Nvidia Uses Own Chips to Accelerate Chip Design

Nvidia is running chip design software on its own Vera CPUs, partnering with Cadence and Synopsys to optimize their electronic design automation platforms. Early tests show a 1.5x performance boost for key verification and simulation tools, with plans to use the gains to accelerate development of its next-generation Rosa CPU. Nvidia is also expanding AI agent involvement in chip design by integrating its PhysicsNeMo and CUDA-X libraries into its Agent Toolkit. New libraries including cuISS, cuDSS, and cuEST bring accelerated solvers and quantum chemistry capabilities to agentic workflows, enabling AI agents to run complex simulations far faster than human engineers.