AWS OpenSearch Serverless redesigns cloud infrastructure for AI agent traffic bursts

Cloud infrastructure built for human browsing patterns is being rebuilt for AI agents, which spike unpredictably and vanish just as fast. Cloudflare reports bots account for 31% of HTTP traffic over the last six months, with AI crawlers, search engines, and assistants making up a quarter of that. Cloudflare's senior PM Lai Yi Ohlsen predicts non-human traffic will exceed human traffic by the first half of 2027. This shift is accelerating as enterprises deploy agents internally and for customers, creating new machine-to-machine traffic patterns.

Amazon's answer is the next-generation OpenSearch Serverless, a fully managed search and vector database that decouples compute from storage. Compute scales up in seconds to handle agent bursts and scales down to zero when idle, so customers pay $0 for idle time. G.M. Tia White likens the change from always paying for a parking space to paying for a metered spot. The system integrates natively with Vercel and Kiro for agent backends. Other cloud providers are following suit: Databricks and Snowflake reposition as AI memory systems, Microsoft updates Azure for agent bursts, and Cloudflare offers persistent agent environments. This redesign could make agents cheaper and easier to deploy at scale.