Siemens uses agentic AI to accelerate chip design

The newly unveiled Questa One Agentic Toolkit is powered by five distinct AI agents. These include an RTL Code Agent for generating synthesizable code, a Lint Agent for checking design errors, a CDC Agent for clock domain crossing verification, a Verification Planning Agent, and a Debug Agent to assist in failure analysis. This allows the system to move beyond simple generative AI, which creates content, to agentic AI, which autonomously reasons and acts to solve complex problems. This initiative is a direct response to a critical bottleneck in semiconductor development. Finding and fixing bugs late in the process, particularly after the register-transfer level (RTL) stage, can lead to costly and time-consuming rework, derailing project schedules. The "shift left" approach aims to catch these structural and functional issues much earlier, when they are significantly easier and faster to fix. Siemens is collaborating with NVIDIA, using its Nemotron reasoning models and NIM microservices as a foundation for the toolkit's AI capabilities. This partnership aims to provide the robust, AI-driven workflows necessary to handle the increasing complexity of modern chip designs. The system is designed to be open, allowing customers to integrate their own data and utilize various large language models. Early feedback from partners like MediaTek suggests rapid adoption, with engineers becoming proficient within hours. Similarly, Tsavorite Scalable Intelligence reported that the toolkit enabled a swift uptake of formal property verification and automated the resolution of issues identified by the Lint Agent. This aligns with Siemens' broader strategy of using AI to address the industry's growing shortage of experienced engineers by embedding expertise directly into the design tools.