Action tokens, real-time tokenization…physical AI…robots
Jensen Explains It All @CES2025
Two-minutes and forty seconds before NVIDIA’s co-founder and CEO Jensen Huang took to the stage at CES 2025, NVIDIA screened a video, which was all about tokens and robotics; tokens are what Andrew Ng has tabbed as the “new electricity” of AI.
If Huang’s keynote from last March was about everyone becoming a GenAI prompt engineer since the capitulation of computer code to GenAI prompting, then his lengthy CES 2025 keynote was another such watershed moment for accelerated computing via GPU’s, with a very special tilt toward robotics. Especially in view of the emergence of physical AI that’s driven along by what Huang calls “action tokens” and “real-time tokenization”.
“Robotics is here. Physical AI is here. This is not science fiction. It’s just really, really exciting. All the factories will be robotic. The factories will orchestrate robots, and those robots will be building products that are robotic.” —Jensen Huang
As in emphasis, the opening video concluded with: “[Tokens] is how intelligence is made. A new kind of Factory. Generator of tokens. The building blocks of AI.”
Immediately following the video’s conclusion, out strode Huang carrying a bagful of amazing new-product announcements that he doled out to the audience during his one-hour-plus oration.
Thor: Universal Robotics Computer
Huang is getting very effective at these long-form keynotes. Highly congenial, articulate, down-to-earth, brilliant, and a showman, he’s masterful at twisting technology into a new future and then backing it up with revolutionary new products. As he did a few months back in our profile on his newest quest: NVIDIA Set Sights on New Frontier: Advanced Robotics.
One of those new announcements he flashed out at CES 2025 is critically important to robotics. He held up what he called NVIDIA’s “Universal Robotics Computer”, the Jetson Thor, which will go on sale in early 2025 at a cost of $4,000 each. Thor will, he said, as embodied AI, revolutionize intelligence in humanoid robots or even AMRs or AGVs, which in turn will revolutionize manufacturing and logistics.
When he made that announcement, undoubtedly all of the fourteen humanoids that stood on stage with him at the opening of his keynote must all have had very sad looks on their robot faces. Huang’s announcement had just diminished them into instant legacy humanoids.
To wit, none had a Thor computer in its chest!
The Intersections of NVIDIA's Jensen Huang,
CES 2025, and the Rise of Intelligent Robots
Jensen Huang: the architect of AI acceleration
Huang and his co-founders, arguably the founding fathers of the GPU (graphics processing unit), have been on the hunt for more accelerated computing since 1999 when the company’s GeForce 256 hit the market.
Twenty-five years later (2025), NVIDIA has vaulted from those early graphics processing units (GPUs) for gaming to being the cornerstone of AI and machine learning innovation. “In fiscal year 2024, NVIDIA’s Compute & Networking segment, which includes data center products, generated more than $45 billion, a significant increase from $15.1 billion in fiscal year 2023. In the process, NVIDIA sold millions of H100 processors for $20-$40k each, depending on the configuration. Now valued at an astounding $3.3 trillion, with over 80% of its revenue derived from data centers, a worry has suddenly appeared in its idyllic AI space, one that NVIDIA has said it will address immediately: competition.
In a recent article from the Financial Times, Nvidia Bets on Robotics to Drive Future Growth: As Competition in the AI Chips Segment Stiffens, NVIDIA’s VP in charge of Robotics, Deepu Talla, “revealed that the market for robotics was close to ‘a tipping point’ and that ‘a ChatGPT moment’ for robotics and physical artificial intelligence was moments away…[and that] the company is looking to dominate the emerging AI robotics space.” Enter, Jetson Thor.
With an AI GPU industry that NVIDIA virtually built, Huang would be somewhat remiss to let the world of robotics and its convergence with AI pass it by in favor of catering to data centers only. More to the point, with competition in data centers continually on the rise, the real prize, he said during his CES 2025 keynote is worldwide GDP valued at $51 trillion. That’s the space where intelligent robots will one day rule.
Huang’s keynote addresses have become a staple for understanding the trajectory of AI. At CES 2025, his presentation went beyond showcasing NVIDIA’s latest hardware and software. He gave a masterclass on how AI-powered robots could redefine industries, ranging from healthcare and logistics to agriculture and entertainment. By democratizing access to AI technologies, NVIDIA is empowering developers and enterprises to create increasingly intelligent and capable robotic systems.
He made clear that NVIDIA wants to play a huge role in it all.
Others are rushing to play a role as well. OpenAI, for one, the originator of ChatGPT wants a piece of physical AI and is hurrying out a new robotics division that could one day compete with its existing partnership with Figure, a humanoid startup.
CES 2025: A showcase of intelligent robotics
The Consumer Electronics Show (CES) has long been a barometer for technological trends, and the 2025 edition was no exception. This year’s CES saw an unprecedented focus on intelligent robots, with companies unveiling systems that blend advanced AI, robotics, and real-world applications.
It seemed that every keynote given at CES was just another interesting AI embellishment to Huang’s masterclass. From warehouse robots that optimize supply chains to agricultural drones that analyze crop health, the applications further demonstrated the versatility and transformative potential of AI-powered robots.
The rise of intelligent robots
Intelligent robots—those capable of perceiving, reasoning, and interacting with their environments—are no longer confined to science fiction. Advances in AI, machine learning, and sensor technologies have enabled robots to become increasingly autonomous and capable.
And at the center of those enablements sits NVIDIA’s innovations.
Jensen Huang made sure during his keynote that everyone in the audience fully understood who made the AI GPU industry and that NVIDIA would be at the head of it for a long time to come.