SAN FRANCISCO, July 15 (Reuters) - Cadence Design Systems on Wednesday launched an artificial-intelligence "super agent" that designs printed circuit boards and chip packages, extending the company's push to automate more of the engineering process.
The tool, called AuraStack, lets engineers describe their goals in plain language, then plans and carries out the work using Cadence's existing software tools to lay out and virtually test circuit designs. Cadence said Nvidia chips will accelerate the AI work.
Cadence said AuraStack can cut time to market by up to half and lift productivity on individual tasks as much as 15-fold. The AI agent for circuit boards and chip packaging follows similar offerings earlier this year to help speed up the design of chips themselves.
In a demonstration, Cadence showed an engineer using the tool to rework a 5G smartphone's circuit board to build a cheaper version for a new market. It recommended consolidating components for a 28% cost saving, then found a lower-cost power-management chip that worked with circuit board design.
"The bottleneck isn't automation. It's really engineering intelligence," Michael Jackson, Cadence's corporate vice president and general manager for system design and analysis, said in an interview, referring to the reasoning across cost and performance trade-offs that the system is designed to handle.
Cadence named Nvidia, Taiwan Semiconductor Manufacturing Co and Schneider Electric among early users.
Jackson said Cadence customers can pair AuraStack with the AI model of their choice, including OpenAI's ChatGPT, Google's Gemini or Anthropic's Claude, or open-source models. Pricing will follow a consumption-based model based on how hard the AI models work, and still require Cadence's underlying tools, Jackson said.
AuraStack will be available this year, with the rollout to be completed in September, Jackson said.
(Reporting by Stephen Nellis in San Francisco; Editing by Sanjeev Miglani)
