# Bots now generate more web traffic than humans and startups are about to feel it > Source: > Published: 2026-07-23 21:22:22+00:00 *Cloudflare says bots now make up most of the web page requests it sees, and that should make any startup built on traffic metrics slow down and check what is actually human.* The number Cloudflare CEO Matthew Prince posted on June 3 is the one founders should not wave away: 57.4% of HTTP requests to web content on Cloudflare Radar were automated, against 42.6% from humans. Prince had expected that threshold near the end of 2027. It arrived early. "Bots have now passed human traffic online," he wrote on X. That is not a small analytics quirk. It changes what traffic means. Start with the scope. This is Cloudflare's view of web page requests, not a census of every app, private server, API call, or website on earth. But Cloudflare sits in front of enough of the web that the signal matters. If your dashboard treats every request like a person, you're already kidding yourself. The growth behind it is not only old-fashioned spam. According to HUMAN Security's 2026 State of AI Traffic and Cyberthreat Benchmark Report, which analyzed more than one quadrillion interactions in 2025, automated traffic grew eight times faster than human traffic, and traffic from AI agents and agentic browsers rose 7,851% year over year. Cloudflare's July bot report adds another hard detail: 52% of crawler requests were for AI training as of June 2026, up from 22% in spring 2025. These aren't search bots doing the old bargain of indexing pages and sending readers back. Many are training crawlers and mixed-use crawlers taking content, refreshing it, and moving on. They don't read your ad. They don't sign up. They don't sit through your onboarding flow. They hit the server and leave a mark in the log. Traffic has always been a proxy for traction. Investors use it, acquirers use it, and founders lean on it in every pitch deck. That proxy is now damaged. If more than half your traffic is automated, your monthly active user count may be inflated, your funnel conversion rate may be distorted, and your cost-per-acquisition math may be wrong. A SaaS product, a media property, or an API business built on the assumption that a pageview is a human being has a measurement problem before it has a growth problem. The cost is real too. Publishers pay for servers while AI crawlers fetch content at scale. API businesses absorb automated calls that may never become subscriptions. E-commerce sites can see catalog requests inflate traffic dashboards while adding nothing to gross merchandise volume. The ugly part is that none of this shows up cleanly until someone separates human traffic from automated traffic and asks what each group is worth. For investors valuing web-native companies, this is no longer a footnote. A platform showing 40% traffic growth year over year might be showing flat human growth under a layer of automated noise. Don't underwrite that blindly. Any diligence process that still asks only for traffic growth, bounce rate, and conversion needs a new line item: bot share by source, by purpose, and by revenue impact. ## Cloudflare wants crawlers to pay Cloudflare's answer is a payment and permission layer. On July 1, the company said Pay Per Crawl is evolving into Pay Per Use, and TechCrunch reported that Cloudflare will start blocking mixed-use crawlers from ad-supported pages by default on September 15, 2026, for new customers, new sites from existing customers, and existing free customers that have not changed their settings. Search can still be allowed. Training and agent use become separate choices. The distinction matters. A publisher may want Google Search to find an article while refusing to let the same crawler feed a commercial AI system without payment. Cloudflare is trying to turn that into a dashboard setting rather than a philosophical argument. Ceramic.ai and You.com are the first named partners. Cloudflare says Ceramic will pay participating publishers when their content appears in Ceramic search results, while You.com will let an agent pay for premium content it needs. That's the beginning of a market, not the end of one. Clutch's March guide to Pay Per Crawl estimated that high-traffic sites could make $50,000 to $200,000 a month if crawl volume and per-request pricing line up, while smaller sites may earn little or nothing. That is the right caveat. This will not save every blog. It may, however, create a real revenue line for publishers with scarce information and large crawler demand. Frankly, the bigger startup opportunity may sit below the content itself. Someone has to build metering, authentication, crawler identity, settlement, reporting, and dispute handling for machine-to-web transactions. Cloudflare has the strongest starting position because it already sits in the CDN layer. But this category is still young. A founder looking for a problem with pressure on both sides should look hard at bot monetization. The internet has tipped into a majority-automated environment faster than Prince expected, and the business models sitting on top of human-traffic assumptions have not caught up. That is where the damage lands first. It is also where the opportunity opens, for teams willing to measure the web as it now works rather than the way their dashboards still pretend it does. **Also read:** [SAP's cloud backlog hits €22.9 billion as the ECC migration countdown enters its final stretch](https://startupfortune.com/saps-cloud-backlog-hits-229-billion-as-the-ecc-migration-countdown-enters-its-final-stretch/) • [Trump's expanded AI data center pledge covers 80% of the U.S. grid but has no teeth to enforce it](https://startupfortune.com/trumps-expanded-ai-data-center-pledge-covers-80-of-the-us-grid-but-has-no-teeth-to-enforce-it/) • [Devin AI and GPT-5.6 Sol Ultra crack math problems that stumped researchers for half a century](https://startupfortune.com/devin-ai-and-gpt-56-sol-ultra-crack-math-problems-that-stumped-researchers-for-half-a-century/)