CES 2026 opened in Las Vegas in early January, and NVIDIA CEO Jensen Huang set the tone with the opening keynote, declaring the arrival of the Physical/Embodied AI Era. The session ran on two clear tracks: the next-generation Rubin platform and new autonomous-driving announcements.
The framing matters as much as the content. Physical AI repositions NVIDIA’s addressable market from the data center — where it already dominates — toward factories, roads, warehouses, and machines, where the incumbents are mechanical and electrical rather than computational.
From Generative AI to Physical AI
Huang’s declaration extends a narrative NVIDIA has been building for years: from GPU computing platform, to the training and inference infrastructure behind generative AI, to AI that operates in the physical world. In physical AI, the automated actor is no longer a software workflow but a robot, a self-driving vehicle, or any machine that senses and acts. Putting that declaration on the opening stage of CES tells the consumer-electronics and manufacturing supply chain directly: this is the theme of 2026.
The Keynote’s Two Concrete Tracks
- The Rubin platform: NVIDIA’s next-generation platform, the core of its going-forward compute roadmap and the computing base behind physical AI
- Autonomous-driving news: self-driving is one of the clearest commercialization paths for physical AI and overlaps heavily with CES’s automotive ecosystem
Neither track is surprising in isolation; together they describe a stack. Robots and vehicles need training compute in the cloud and inference compute at the edge, and NVIDIA is positioning to sell both ends of that pipeline.
Why Announce It at CES
Physical AI has a different customer base than cloud AI: the buyers are automakers, robotics companies, and equipment manufacturers — precisely the audience gathered in Las Vegas. Choosing CES over a developer conference is itself a market signal. For engineering teams, physical AI means integrating models, sensors, and real-time computing becomes a mainstream workload. For industry watchers, the rollout timelines of the Rubin platform and the driving announcements are the two threads to track on NVIDIA this year.
The gap to watch is execution. Declaring an era is cheap; shipping platforms that automakers and robot builders standardize on is not. Rubin’s adoption rate will be the measurable test of whether the Physical AI era is a schedule or a slogan.
Sources
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.