Cadence Design Systems has launched Cadence Maestro, a portfolio of AI models tuned for designing computer chips, deepening its open-weight work with Nvidia. The suite is built in part on the Nvidia Nemotron model and secured with Nvidia OpenShell safety technology.
The key pitch is control. Customers can run the models on the Cadence cloud or any public cloud, or keep them entirely in-house behind their own firewalls and data centers. They can also fine-tune the models on their own chip designs, an important point for companies guarding proprietary layouts.
Early results cited by Cadence point to up to 30 percent higher quality and up to five times better cost-efficiency than general-purpose models. Shares in Cadence rose 3.9 percent after the announcement, and the first Maestro models reach select early-access customers in the fourth quarter.
The approach contrasts with rival Synopsys, which announced on September 30 that it would develop a model with OpenAI, called GPT-Synopsys, running on OpenAI-hosted infrastructure. Cadence is positioning Maestro as model-agnostic, with open-weight and closed-weight approaches both in play.
Chip design is a demanding test for AI agents, which must arrange billions of transistors to meet cost and performance targets. As general-purpose models such as ChatGPT and Claude move into engineering work, and coding assistants such as Cursor become standard tools, purpose-built models that run behind a firewall could find a ready audience.
Source: Reuters via WNCY
