Nvidia has announced a significant expansion of its robotics initiatives at CES 2026, unveiling a comprehensive suite of robot foundation models, advanced simulation tools, and cutting-edge edge hardware. This strategic move signals the company's ambition to establish itself as the default platform for generalist robotics, drawing a parallel to how Android became the dominant operating system for smartphones.

This push into robotics reflects a broader industry shift, as artificial intelligence transitions from cloud-based operations to machines capable of learning and reasoning within the physical world. This evolution is driven by advancements in cheaper sensors, sophisticated simulation technologies, and AI models that can increasingly generalize across diverse tasks.

Nvidia's Full-Stack Ecosystem for Physical AI

Nvidia provided detailed insights into its full-stack ecosystem for physical AI, which includes new open foundation models designed to empower robots with the ability to reason, plan, and adapt across various tasks and environments. These innovations aim to move beyond narrow, task-specific bots, with all models readily available on Hugging Face.

Key models introduced include:

  • Cosmos Transfer 2.5 and Cosmos Predict 2.5: These two world models are crucial for synthetic data generation and evaluating robot policies within simulation environments.
  • Cosmos Reason 2: A reasoning vision language model (VLM) that enables AI systems to perceive, comprehend, and interact with the physical world.
  • Isaac GR00T N1.6: Nvidia's next-generation vision language action (VLA) model, specifically engineered for humanoid robots. GR00T leverages Cosmos Reason as its core intelligence, facilitating whole-body control for humanoids to simultaneously move and manipulate objects.

Advanced Simulation and Workflow Management

Further enhancing its physical AI platform, Nvidia also introduced Isaac Lab-Arena at CES. This open-source simulation framework, hosted on GitHub, provides a critical environment for the safe virtual testing of robotic capabilities.

The platform directly addresses a significant industry challenge: the high cost, time consumption, and inherent risks associated with validating complex robotic tasks—from precise object handling to intricate cable installations—in physical environments. Isaac Lab-Arena tackles this by consolidating essential resources, task scenarios, training tools, and established benchmarks like Libero, RoboCasa, and RoboTwin, thereby creating a unified standard where one was previously lacking.

Supporting this extensive ecosystem is Nvidia OSMO, an open-source command center. OSMO acts as the connective infrastructure, integrating the entire workflow from data generation through training across both desktop and cloud environments.

Powerful Hardware and Strategic Partnerships

To power these advanced systems, Nvidia unveiled the new Blackwell-powered Jetson T4000 graphics card, the latest addition to its Thor family. Positioned as a cost-effective on-device compute upgrade, the Jetson T4000 delivers 1200 teraflops of AI compute and 64 gigabytes of memory, all while operating efficiently at 40 to 70 watts.

Nvidia is also strengthening its partnership with Hugging Face, aiming to make robot training more accessible without the need for expensive hardware or specialized expertise. This collaboration integrates Nvidia's Isaac and GR00T technologies into Hugging Face's LeRobot framework, connecting Nvidia's 2 million robotics developers with Hugging Face's 13 million AI builders. Notably, the developer platform's open-source Reachy 2 humanoid now directly supports Nvidia's Jetson Thor chip, allowing developers to experiment with various AI models without being tied to proprietary systems.

Nvidia's Vision for Accessible Robotics

The overarching goal behind Nvidia's extensive new offerings is to democratize robotics development and position itself as the foundational hardware and software vendor, much like Android's role in the smartphone market. Early indicators suggest this strategy is gaining traction.

Robotics is currently the fastest-growing category on Hugging Face, with Nvidia's models leading in downloads. Furthermore, prominent robotics companies, including Boston Dynamics, Caterpillar, Franka Robots, and NEURA Robotics, are already incorporating Nvidia's technology into their operations.