Cadence Launches AuraStack Agentic AI Platform for PCB and Advanced Packaging Design

Cadence has introduced the AuraStack Agentic AI platform to simplify PCB and advanced packaging design with intelligent automation and faster engineering workflows. The platform helps engineers reduce design time, improve productivity, and accelerate product development across multiple industries.

Published Date: 20 July 2026
Share : linkedin twitter facebook

In July 2026, Candence announced the launch of its new AuraStack AI Super Agent, with an AI-enabled design for PCB design and advanced packaging. It will also offer a full-flow agentic environment, boosting engineering productivity by 15 times successfully.

The Agentic AI runs on the Allegro AI Studio platform. It introduces the industry's first full-flow agentic AI environment for PCB and advanced packaging design, delivering two times faster time to market and up to a 15X leap in engineering productivity. The next-generation AI infrastructure is scaling its standards across hyperscale data centers, autonomous vehicles, and aerospace applications. The core engineering challenge has expanded way beyond the silicon die. The modern hardware performance is majorly interrupted by complex systems that connect, power, and cool those chips.

Engineering teams involved in designing advanced packaging and printed circuit boards (PCBs) often find themselves caught in a web of limited design tools, siloed workflows, and manual interactions. As per the data provided by the company, hardware engineers spend roughly 65% of their day simply navigating these disconnected tasks and cross-domain handoffs rather than actively designing.

A complex 3-nanometer or 2-nanocenter chip takes approximately 9 to 10 months to realize due to mature design automation, and finalizing the accompanying advanced system package can drag on for over 12 months. Aurastack PCB was designed to break this systemic issue.

The Bottleneck: Engineering Intelligence vs. Point Automation

For years, EDA tools have aided automated isolated point tasks- such as running an IR drop simulation or calculating trace impedance. Still, the major issue of human orchestration remains to stitch such workflows together. When a layout engineer finishes a board layout, it is traditionally "thrown over the wall" to a mechanical or thermal specialist. If a Celsius thermal analysis or a Sigrity signal integrity simulation flags a violation, the layout must be manually modified, restarting the verification loop.

Aurastack helps to shift this process easily from simple task automation to true engineering intelligence. The newly launched platform developed by Candence leverages an underlying “mental model” of design intent. The intelligent system of the platform immediately understands the relationship between design constraints, supply chain realities, and physical layouts, orchestrating workflows concurrently, instead of waiting for a user to trigger individual tools to do the same.

"The agentic AI democratizes engineering knowledge. It can ask ‘Do you want me to set up your thermal analysis?' and then it will set up your thermal analysis for you, whereas before that, the engineer might not know how to do it," says Vivek Mishra, Corporate VP, System Design and Analysis Group at Cadence.

Limelight over the Three-Layer Architecture

  • AuraStack’s framework is structured as a unified "three-layer cake" that integrates accelerated computing, native physics engines, and conversational AI:
  • The Orchestration Layer: The top layer features the agentic AI workflow. AuraStack accepts conversational natural language prompts (supporting multiple languages, including English, Japanese, and Korean) to autonomously establish goals, evaluate engineering tradeoffs, and execute multi-step modifications across domains.
  • The Physics Layer: The core consists of Cadence’s trusted multiphysics simulation and signoff engines. This includes the Celsius Thermal Solver, Clarity 3D Solver for 3D-EM, Sigrity X for signal and power integrity, alongside MSC Nastran and Marc for linear and non-linear mechanical stress, drop, vibration, and fatigue analysis.
  • The Compute Layer: The foundation is powered by custom accelerated hardware, leveraging NVIDIA Blackwell GPUs and NVIDIA CUDA architectures via the Cadence Millennium M2000 supercomputer platform.
  • By running these layers within a closed-loop environment, multiphysics analysis transitions from a late-stage gatekeeper to an active, real-time design guardrail.

Industry Leaders accepting the Agentic Shift

The gigantic software company is actively deploying AuraStack across prominent partners in the semiconductor, automotive, and industrial sectors, to reveal significant performance gains. NVIDIA has said that the Cadence AuraStack Al Super Agent and the Millennium M2000 Supercomputer enabled up to 20X faster multiphysics performance.

In advanced packaging environments, multidie system layout times help to highlight a major schedule risk. Through a multi-year development collaboration focused on automated substrate routing, the company reports that TSMC notes its deployment has allowed customers to unlock a 100X productivity boost in trace implementation while matching the quality of rigorous manual routing.

The automotive sector also observes a similar structural improvement. As per Candence, Forvia Hella reported that utilizing the platform’s AI-assisted component placement helped to reduce a 300-component placement task that required only four days of manual engineering layout down to just four minutes.

Mishra notes that AuraStack can meet the needs of engineers across all levels.

“AuraStack makes junior engineers more like middle-level engineers, and middle-level engineers into senior engineers, senior engineers into super engineers."

Latest Insights