Qualcomm sees 6G commercialization being shaped by sensing, distributed intelligence and new multi-device experiences that merge the physical and digital worlds
For Qualcomm, the case for 6G is about building a network platform that can sense the environment, support distributed AI inference and model the real world in software. Together, those capabilities begin to define how 6G could support physical AI, a broad vision that connects robots, sensors, networks and edge compute into a coordinated system.
One of the clearest examples is ISAC, or integrated sensing and communications. Qualcomm Engineer Sam Kotla explained: “ISAC can be used for a variety of applications.” One of those is drone sensing, “where the base station is transmitting reference beams and then studying the receive signals, and from these signals you can actually extract information.” That can include identifying characteristics such as the size of an object or even a specific drone. As Kotla put it, “It’s really great for tracking aerial activity.”
Qualcomm demonstrated that idea by identifying the dimensions of two different drones using a base station with 100 megahertz transmit and receive channels. The setup uses monostatic sensing wherein a downlink subpanel transmits the reference signal while an uplink subpanel simultaneously captures the reflected signal. A background clutter cancellation step helps isolate the relevant information. The larger point is that 6G networks can both move data while also observing and interpreting the physical environment.
That same sensing capability feeds directly into another 6G use case: digital twins. Kotla told RCR Wireless News, “The sensing data that we want to collect from all these base stations can be used for a digital twin…It’s just a replica of the real world.” The value of that replica is practical. “What happens is you are creating a digital twin of this environment and whenever you want to deploy your resources in the network, you don’t actually have to deploy first.” Instead, operators can use the model to understand what they want to deploy and what resources they need. “This saves costs and improves efficiency for the network.”
Qualcomm is also framing 6G as an AI inference fabric that can support increasingly complex multi-modal and multi-device experiences. As Kotla said, “6G offers compute and inference as a service by enabling 6G devices to dynamically offload complex computational tasks to edge servers.” That approach gives devices access to more complex models while conserving local computational and power resources. Rather than forcing every AI workload onto the device, 6G creates a framework for distributing compute between endpoints and the network edge.
Taken together, ISAC, digital twins and the network as an inference fabric point toward a larger vision of physical AI. Kotla connected those dots through robotics. “Robots are something that are going to be used both in consumer applications and enterprises…They can be used for a variety of activities and they can be used indoors and outdoors…” She said these systems will rely on VLA models — vision, language, action models — running on compute servers that process data from robots and sensors, fuse it and generate actions. “But at the same time we also need some computation on the AI device mainly to run low-latency applications and mission-critical applications,” Kotla said. “You’re kind of distributing this compute between the devices and these compute models that are running on the edge servers.”
Taken together, 6G presents an opportunity to build a converged network platform that supports connectivity, sensing and inference, all of which will be required to deliver on the vision of physical AI.
