Robots navigating factory floors, drones inspecting power lines, and autonomous vehicles reading traffic all share the same constraint: They need to act in milliseconds. Sending data to a distant computing facility and waiting for a response takes too long. For machines operating in the physical world, that delay can mean a missed obstacle, a wrong turn or a failed inspection. The Robot Report noted that physical artificial intelligence systems can’t depend on remote computing for real-time decisioning. That problem has no software fix. It requires different infrastructure, built closer to where machines operate. That gap is what Nvidia and T-Mobile are moving to fill. Working with Nokia, the two companies are running computing directly through cell tower infrastructure. Instead of routing data to a remote facility, the system processes it at the tower, near where the robot or vehicle operates. T-Mobile was the first U.S. carrier to run the system in a live pilot. T-Mobile’s Tower AI Is Already on the Job Physical AI deployments have been moving out of controlled environments and into live industrial settings. Robots update their behavior on the fly. Drones reroute around unexpected obstacles. None of that can wait. Sending data back and forth to a remote facility introduces delays that make the system unreliable. The issue isn’t bandwidth. It’s distance. Computing needs to happen closer to the machine. Carriers are positioned to provide that. They already own tower infrastructure nationwide. T-Mobile CEO Srini Gopalan said the company built its network for exactly this requirement. Nvidia CEO Jensen Huang said at Nvidia’s GTC 2026 conference that telecoms, a roughly $2 trillion industry, are about to be reinvented around that opportunity. “The radio tower used to be just a radio tower,” Huang said. “In the future, it’s going to be an AI infrastructure platform.” T-Mobile’s AI