Automated traffic overtook human traffic: May 2026 (Source: Cloudflare, 2026 Annual Founders' Letter (Sept. 27, 2026))
A crossover arrived two years early
Cloudflare's 2026 Annual Founders' Letter, published to mark the company's 16th birthday, revises a forecast it had held for years: that automated traffic — bots, crawlers, and AI agents — would overtake human traffic on the Internet in the second half of 2027. That crossover happened in May 2026 instead, Cloudflare says, pulled forward by the rapid rise of AI agents and crawlers. For any organization that budgets for web infrastructure on the assumption that visitors are mostly people, the letter argues that assumption no longer holds.
The repeat-visit problem
Cloudflare frames the cost of this shift as a tragedy of the commons: an AI agent tasked with finding a lunch spot might survey a thousand restaurant menus to recommend one, and the other 999 sites absorb the crawl load without any return. Compounding that, Cloudflare says its own data shows a majority of what well-behaved bots fetch is content that hasn't changed since their previous visit — meaning much of the crawl traffic hitting a site's servers is redundant by the crawler's own admission. Cloudflare says it has been building infrastructure so crawlers can identify what's new without re-fetching everything, aimed at cutting that load for site operators.
Discovery now favors whoever an agent already knows
The letter's sharper point for budget-holders is about who gets found. Cloudflare argues that where a human might choose a local business out of loyalty or convenience, an agent has no such preference — it defaults to whichever business it has the most information about, which is typically the incumbent that has been online longest. Cloudflare warns this dynamic could make it harder for new entrants to gain visibility as commerce shifts toward agent intermediaries, a risk the letter ties to possible market consolidation. It's a live concern set against a backdrop of rapid new-site creation: Cloudflare reports more than 7 million developers are now building on its developer platform, many using AI-assisted 'vibe coding' tools that lower the barrier to launching something new — even as the letter argues agents make that new creation harder to surface.
What to ask your team
Cloudflare says it plans to introduce ways for site and application owners to be compensated when agents use their content, alongside the crawl-efficiency work already underway — part of what it frames as a broader shift in how the web's business model has to adapt to agent-mediated traffic. For a budget-holder, the letter is a prompt to get current answers rather than assumptions: What share of inbound traffic to our properties is already automated, and is it billed differently than human traffic under our current hosting and CDN contracts? Do we have visibility into how much of that traffic is redundant re-crawling versus genuine new requests? And separately from SEO, is our content structured so that an AI agent evaluating options can actually find and cite us — or does our current setup depend on the kind of human loyalty and local awareness that an agent doesn't factor in?
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: Cloudflare.
