The Cadence-Nvidia partnership in robotics

      The two companies revealed an expanded partnership during a Cadence conference in Santa Clara on Wednesday. The aim is to enhance the accuracy of robot training data, enabling physical AI systems to achieve real-world deployment more swiftly.

      Cadence Design Systems and Nvidia have formed a partnership designed to address a longstanding issue in robotics: the disparity between how robots learn in computer simulations and their actual performance in the real world.

      The collaboration was announced by the CEOs of both companies at the Cadence conference in Santa Clara, California. It combines Cadence’s advanced physics simulation engines with Nvidia’s AI training platforms, including the open-source simulation libraries of Isaac and the open-world models of Cosmos.

      Cadence is primarily recognized as a leading provider of software for designing cutting-edge computing chips. However, the company also produces physics engines that simulate real-world material interactions, metal deformation, fluid dynamics, and surface contact.

      These simulations have applications in aerospace, automotive, and semiconductor design, and they are now being utilized to address the challenge of generating the training data necessary for robot AI systems to learn how to manipulate objects and navigate physical spaces.

      Training robots in simulations is more efficient and cost-effective than training them in the real world, but the utility of the training data depends on the accuracy of the physics engine used.

      “The more precise the generated training data, the higher the quality of the model,” stated Cadence CEO Anirudh Devgan at the Santa Clara conference.

      Nvidia CEO Jensen Huang outlined the partnership’s extent: “We’re collaborating with you on robotic systems across the board.”

      The integrated solution will combine Cadence’s multiphysics simulation with Nvidia’s model training pipelines, deploying the outcomes on Nvidia’s Jetson robotics and edge AI hardware.

      The result is a streamlined workflow that moves from world-model training through physics simulation to real-world deployment feedback, all coordinated by AI agents throughout the entire process.

      This announcement fits into a larger trend of Nvidia forming extensive simulation partnerships within industrial engineering. The company has also established partnerships with Siemens and Dassault Systèmes to create industrial AI platforms and virtual twins.

      For Cadence, entering the robotics application domain marks a significant extension of its simulation software into the AI infrastructure layer at a time when the demand for precise robot training data is rapidly increasing.