Serving technology enthusiasts for more than 25 years. TechSpot is the place to go for tech advice and analysis.
In one of the most highly anticipated and attended CES keynotes ever, Nvidia CEO Jensen Huang announced a wide range of announcements spanning a number of the hottest tech topics, including AI and robotics.
Clad with a Las Vegas-glitzed version of his trademark leather jacket, the tech leader discussed the company’s new GeForce 50 series graphics cards as well as the new Nemotron AI Foundation model families and AI blueprints to power AI-powered agents.
The tech industry leader also highlighted extensions to Omniverse’s digital twin and simulation platform which extends AI in the physical world as well as new certifications for its auto-driving platform. He also introduced a desktop-sized mini AI supercomputer powered by the Grace Blackwell graphics processor, called Project Digits. It was a lot of information to take in.
Cosmos, a set foundation models and platform capabilities, was one of the most interesting – but probably least understood – announcements. Cosmos is a set of world foundations models, advanced tokenizers and safety guardrails as well as an advanced video processing pipeline. Its goal is to bring the advanced outcomes and training capabilities of generative AI into the physical realm.
In other words, instead of using generative AI to create new digital outputs based on training across billions of documents, images, and other digital content, Cosmos can generate new physical actions – let’s call them analog outputs – by leveraging data it has been trained on from digitally simulated environments.
While the concept is complex, the real-world implications are both simple and profound. For applications like robotics, autonomous vehicles, and other mechanical systems, Cosmos enables these systems to react to physical stimuli in more accurate, safe, and helpful ways. For instance, humanoid robots can be trained to physically replicate the most effective or safest way to perform a task, whether it’s flipping an omelet or handling parts on a production line. Similarly, an autonomous car can dynamically adapt to varying situations and environments.
Also see: AI Agents Explained: The Next Evolution in Artificial Intelligence
Much of this type of training currently relies on manual efforts, such as filming humans performing the same action hundreds of times or having autonomous cars drive millions of miles. Even then, thousands of people must spend significant time hand-labeling and tagging those videos. With Cosmos, these training methods can be automated, dramatically reducing costs, saving time, and expanding the range of data available for the training process.
Nvidia Cosmos combines generative models with a data curator, tokenizers and a framework for accelerating physical AI development.
Cosmos is an extension of Nvidia’s Omniverse digital simulator environment. It translates digital physics from models and systems created in Omniverse to physical actions in the world. This distinction, while seemingly insignificant, is crucial because it allows Cosmos to produce GenAI powered physical outputs.
The world foundation models are at the core of Cosmos. They are built from millions hours of video content and possess a deep understanding of the physical universe. Cosmos integrates the digital models created in Omniverse of physical objects and environment into these world-foundation models and produces photorealistic videos of how the models will behave in real scenarios. These videos can then be used as synthetic data to train models in robotic systems, autonomous vehicles, and other GPU powered mechanical systems. The result is a system that can respond more effectively to diverse environments.
Nvidia’s CEO Jensen Huang clearly wanted to tell us something when he outlined the development of AI technologies during his keynote speech at CES 2025. From perception AI to generative AI to agentic AI and the rise of the physical AI.
Nvidia’s Cosmos world foundation model is also available for free, to encourage robotics and autonomous vehicle advancements as well as further experimentation.
The immediate impact of Cosmos is limited in the short-term, as it targets a niche audience that develops advanced robotics and autonomous vehicle applications. In the long run, however, its impact could be profound. It could speed up the development of this product category and improve the accuracy and safety.
But more importantly, it shows Nvidia’s ability in anticipating and preparing for emerging tech trends like robotics. This also highlights the ongoing, but often overlooked, transformation of Nvidia from a hardware company to a software firm building platforms for new applications. These developments provide important and intriguing insights for those who are curious about the direction of the company and how it plans on maintaining its impressive growth.
Bob O’Donnell, founder and chief analyst at TECHnalysis Research, LLC, is a technology consultancy firm that offers strategic consulting and market-research services to the professional financial community and the technology industry. You can follow him @bobodtech (19659021]