already surpassed human traffic and continues to grow as AI companies scrape the web and agents perform tasks on behalf of users. For website operators, this is not merely a change in analytics. It challenges the economic model that has supported the web for decades.

Key takeaways

• Cloudflare wants to be more than a security provider. Prince described the company not simply as a service that speeds up websites and blocks malicious activity, but as part of the infrastructure needed to rebuild the internet. Cloudflare operates a large global network and provides security, application scaling, and edge-based computing services.
• Machine traffic is becoming the default background of the web. Because Cloudflare sits between websites and visitors, it can observe a broad portion of internet activity. Prince said the company’s earlier forecasts for automated traffic overtaking human traffic were repeatedly moved forward as AI usage expanded. Scrapers and agents are now major sources of demand.
• Advertising does not translate cleanly to bots. Search, display advertising, and analytics helped build the modern web, with Google playing a central role in that ecosystem. But bots do not click ads or respond to branding in the same way people do. More requests can therefore create more bandwidth and computing costs without creating equivalent revenue.
• Access may become a paid and negotiated service. Cloudflare is giving website owners options to block AI tools, allow them, or permit only tools that may pay for access. This points toward a system based on identification, authorization, and billing rather than a simple choice between an entirely public page and a fully blocked one.

Why it matters

The shift will be especially important for publishers, data providers, and AI companies that depend on public web information. If scraping produces costs without meaningful compensation, websites may tighten their bot policies or place more content behind controlled interfaces. AI companies that want reliable, high-quality information may need to demonstrate the value of their traffic and eventually share the cost of infrastructure or content rights. The web could move from an advertising-subsidized commons toward a tiered system in which machine access is priced differently from human access.

Cloudflare is well placed to influence that transition because it operates between sites and their visitors. It can identify traffic, enforce rules, and use network-level data to help websites make decisions. Yet the same position raises difficult questions. Who should determine whether an agent is trustworthy? How should a publisher price access? Will creators receive a fair share of the value generated by their material? And could more granular controls weaken the openness that made the web useful in the first place?

The AI disruption also reaches inside Cloudflare itself. The material notes that the company laid off more than a thousand employees, representing about one-fifth of its workforce, and that Prince wrote about deciding which roles could be replaced by AI. This connects the external transformation of web traffic with a broader internal transformation of organizations, costs, and job design.

Cloudflare cannot by itself decide whether the next web will be healthier. But its strategy highlights the central question: when the web’s fastest-growing users are agents and models rather than people, who should pay for the infrastructure, information, and access that make the system work?

Source: The Verge AI