NVIDIA Unveils New AI Technologies and Robotics at CES 2026
LAS VEGAS, Jan. 5, 2026 (GLOBE NEWSWIRE) — NVIDIA (NVDA) has made significant strides in the field of robotics by announcing new open models, frameworks, and AI infrastructure designed for physical AI at the Consumer Electronics Show (CES) 2026. This development is poised to accelerate the robotics industry, enabling the creation of next-generation robots capable of performing multiple tasks swiftly and efficiently.
New Collaborations and Innovations
Leading robotics firms such as Boston Dynamics, Caterpillar, Franka Robots, Humanoid, LG Electronics, and NEURA Robotics are leveraging NVIDIA technologies to present a diverse lineup of AI-driven robots. CEO Jensen Huang remarked, “The ChatGPT moment for robotics is here. Breakthroughs in physical AI — models that understand the real world, reason, and plan actions — are unlocking entirely new applications.”
- NVIDIA Cosmos: New open models including Cosmos Transfer 2.5 and Cosmos Predict 2.5 enable efficient synthetic data generation for training robots.
- NVIDIA Isaac GR00T: This model enhances reasoning capabilities in humanoid robots, offering full body control.
- Collaboration with Hugging Face: Integration of NVIDIA's open models into the LeRobot framework supports the open-source robotics community.
Advancements in Robot Learning and Reasoning
Transforming traditional fixed-function machines into reasoning generalist-specialist robots requires extensive resources. To facilitate this transition, NVIDIA is developing open models to streamline development and reduce the associated costs. These models are available on Hugging Face, further democratizing access to advanced robotics technology.
- NVIDIA Cosmos Reason 2: An innovative vision language model that allows machines to visually perceive and interact with their environment.
- Franka Robotics and NEURA Robotics: These companies are utilizing GR00T-enabled workflows to enhance robot capabilities.
- Salesforce's Use Case: Employing Cosmos Reason to efficiently analyze video footage, decreasing incident resolution times by 50%.
Open-Source Frameworks for Robotics Development
Current robotics development workflows often face complexities and fragmentation. To address this, NVIDIA has released open-source frameworks such as Isaac Lab-Arena and OSMO, which simplify robotics training and simulation processes.
- Isaac Lab-Arena: This open-source framework enables large-scale robot policy evaluation and benchmarking.
- OSMO: A cloud-native orchestration framework that streamlines workflow management for developers.
Next-Level Humanoid Robots Powered by NVIDIA Jetson Thor
The introduction of NVIDIA Jetson Thor is revolutionizing the capabilities of humanoid robots. Many developers are showcasing their latest innovations integrated with Jetson Thor at CES, indicating a considerable leap in performance and functionality.
- NEURA Robotics: Launching a Gen 3 humanoid with Porsche design for enhanced dexterity.
- LG Electronics: Unveiling a versatile home robot designed for various indoor tasks.
Introducing the NVIDIA Jetson T4000 Module
NVIDIA’s latest Jetson T4000 module, built on the Blackwell architecture, provides a robust platform for high-performance AI computing in robotics. Priced at $1,999, this module offers a substantial performance increase, delivering 1,200 FP4 TFLOPS and 64GB of memory while maintaining an energy-efficient profile.
Conclusion
The advancements made by NVIDIA (NVDA) at CES 2026 highlight the company’s commitment to driving innovation in the robotics sector. By providing industry partners with powerful tools and technologies, NVIDIA is paving the way for more sophisticated, AI-driven solutions across various applications.