Socionext Inc. is expanding its agentic AI initiatives into physical design, applying the Cadence InnoStack AI Super Agent to the physical design of a large-scale 3nm SoC.
Socionext said physical design is becoming more complex for large-scale SoCs using advanced process technologies. It said effectively integrating AI agents with advanced EDA tools and design flows is critical to addressing these challenges and improving design efficiency. The company is collaborating with Cadence on the work.
Socionext uses retrieval-augmented generation (RAG), Model Context Protocol (MCP) support, customized AI skills and AI agents to integrate internal design knowledge into physical design workflows. Current and planned applications include floorplanning, custom timing-analysis checks and future checklist automation, making proprietary expertise more accessible and reusable across projects and teams.
Hisato Yoshida, president and COO of Socionext, said the company expects to achieve up to 5X overall productivity gains through Cadence's agentic AI workflow. He said Socionext aims to combine its accumulated design knowledge, methodologies and engineering experience with Cadence technology to improve design quality and enable engineers to focus on higher-value innovation across the SoC development flow.
Socionext plans to connect its front-end design AI initiatives with its expanding physical design capabilities and apply the resulting work to additional SoC development projects. The company is also developing custom AI agents and engineering assistants that use internal expertise to strengthen productivity and knowledge reuse.
