Cadence Design And Nvidia Team To Create AI Data Center Digital Twin

The data center industry is experiencing unprecedented growth, with demand skyrocketing and rack power requirements increasing exponentially. In just the last year, rack power requirements have surged from approximately 12 kilowatts per rack to over 125 kilowatts, with predictions of reaching a gigawatt per rack in the next two to three years. This rapid escalation is driven by the increasing adoption of artificial intelligence (AI) and the need for more powerful computing systems, such as Nvidia’s Rubin Ultra.

As power demands rise, so do the cooling requirements, leading to the adoption of liquid cooling systems and the need for more space to accommodate cooling distribution units. The modern AI factory is undergoing a significant transformation, with complex interactions between various components, including power and cooling systems, that cannot be simulated by humans alone. This is where AI and digital twins come into play, enabling the simulation and optimization of these complex systems.

Digital Twins: The Future of Data Center Simulation

Cadence, a leading provider of digital twin technology, sees the data center as just the starting point for applying digital twins to simulate and optimize a broad spectrum of systems. The company’s Reality Digital Twins Platform is being used by Nvidia to simulate its own supercomputing data centers. Cadence has also adopted Nvidia’s Omniverse as its digital twin collaboration platform, demonstrating the growing importance of digital twins in the data center industry.

While digital twins are widely used in semiconductor design, their adoption in data center simulation is still in its early stages. According to Cadence, only 20% of data center systems are currently being simulated before construction or in operation. However, as the industry prepares for the future of AI, it is likely that this number will increase dramatically over the next three to five years. Cadence believes that digital twins will become an absolute requirement for future data centers, especially AI factories.

Simulating Complex Systems

The recent AI Infra Summit in Sunnyvale, California, saw the announcement of a comprehensive physics model developed by Nvidia and Cadence to enable the simulation of a DGX SuperPOD with GB200 GPUs. This development brings the Cadence Reality DC platform to contain over 14,000 devices from over 750 vendors in its library of reference designs and workflows. This database enables the simulation of customized, accurate operational behavior of air- and water-cooled components in the AI factory, resulting in faster design, lower risk, and better operational efficiency of gigawatt AI factories.

A complete database of AI factory models and reference designs ensures an accurate representation of how the Nvidia DGX system will behave in the data center. This level of simulation and modeling will become increasingly important as the industry continues to evolve and adopt more powerful computing systems. In the near future, designers will likely wonder how the early days of AI data centers were possible without rigorous physics-based simulation.

What’s Next for Digital Twins?

The addition of support for the DGX SuperPOD to Cadence Reality DC is just the beginning of a more comprehensive capability that will extend to NVL72, Rubin, and other roadmap technologies that populate the AI factory. As the industry continues to advance, it is likely that digital twins will play an increasingly important role in the design, simulation, and optimization of complex systems. With the growing demand for more powerful and efficient computing systems, the adoption of digital twins is poised to revolutionize the data center industry and beyond.