Cadence and Nvidia expand partnership to close robotics simulation gap

At a glance:

• Cadence and Nvidia announced an expanded partnership at a Cadence conference in Santa Clara.
• The collaboration ties Cadence’s high‑fidelity physics engines to Nvidia’s Isaac and Cosmos simulation platforms.
• Goal is to generate more accurate robot training data and speed real‑world deployment of AI‑driven robots.

What the partnership delivers

The two CEOs unveiled the deal at a Cadence conference in Santa Clara, California, describing a joint stack that fuses Cadence’s multiphysics simulation engines with Nvidia’s AI training pipelines. Cadence, traditionally known for chip‑design software, contributes physics models that capture material deformation, fluid flow, and surface contact. Nvidia brings its Isaac open‑source simulation libraries and Cosmos open‑world models, as well as the Jetson family of edge AI hardware for deployment.

The workflow is designed to run from world‑model training through physics‑accurate simulation, then feed real‑world deployment feedback into the loop. AI agents coordinate each stage, ensuring that the generated training data reflects the nuances of physical interaction. As Cadence CEO Anirudh Devgan put it, “The more accurate the generated training data is, the better the model will be.” Nvidia CEO Jensen Huang added, “We’re working with you across the board on robotic systems.”

Why accuracy matters for robot AI

Training robots in simulation has long been a cost‑effective shortcut compared with physical trials, but the fidelity of the simulation directly limits how well a model transfers to reality. Inadequate physics can lead to brittle behaviors when a robot encounters real‑world friction, compliance, or fluid dynamics. By leveraging Cadence’s proven multiphysics engines—originally used in aerospace, automotive, and semiconductor design—the partnership promises training data that mirrors true material behavior.

Better data translates to faster iteration cycles: developers can train models on virtual environments, validate them against high‑precision physics, and then deploy on Nvidia Jetson devices with confidence. This reduces the need for expensive hardware‑in‑the‑loop testing and accelerates time‑to‑market for applications ranging from warehouse automation to autonomous inspection.

Broader industry context

Nvidia’s move fits a larger strategy of building deep simulation ecosystems across industrial engineering. The company recently announced collaborations with Siemens and Dassault Systèmes to create industrial AI platforms and digital twins. Those ties, together with the Cadence partnership, signal a push to make simulation a core component of AI‑driven product development.

For Cadence, the deal marks a significant expansion beyond its traditional chip‑design market into the AI infrastructure layer. Demand for realistic robot training data is rising as manufacturers seek to scale autonomous systems, and Cadence’s physics expertise positions it to capture a slice of that emerging market.

Looking ahead

Both firms indicated that the joint stack will be available to developers soon, though exact release dates were not disclosed. The collaboration is expected to evolve with additional modules, such as reinforcement‑learning environments and tighter integration with Nvidia’s Omniverse platform. Stakeholders will be watching for benchmark results that quantify the reduction in sim‑to‑real transfer error, a metric that could become a new industry standard.

The partnership underscores a broader trend: as AI models become more capable, the quality of the simulated world they train in becomes a decisive competitive factor. Companies that can deliver high‑fidelity, scalable simulation pipelines are likely to shape the next wave of robotics deployment.