CES 2026 showcases Nvidia, AMD and robots in ‘physical AI’ shift

The CES 2026 coverage makes clear that AI’s center of gravity is shifting from abstract model demos to machines that sense, think and act in the physical world. Nvidia touts its Rubin platform and new chips; AMD pushes MI455X accelerators and Venice CPUs; Qualcomm introduces a full-stack robotics SoC. Around them, a swarm of humanoid robots and autonomous systems from Boston Dynamics, LG and a long list of Chinese vendors turn “embodied AI” into the show’s main character. ([eet-china.com](https://www.eet-china.com/mp/a466439.html))

For the race to AGI, this matters because general intelligence ultimately has to cope with reality’s messiness, not just tokens on a server. The CES lineup shows an industry converging on a stack where large models run across data centers, vehicles, and mobile robots, connected by specialized sensors and edge compute. That accelerates feedback loops between simulation, deployment and real‑world data, which is exactly what you’d design if you were trying to bootstrap more general capabilities from today’s narrow systems. ([chinanews.com.cn](https://www.chinanews.com.cn/cj/2026/01-09/10548704.shtml?utm_source=openai))

Competitively, Chinese hardware and robotics firms are no longer bit‑players; they dominate the humanoid aisles and key perception components like 3D cameras and LiDAR. That suggests future AGI‑adjacent platforms—household robots, logistics systems, autonomous fleets—may be as dependent on Chinese hardware ecosystems as cloud AI is on US chips today.