Enabling Agentic AI for Semiconductor Testing

Agentic AI for semiconductor testing demands a test-centric infrastructure that can reason, decide, and act in real time at the tester, not just offline in the cloud. Modern test floors face rising complexity from chiplet-based packages, exploding data volumes, and stricter data-security and data-movement constraints across multi-party manufacturing. 

This talk introduces an architecture that spans cloud, secure factory servers, and embedded AI at the test cell to enable adaptive test, intelligent binning, and real-time anomaly detection within tight in-socket latency budgets. It will highlight practical patterns for orchestrating multiple agents per tester, securing data across OSATs and foundries, and continuously improving models so test becomes an autonomous, continuously optimizing system that delivers higher quality and yield at lower cost.