Amazon Web Services launched a new generation of its OpenSearch Serverless platform in May 2026, redesigning the search and vector database service specifically to handle the unpredictable traffic patterns generated by AI agents.
The key technical change decouples compute from storage, allowing the system to scale up within seconds when agents trigger activity and scale back down to zero when idle — meaning customers pay nothing during quiet periods. Previously, even AWS’s earlier Serverless version required at least one instance running at all times because compute and storage were linked, leaving customers paying for idle capacity.
“Agents are moving from experimentation into production, and they create traffic patterns that previous infrastructure simply wasn’t designed for,” said Tia White, general manager for Amazon OpenSearch Service. “They spike without warning, they go idle without notice, and enterprise needs search that keeps up without paying for empty or idle compute.”
At launch, the service integrates natively with AI development platforms Vercel and Kiro. The move reflects a broader shift underway across the cloud industry. Databricks and Snowflake are repositioning as AI memory and retrieval systems, Microsoft has updated Azure to handle agent workloads, and Cloudflare introduced persistent agent environments last month.
The urgency stems from a measurable change in internet traffic. Cloudflare reports that bots accounted for 31% of all HTTP traffic over the past six months, with AI crawlers, search engines, and assistants making up roughly a quarter of bot requests. Li Yi Ohlsen, senior product manager at Cloudflare, told TechCrunch that non-human traffic is expected to exceed human traffic sometime in the first half of 2027.
Enterprises are also deploying agents internally and for customers, adding machine-generated traffic behind the scenes on top of consumer-facing AI tools. As more companies adopt agents at scale, this could increase pressure on infrastructure providers to further adapt their systems — which in turn may make agents cheaper and easier to deploy at larger scales.
Source: TechCrunch