Siemens Executive Highlights Semiconductors as the Core of Humanoid Robot Intelligence

Siemens Semiconductor Vice President Michael Mancy emphasized that the true essence of humanoid robots lies not in their physical appearance but in the semiconductors that enable their functionality. These robots must process real-time sensor data, run AI models, and coordinate joint movements simultaneously, requiring a robust semiconductor computing platform to avoid being mere 'puppets'.

Mancy pointed out that humanoid robots are fundamentally software-defined systems, with their capabilities stemming from the integration of software, electronics, and mechanics rather than just mechanical hardware. The intelligence of these robots must extend to safely responding to various situations, necessitating advanced physical AI that coordinates perception, feedback, and actuation, which is significantly more complex than executing a large language model.

As AI workloads increase on processors, heat management becomes critical, especially in high-temperature and electrically noisy factory environments. Mancy proposed heterogeneous integrated circuits as a solution, allowing multiple chips to be stacked in one package to optimize thermal performance. He also highlighted the importance of digital twins for modeling and validating humanoid systems before physical construction, enabling engineers to test software and hardware collaboratively and share verified improvements across fleets, ensuring continuous evolution and reliable operation in real-world environments.