AWS just released OpenSearch Serverless, a search and vector database that can spin up compute in seconds and shut it down to zero when idle, a first for AI agent workloads.
Agent Traffic is Exploding
Cloudflare reports bots now make up 31% of all HTTP traffic, with AI crawlers, search engines, and assistants contributing roughly a quarter of those bot requests. Li Yi Ohlsen, Cloudflare’s senior product manager, predicts non‑human traffic will exceed human traffic sometime in the first half of 2027. AI agents, which spawn sub‑agents that query hundreds of databases in seconds, are already a significant fraction of that traffic and are poised to grow.
AWS’s Serverless Breakthrough
Tia White, general manager for Amazon OpenSearch Service, explained that the new generation decouples compute from storage. Previous serverless versions required at least one instance to stay operational because compute and storage were coupled, forcing customers to pay for idle compute. The upgraded model lets compute scale up instantly to accommodate agent bursts and scale back to zero when idle, meaning customers pay $0 for unused resources.
This design mirrors a metered parking spot rather than a reserved space: you only pay for the compute you actually use. At launch, OpenSearch Serverless will integrate natively with AI development platforms like Vercel and Kiro, allowing developers to deploy production‑ready search and vector backends for agents without managing infrastructure.
Industry Ripple Effect
Databricks and Snowflake are repositioning themselves as AI memory and retrieval systems for enterprise data. Microsoft has rolled out Azure updates that handle AI agent bursts and share memory between agents. Cloudflare, following Amazon’s lead, introduced infrastructure that gives agents persistent environments and instant scalability. These moves signal a broader industry shift: as AI agents become mainstream, infrastructure originally built for human‑driven traffic will no longer suffice.
What Comes Next
With serverless search now able to meet the unpredictable, bursty demands of autonomous agents, enterprises can deploy AI assistants at scale without the cost and operational overhead of traditional search clusters. As the internet increasingly becomes a machine‑centric ecosystem, tools like OpenSearch Serverless will likely become the default backbone for autonomous data retrieval, lowering entry barriers for startups and large organizations alike.
Source: The internet is being rebuilt for machines
Domain: techcrunch.com
Comments load interactively on the live page.