Cadence introduced Maestro, a family of AI models for chip design. Reporting citing Reuters says it partly uses Nvidia Nemotron, supports further training on customer design data, and offers deployment through Cadence’s cloud, public clouds or customers’ own data centers. OpenShell technology is planned for agent security. The first models are for selected customers in Q4 2026, not an unrestricted public model download. Reported quality and efficiency gains are early company results; design-tool permissions, confidential-data boundaries and internal validation still require separate assessment.