Mostly True
Siemens and TSMC have deployed AI agents to automate the correction of design rule check (DRC) errors in semiconductor design, an advancement that shortens tape-out timelines, according to information presented around TSMC's OIP Ecosystem Forum.
The collaboration applies AI agent technology to one of the most repetitive stages of chip design: resolving DRC violations, where a layout fails to meet the foundry's manufacturing rules. Instead of engineers manually fixing each violation, AI agents now automate the correction process.
By reducing the manual workload in the DRC stage, the two companies say the approach cuts the time required to reach tape-out — the final handoff of a completed chip design to the fab for production.
The development was presented in connection with TSMC's Open Innovation Platform (OIP) Ecosystem Forum, where the foundry and its design-automation partners showcase collaborative work on design tools and methodologies. Details on the specific magnitude of the tape-out period reduction were not provided in the reviewed materials.
Siemens, whose EDA division supplies design automation tools to the semiconductor industry, partnered with TSMC on applying agentic AI to this stage of the design flow.
The precise scope of the deployment — including which process nodes or design types are covered and how extensively AI agents have been adopted across production flows — has not been disclosed in the available information.
As AI agents move from experimentation into core EDA workflows, automated DRC correction stands out as one of the first production-ready applications, and the claim that Siemens and TSMC are automating this step while shortening tape-out periods is Mostly True.