Cloud infrastructure has long been designed around humans searching, clicking, scrolling, and streaming in a steady and predictable manner. AI agents behave differently. You can trigger countless activities by querying hundreds of databases, retrieving documents, calling APIs in seconds, and then spinning up multiple subagents that disappear as quickly as they arrive.
With this premise in mind, Amazon is redesigning key parts of its cloud infrastructure. AWS on Thursday launched the next generation of OpenSearch Serverless, a fully managed search and vector database (essentially a system for storing and retrieving information at scale) designed specifically for agent workloads. AWS says the new system can instantly scale up when agents trigger actions and scale back down to zero when idle.
This launch reflects a growing awareness across the technology industry. The infrastructure originally designed for a human-centric Internet does not work well in a world with more and more agents.
Although AI agents still account for a relatively small portion of Internet activity, machine-generated traffic is already significant and poised to grow. According to Cloudflare, bots accounted for 31% of all HTTP traffic over the past six months. AI crawlers, search engines, and helpers accounted for about a quarter of all bot requests during that period.
“Non-human traffic will exceed human traffic in the first half of 2027,” Lai Yi Ohlsen, senior product manager at Cloudflare, told TechCrunch.
At Google’s I/O developer conference last week, the company revealed that users will be able to delegate tasks such as researching purchases, booking travel, searching the web, and interacting with apps to AI systems. But costs are not limited to consumer-focused AI agents. Enterprises are increasingly deploying agents, both internally and for their customers, to generate new kinds of machine-generated traffic behind the scenes.
As a result, cloud providers and infrastructure companies have been thinking about how to apply systems built for humans to a world of agents that continuously and autonomously retrieve information, invoke tools, and generate machine-to-machine traffic.
AWS’s new OpenSearch Serverless is here.
“The timing is simple: Agents are moving from experimentation to production and are generating traffic patterns that our previous infrastructure could not have designed for,” Tia White, general manager of Amazon OpenSearch Service, told TechCrunch. “With volume spiking and idling without warning, businesses need search that can continue without paying for empty or idle compute.”
The key technology change in this new generation is to decouple compute from storage, allowing compute to scale up and scale down to zero in seconds to accommodate spikes in agent traffic, ensuring customers pay $0 when agents are idle.
“Previously, even in older serverless versions, storage and compute were combined, so you had to have more than one instance up and running,” White said. “We couldn’t automatically spin up (compute) at the speed we needed, so we always reserved idle compute for our workloads, whether they were being used or not.”
Think of it like always paying for parking, even when you’re not using the space. Using AWS’s upgraded serverless is like paying for a metered parking space.
OpenSearch Serverless will natively integrate with AI development platforms like Vercel and Kiro at launch, allowing developers to deploy production-ready search and vector backends for agents without having to manage infrastructure.
Change is occurring throughout the cloud industry. Databricks and Snowflake are positioning themselves as AI memory and search systems for enterprise data. Microsoft has released an update to Azure designed to handle AI agent bursts and share memory between agents. Cloudflare, like Amazon, introduced infrastructure last month to provide agents with a persistent experience and immediate scalability.
As more companies deploy AI agents, there will be greater pressure to re-architect their infrastructure around machine-generated workloads, ultimately making agents cheaper and easier to deploy at scale.
If you purchase through links in our articles, we may receive a small commission. This does not affect our editorial independence.
