AI Agents Are Now the Primary Users of Databases, and the Architecture Must Change
ET Edge Insights, Wednesday, September 9th, 2026
TiDB CTO Ed Huang says AI agents now create most new databases, requiring serverless, branchable, scale-to-zero architectures.
In this interview, TiDB co-founder and CTO Ed Huang argues database scalability is largely solved and the new frontier is architectural simplification for AI agents. He says over 90% of new TiDB Cloud clusters are created by AI agents, which spin up, mutate, and discard databases at machine speed.
Huang lists requirements including second-level provisioning, S3-backed compute-storage separation, non-blocking schema changes, unified transactional, analytical, and vector workloads, and native branching.
He also calls for cost governance through scale-to-zero, request-unit billing, and per-agent metering that acts as a circuit breaker.