Synopsys unveils agentic chip design tools built with Nvidia

Synopsys, Inc. (NASDAQ: SNPS) announced a set of autonomous, AI-driven engineering workflows at the 2026 DAC Chips to Systems Conference, developed in collaboration with Nvidia.

The announcements center on a fully autonomous design verification agent built on Synopsys' AgentEngineer technology and Nvidia's agentic AI infrastructure, including the Nvidia Agent Toolkit, Nvidia Nemotron 3 Ultra model, and OpenShell runtime. According to the company, the workflow compresses chip verification from weeks to hours, delivering up to 50 times faster time-to-validated RTL and a 20% improvement in coverage compared to traditional verification workflows not powered by AgentEngineer technology.

Synopsys also demonstrated its first autonomous computer-aided engineering workflow for electronics thermal analysis, using Ansys Icepak simulation software and Nvidia CUDA-X libraries. The workflow autonomously handles simulation setup, pre-processing, and post-processing.

In analog and mixed-signal design, Synopsys' Custom Compiler Layout Synthesis tool, combined with AgentEngineer, claims up to 3 times productivity improvement across multi-step analog design flows.

The company said its GPU-accelerated product portfolio now spans more than 20 EDA and multiphysics tools. Specific performance figures cited include an 18 times speedup for PrimeSim SPICE circuit simulations using Nvidia GPUs, a 10 times speedup for Ansys Lumerical FDTD electromagnetic simulation, and up to 200 times faster machine-learned force field simulations in Synopsys QuantumATK using Nvidia Blackwell GPU infrastructure.

Synopsys said customers are currently evaluating the agentic capabilities, with general availability planned for the second half of 2026. The information is based on a press release issued by Synopsys.

You May Also Be Interested In

• ADARx Pharmaceuticals files for proposed IPO
• Bank of America to redeem CAD1.425 billion in Canadian senior notes
• Novartis pelacarsen trial fails to meet primary endpoint