Cadence unveils Maestro AI chip-design models, first units in Q4
Cadence says early use showed up to 30% higher quality and up to 5 times better cost efficiency than general-purpose models.
Cadence Design Systems has unveiled Cadence Maestro, a family of AI models for semiconductor design, with the first models to be supplied to some customers in the fourth quarter, according to Reuters.
The models were developed partly using Nvidia's open-weight Nemotron models, and Nvidia's OpenShell security technology was applied to Maestro. Customers can run the models on Cadence's cloud or on public clouds, or on-premises inside their own firewalls and data centers, and can further train them on their own chip design data.
Cadence said early use results showed up to 30% higher quality and up to 5 times better cost efficiency than general-purpose models.
Not tied to a single model
Paul Cunningham, a Cadence senior vice president, said Maestro is not tied to any specific model and that the company is considering partnerships with commercial AI research and development companies.
The launch comes as rival Synopsys pursues its own AI work in chip design. In September 2026, Synopsys agreed with OpenAI to jointly develop AI models and share revenue, with their GPT-Synopsys model based on OpenAI infrastructure. Both Cadence and Synopsys have recently been working on enabling AI agents to operate design tools on engineers' behalf to speed up work.
Quick answers
When will the first Cadence Maestro models be available?
The first Maestro models will be supplied to some customers in the fourth quarter.
Where can customers run Cadence Maestro?
On Cadence's cloud or public clouds, or on-premises inside their own firewalls and data centers.
What results did Cadence report from early use?
Cadence said early use showed up to 30% higher quality and up to 5 times better cost efficiency than general-purpose models.