Fastly has published research showing AI-generated traffic on its network grew about 30% between January and May 2026, about 6.5 times faster than human traffic over the same period.
The findings point to a shift in how companies must treat automated internet activity as AI systems move from being mainly a security concern to becoming a factor in content distribution, infrastructure planning, and customer acquisition.
Fastly's analysis of activity across its global network found that machine-driven requests now account for a meaningful share of internet traffic, spanning AI crawlers, fetchers, bots, agents, and API-based systems. More significant than the increase in volume, the report said, is the widening variety of ways AI systems interact with websites, applications, and online services.
That distinction matters because different forms of AI traffic place different burdens on digital infrastructure. More than half of AI requests, or 51%, required origin access in May 2026, compared with less than 9% of human requests, according to the research. Origin access refers to traffic that reaches the underlying server rather than being handled closer to the network edge.
Traffic split
The report distinguishes between two main forms of AI activity: crawlers and fetchers. Crawlers collect information from the web to build and refresh AI models, while fetchers retrieve information in response to a user request made through an AI assistant or software agent.
Fetchers may prove particularly important for publishers, retailers, and service providers because they can influence whether a company's information appears in answers delivered by AI tools. That means decisions on whether to allow, restrict, or block such traffic may have commercial consequences beyond server load and cyber defence.
Organisations are increasingly weighing content protection, visibility, digital distribution, and customer acquisition when setting rules for AI systems, Fastly said. It cited examples from its network in which one large company imposed a hard block after a sudden rise in AI fetcher traffic, while another allowed AI agents and then saw fetcher volumes rise over several months.
Those contrasting responses reflect wider uncertainty in the market. For some businesses, limiting AI access may help them retain control over content and how it is used. For others, allowing access may improve the chance of being surfaced in AI-driven search and recommendation tools.
Fastly also highlighted sharp growth in traffic linked to Claude, which it said rose by more than 555% from its January 2026 baseline. The data suggests that usage patterns tied to individual AI services can shift quickly, adding another layer of complexity for network operators and site owners.
Business choices
The report frames AI traffic management as a broader business issue rather than a narrow technical one. Companies are being pushed to decide which machine interactions create value, which should be limited, and which should be blocked outright.
That requires more detailed monitoring of who is accessing digital properties, how those systems behave, and whether they serve a commercial purpose. It also suggests a blanket policy for all bots is becoming less workable as AI tools take on more tasks on behalf of users, from fact-checking and price comparison to answering questions in real time.
Artur Bergman, Founder and Chief Technology Officer at Fastly, said the internet is entering a period in which human users are no longer the only audience companies need to consider. "AI traffic is fundamentally changing how the internet operates," Bergman said.
"Businesses are moving beyond a world where humans are the primary users of digital experiences. The challenge is no longer simply blocking bots, it's understanding which machine interactions should be accelerated, managed, challenged, or stopped," he said.
Fastly said an effective approach rests on three elements: visibility into which AI systems are interacting with online assets, context on how they behave and whether they create business value, and the ability to respond differently depending on intent and impact.
The figures underline how quickly AI is becoming a distinct layer of internet activity, with consequences for website operators, cloud providers, and any business that depends on being found online. In May 2026, more than half of AI requests on Fastly's network required origin access, compared with less than 9% of human requests.