Siemens deploys AI for chip design

The newly announced Questa One Agentic Toolkit is part of a broader Siemens EDA AI system that leverages both generative and agentic AI to enhance productivity across the entire electronic design automation (EDA) workflow. This system allows for the integration of a company's own EDA data and the creation of custom AI-driven workflows, deployable either on-premises or in the cloud for data security. Agentic AI represents a shift from simple automation to autonomous software agents that can reason, plan, and execute complex multi-step tasks traditionally requiring significant human engineering involvement. In chip design, these agents can analyze goals, adapt strategies, and manage detailed processes from RTL generation to verification. This approach aims to move engineers away from routine tasks to focus on higher-level architectural decisions. The register-transfer level (RTL) is a critical abstraction in digital circuit design where the flow of signals between hardware registers and the logical operations on those signals are defined using a hardware description language (HDL). RTL sign-off is the process of verifying that this high-level description is structurally sound and ready for the next stages of synthesis and physical implementation, checking for issues like unsynchronized signals between clock domains. Siemens' toolkit specifically targets this RTL phase with five initial AI agents: an RTL Code Agent for code generation, a Lint Agent for error checking, a CDC Agent for clock domain crossing verification, a Verification Planning Agent, and a Debug Agent for failure analysis. Early user MediaTek reported that its engineers, even with limited prior experience, became proficient with the toolkit within hours and saw "immediate and significant" productivity gains. The system integrates NVIDIA NIM microservices and NVIDIA Nemotron models to power its AI-driven workflows. This collaboration is part of a wider industry trend where AI is being used to tackle the increasing complexity and cost of designing smaller, faster, and more efficient chips. The global semiconductor market is projected to grow by 15% in 2025, largely driven by the influence of AI. Siemens has been incorporating AI into its EDA tools for years, from pattern recognition in 2005 to the reinforcement learning used today. The new architecture moves from individual tool databases to a centralized multimodal data lake, allowing AI engines to access information across different tools and create a "data flywheel" effect that boosts productivity with each interaction.