Physical AI

Robotics silicon: what runs the control loop, and what runs inference?

NXP’s Neural Axis announcement describes a system architecture that can split sensing, control, inference, and communications across devices. Renesas RZ/T2N is a shipping MPU. Its product page puts application processing, motor control, and industrial Ethernet on one chip, and it lists no NPU.

Physical AI

Physical AI chiplets: what emerging architecture work means for selection

Cadence and imec describe a proposed Open Compute Project profile for physical AI chiplets. The post presents it as work still open for discussion, with die-to-die physical and link layers left to UCIe and BOW.

Physical AI

Physical AI: security and connectivity questions before selecting IP