February 12, 2026 -
Future edge AI advances will accelerate embodied intelligence, humanoid robots, heterogeneous SoCs, and real-time sensing.
By Amit Kumar, Cadence Design Systems Inc
There is a tremendous push to process data at edge nodes rather than in the cloud to reduce latency and improve privacy, reliability, and overall cost. The number of platforms at the edge is growing rapidly to accommodate these workloads. In the physical AI domain, the automotive and robotics segments are fueling this growth due to the need for real-time decision-making while keeping compute costs low. Silicon providers are reshaping their product roadmaps to design chips capable of addressing such requirements.
These chips become platforms embedded within intelligent machines, and these “embodied agents”, which have a body and act within a physical or simulated environment, are on a significant growth trajectory. Classic examples include automotive ADAS, digital cockpits, Industry 4.0 for factory automation, and consumer robots.
More recently, there’s a new growth engine—humanoid robots. These physical AI machines rely on embodied intelligence, vary in shape and size, and sense, interpret, and act in the physical world. They also demand unprecedented levels of compute, sensing, and system integration. As these domains mature, expect to see greater demand for heterogeneous compute, custom accelerators, advanced packaging, and high‑bandwidth interconnects.