Accelerating System Design with Real-Time Simulation, Powered by AI Physics

28 Oct 2025 • 4 minute read

Rising demand for AI infrastructure is driving faster innovation and smarter use of resources throughout the design lifecycle. Accelerated computing shortens design and simulation cycles, as well as streamlines workflows and amplifies human creativity through data-driven insights. Together, AI and accelerated computing empower engineers to explore ideas in real time and bring their visions to life. Cadence, with its GPU-accelerated Cadence Fidelity CFD Software, collaborated with NVIDIA to generate high-fidelity simulation data for airframe simulations, generating thousands of simulations in the span of weeks using NVIDIA GB200, available through the Cadence Millennium M2000 AI Supercomputer. This was followed by using NVIDIA PhysicsNeMo, an AI physics framework, to train a physics-accurate AI surrogate model for a digital twin that provides interactive what-if design changes and analyses for aircraft design.

This breakthrough in real-time simulation is a powerful example of the Cadence strategy for innovation, "The Three Layer Cake," in action. This strategic framework unifies Cadence's technology stack and drives our solutions. At the foundation is accelerated compute, exemplified by the Millennium M2000 AI Supercomputer, built with NVIDIA Blackwell systems. In the middle is Cadence's Fidelity CFD Software, enabling high-fidelity, physics-based modeling of the system under design. At the top sits AI, where frameworks like NVIDIA PhysicsNeMo and Cadence's AI-driven design intelligence transform simulation data into interactive, predictive digital twins. Combined, these layers form a cohesive platform that empowers engineers to design, simulate, and optimize complex systems faster and more intelligently than ever before.

A demonstration of the technology shows real-time airframe performance simulation while varying the design configuration. Other applications, including automotive aerodynamics or aeroacoustics, 3D-IC thermal and electromagnetic analysis, and data center thermal analysis, are possible. See the first demos at the NVIDIA GTC in Washington, D.C., October 27 – 29.

How AI for Physics Is Transforming Engineering Design

Computational fluid dynamics (CFD) is a cornerstone of modern engineering. It allows designers to simulate the flow of fluids—like air over a plane's wings or fuel through an engine—to predict performance, identify issues, and optimize designs. However, traditional CFD methods are incredibly resource-intensive. Historically, running a single high-fidelity simulation on conventional computing systems can take days or even weeks, limiting the number of design iterations engineers can perform. Applying AI technology speeds the calculations and turnaround time, making real-time what-if design analysis practical.

High-quality results from AI are dependent on accurate and representative training data, in sufficient quantities. The availability of such data for computational engineering purposes is relatively limited in comparison to typical data used to train foundational AI models. In this example, the Cadence Fidelity CFD Software, accelerated on the Millennium M2000 AI Supercomputer, produced the high-quality dataset for the NVIDIA PhysicsNeMo framework. Thousands of detailed, time-dependent simulations were computed in a matter of weeks, with each simulation comprising hundreds of millions of degrees of freedom. This volume of high-quality data, generated from the design itself, is critical to being able to trust the AI's predictions.

The collaboration between NVIDIA and Cadence addresses these challenges head-on. By leveraging GPU acceleration and AI, this integrated solution fundamentally changes the speed and scale of engineering simulation.

Cadence and NVIDIA Transform Aerospace and Automotive Design with AI Physics

At GTC Washington, D.C., NVIDIA will unveil groundbreaking advancements in AI-powered simulation, transforming aerospace and automotive design with up to 500X faster engineering workflows. Cadence is at the forefront of this transformation, leveraging its Fidelity CFD Software with Millennium M2000 AI Supercomputer built on NVIDIA Blackwell to empower aerospace leaders. By combining high-fidelity multiphysics simulations with modern accelerated computing, Cadence enables rapid design iteration, enhanced efficiency, and optimized performance for next-generation systems. Cadence, with NVIDIA, is accelerating innovation and redefining the future of computational engineering.

Shaping the Future of AI Infrastructure

In addition, NVIDIA unveiled the NVIDIA AI Factory Research Center in Virginia, which will leverage the NVIDIA Vera Rubin platform and the NVIDIA DGX blueprint to enable gigawatt-scale AI factory design and build-out.

To support the design and operational excellence of this groundbreaking facility, Cadence is creating a Reality DC Platform of a high-fidelity digital twin of the research center for design and operations. The Cadence Reality DC Platform and integrated NVIDIA Omniverse libraries provide a physics-based simulation environment that enables engineers to model thermal, energy, and airflow dynamics across the entire infrastructure—from chip to chiller. Cadence is integrating the NVIDIA Omniverse libraries across our system, AI, and bio simulation products. By integrating computational fluid dynamics (CFD) and multiphysics analysis, the Cadence Reality DC Platform allows teams to explore design configurations, predict failure scenarios, and optimize operational strategies before physical implementation. The result is a resilient, energy-efficient AI factory blueprint that accelerates development, reduces costs, and supports sustainability goals.

See the Latest Demos at NVIDIA GTC Washington, D.C.

Join us at NVIDIA GTC in Washington, D.C., from October 27 – 29 at the Walter E. Washington Convention Center, Booth 728.

We will be highlighting our AI and system design solutions:

AI for Design and CAE Acceleration: Explore Cadence Fidelity CFD Software plus Millennium M2000 AI Supercomputer with the NVIDIA PhysicsNeMo framework
Digital Twins: Learn about designing and operating data center digital twins with the Cadence Reality Digital Twin Platform and the NVIDIA DGX GB200 NVL72 SuperPOD model
Agentic AI: Discover how Cadence agentic AI is transforming silicon, system, and drug design—using NVIDIA AI technology