SAN JOSE: Cadence Design Systems launched its Maestro AI models on Monday, letting semiconductor companies run specialised chip-design software in their own data centres or via cloud services.
The Nvidia-backed portfolio uses technology from the chipmaker’s open-weight Nemotron models. It also incorporates Nvidia’s OpenShell security architecture. Cadence said customers could operate the models behind their own firewalls.
Companies can alternatively use Cadence’s cloud infrastructure or supported public cloud services. The system lets chipmakers customise models with their proprietary designs and engineering data.
Cadence claims initial testing delivered design results up to 30% better than general-purpose AI models. It also reported up to five times greater token efficiency.
The company has not published detailed, independent benchmarks to support those figures. Maestro works with Cadence’s AI agents and electronic design automation (EDA) tools.
These systems can interpret engineering requirements, generate design content and assess results against technical constraints.
The launch expands Cadence’s partnership with Nvidia as chip-design software companies introduce AI agents to automate engineering tasks.
Read: Huawei Ascend 960 Starts Annual AI Chip Upgrade Cycle
Rival Synopsys announced a different approach on September 30, partnering with OpenAI. Their planned GPT-Synopsys model will operate on OpenAI-hosted infrastructure.
The companies also agreed to share revenue from the product. Cadence Senior Vice President Paul Cunningham said the Maestro platform would support different AI model providers.
Read: AMD Personal Supercomputer; Local AI Workstation; Instinct MI350P.
He said the company was exploring both open-weight and proprietary models, but he gave no details on potential commercial partnerships.
Cadence plans to provide the first Maestro models to selected early-access customers during the fourth quarter of 2026.
