Cadence Design Systems just showed up to Computex 2026 with something that makes traditional chip verification look like sending a fax. The company unveiled what it calls the industry’s first fully autonomous Level-5 ChipStack AI Super Agent on June 1, 2026, built on Nvidia’s Nemotron foundation models and secured by the Nvidia OpenShell runtime.

The headline number: RTL validation cycles reduced by over 40 times. In English, that means a standard five-week verification process now takes less than a day.

What the ChipStack AI Super Agent actually does

Cadence’s new agent operates at what the company designates “Level-5 autonomy,” meaning it can independently execute complex chip design and verification workflows without constant human hand-holding. The system can run hundreds of dynamic simulations autonomously, requiring only human oversight rather than human direction at every step.

To put that in perspective, thousands of Nvidia engineers already run billions of compute hours annually on chip verification. The technology is powered by Nvidia’s Nemotron foundation models, which provide the underlying AI reasoning capabilities. The Nvidia OpenShell runtime handles the security layer.

From 10X to 40X in four months

This isn’t Cadence’s first swing at autonomous chip design. The company debuted an initial version of the ChipStack AI Super Agent back in February 2026, which reported a tenfold productivity increase among early users including Nvidia and Altera.

Cadence acquired the ChipStack startup that formed the foundation of this technology, then layered its existing electronic design automation (EDA) expertise on top. The company also announced in April 2026 a partnership with Google Cloud for enhanced AI-driven engineering solutions, adding another hyperscaler to its alliance network alongside Nvidia.

Early access for the Level-5 capabilities and what Cadence calls AgentStack orchestration is expected in the second half of 2026.

What this means for investors

For Cadence specifically, the strategic positioning here is deliberate. The company has pursued what it describes as a dual strategy: designing tools for AI workloads while simultaneously using AI to improve those very tools.

By building directly on Nvidia’s models and runtime, Cadence is tying its autonomous agent ecosystem to the company that currently dominates AI compute. If Nvidia’s models improve, Cadence’s tools improve almost by default. But it also means Cadence’s most advanced capabilities are linked to a single AI platform provider’s roadmap.