Siemens Launches AI for Chip Design

The new system, called the Questa One Agentic Toolkit, introduces five types of AI agents to automate different stages of the chip verification process. These include an agent that generates synthesizable code from natural language, a lint agent for error checking, and a debug agent for root cause analysis. This approach aims to tackle the widening verification productivity gap driven by the increasing complexity of chip designs. Siemens has built the workflows using NVIDIA Llama Nemotron and NVIDIA NIM, enabling the AI agents to reason, plan, and execute complex verification tasks. This allows the system to understand the relationships between designs, testbenches, and specifications in real-time. The toolkit is designed to integrate with mainstream AI coding applications like GitHub Copilot and Claude Code. The register-transfer level (RTL) is a design abstraction that models a digital circuit in terms of the data flow between hardware registers and the logical operations performed on that data. The "sign-off" phase is a critical step that verifies the RTL implementation is structurally sound and ready for manufacturing, checking for issues like unsynchronized signals between different clock domains. Early trials of the toolkit have shown significant productivity gains. Engineers at MediaTek, a major chip designer, were able to become proficient with the tool within hours and complete tasks that would typically take days. This acceleration is crucial as the semiconductor industry faces growing demands for more complex and powerful chips. This move is part of a broader industry trend of integrating AI into Electronic Design Automation (EDA) tools to manage soaring chip complexity. Siemens itself has been actively incorporating AI, with its EDA division reporting a 19% sales increase last year, outpacing competitors. Other AI-powered tools in their portfolio, like the Solido Simulation Suite, have demonstrated up to a 19x speedup in verification for certain designs. The AI agents in the Questa One toolkit are designed to operate within boundaries defined by the customer, maintaining human oversight at key decision points. This ensures that engineers can leverage AI to automate repetitive tasks while still applying their expertise to critical judgments, aiming to enhance, not replace, their role. The toolkit is currently available through an early access program and connects with a suite of Siemens' verification tools, including Tessent software and the Veloce CS hardware-assisted verification system. This integration creates a more cohesive and automated workflow from design to final sign-off.