Physical AI can impress us with an outstanding performance that rivals or surpasses human capabilities in a few tasks. Over time, AI-driven robots, drones and other devices will expand the range of tasks they can perform with accuracy, reliability and cost- and energy-efficiency. But to achieve this, they need levels of latency, reliability and security that today’s public networks do not support. For many use cases, physical AI will need a private network infrastructure that may or may not be managed by mobile operators. In most countries, mobile operators will not be the main infrastructure players.
I sat down with Gianluca Verin, CEO and Co-Founder of Apeiroon, to examine how we should plan and operate mobile networks to successfully and safely operate physical AI in the real world at scale. Gianluca has accumulated in-depth, practical experience on private networks for a long time since he started Athonet, and explains why many physical AI use cases require dedicated, deterministic, edge connectivity, and why centralized cloud architectures are not sufficient:
- How to think about network requirements as physical AI gets deployed
- Edge AI, mobile AI, mobile AI bubbles as the foundation for scalable physical AI deployments
- Physical AI innovation in Europe versus USA: An opportunity that Europe cannot afford to lose
To sum it all up, Gianluca said: “A stationary robot is an IT problem. A moving robot is a network problem.”
To go deeper, check the white paper that Gianluca mentions. And if you want to find out where Apeiroon comes from, see Carlo Rovelli, Anaximander and the nature of science, it is a beautiful and beautifully written short book.
You can watch the video of this podcast and read the transcript on Sparring Partners on Substack or listen to it on your podcast platform.
