Nvidia announced a major expansion of its NVIDIA Agent Toolkit on Monday, integrating the PhysicsNeMo AI physics framework and updated CUDA-X libraries to enable the development of autonomous AI engineers for chip and system design. The update introduces new accelerated solvers and quantum chemistry capabilities, while the Nemotron 3 Ultra model achieved a 100% pass rate on key RTL debugging benchmarks. Major EDA players including Cadence, Synopsys, and Siemens are already embedding the technologies into their workflows, reporting up to 20 times faster multiphysics performance and 10 times faster library characterization. Samsung, TSMC, and Keysight are also leveraging the tools for computational lithography, thermal-stress analysis, and electromagnetic simulations. The move is underpinned by Nvidia's Vera CPU, which delivers up to 1.5 times higher throughput on critical verification workloads, creating a feedback loop where Nvidia uses its own processors to design next-generation chips.