Bots Surpass Human Internet Traffic for the First Time, Cloudflare CEO Says

Leila RahimiLeila RahimiNewsAI12 hours ago

  • Cloudflare CEO Matthew Prince said that bots and AI agents have passed human traffic online for the first time, arriving more than a year before his own 2027 forecast.
  • Cloudflare Radar data put bot traffic near 57% of global HTTP requests, peaking around 62% over the prior week, across a network that sits in front of roughly one-fifth of all websites.
  • The milestone sharpens the “pay to crawl” debate, with Cloudflare embedding the Coinbase-built x402 standard into its pay-per-crawl tooling so sites can charge AI agents for access.

Bots and AI agents have surpassed human traffic on the internet for the first time, according to Cloudflare CEO and co-founder Matthew Prince, who shared the milestone in a post on X on Wednesday, June 3, 2026. Prince said agentic traffic had crossed the threshold sooner than he expected, and he linked to Cloudflare’s Radar dashboard so readers could check the live data themselves.

Bots Surpass Human Traffic Ahead of Cloudflare’s Own Forecast

The crossover arrived well ahead of Prince’s own timeline. In March 2026, speaking at the SXSW conference in Austin, he said that automated traffic would overtake human traffic by 2027. By his account, the rapid growth of agentic AI compressed that estimate by more than a year.

In the post, Prince wrote that “bots have now passed human traffic online for the first time in the Internet’s history.” He added that the shift had beaten his expectations, calling the pace faster than he had predicted.

Cloudflare Radar showed bots generating roughly 57% of HTTP requests worldwide against about 43% from humans, with the bot share climbing as high as 62% during the prior week. The vantage point is broad, since Cloudflare sits in front of close to one-fifth of all websites. Prince described the underlying data as “a bit messy,” while saying the web was clearly on the other side of the line now.

A Web Built for Machines, and the Pay-to-Crawl Question

Before the generative AI wave, bots accounted for about 20% of internet traffic, led by Google’s crawler alongside scrapers and spam systems. The new surge comes from AI agents that browse the web on behalf of users, often querying far more pages than a person would. Prince has illustrated the gap with a shopping example: a human comparing cameras might open five sites, while an agent doing the same task could hit 5,000.

The figures track HTTP requests, a measure of raw server hits. By time spent and engagement, humans still dominate the web. The split also varies by region. North America skews heavily automated at about 68.6%, while the US Midwest still tilts human at 54.5%, according to regional Radar breakdowns reported at the time.

For web3 and crypto builders, the milestone tightens a question Cloudflare has raised for months: how content creators get paid in a web where automated systems do most of the reading. Prince has argued the answer is “pay to crawl,” a model that charges automated visitors for access.

Cloudflare has wired the Coinbase-developed x402 payment standard, built on the dormant HTTP 402 “Payment Required” status code, into its pay-per-crawl tooling. At Consensus in Miami in May, Cloudflare Chief Strategy Officer Stephanie Cohen tied the same shift to agentic-payment rails, arguing that identifying, verifying, and charging bots could preserve the incentive to publish original content.

The next signals to watch are longer Radar time series, clearer bot-classification methods, and whether major assistant providers begin disclosing the web queries their agents make on users’ behalf.


Editorial Note: This news article has been written with assistance from AI. Edited & fact-checked by the Editorial Team.

Interested in advertising with CIM? Talk to us!

0 Votes: 0 Upvotes, 0 Downvotes (0 Points)

Donations

Stay Informed With the Latest & Most Important News

I consent to receive newsletter via email. For further information, please review our Privacy Policy

Loading Next Post...
Follow
Search
Loading

Signing-in 3 seconds...

Signing-up 3 seconds...