Samsung: Physical AI Applications Will Demand Smarter, More Autonomous Networks
Samsung's networks division contends that physical AI applications will require networks to evolve beyond simple connectivity to provide intelligent, distributed computing infrastructure that determines whether workloads should execute on devices, edge nodes, or centralized cloud systems. Industrial trials launching this month in South Korea with carriers KT and SK Telecom will test physical AI use cases such as autonomous robots in shipyards and petrochemical facilities, generating real-world data to inform Samsung's 6G development. The company identifies key network requirements for physical AI as ultra-low latency, high reliability, consistent throughput, expanded uplink capacity, and autonomous operations.
Samsung's networks division is testing autonomous robotic systems in real-world industrial settings to understand how AI applications that physically interact with their environment differ from traditional software-based AI. These trials involve robots performing tasks like welding, painting, and hazard detection while processing high-definition video feeds in real time. The company is investigating how networks should intelligently route computational tasks—deciding whether processing should happen locally on devices, at nearby edge infrastructure, or in distant data centers—based on each application's specific latency and reliability needs.
The initiative stems from recognition that physical AI demands more sophisticated network capabilities than current 5G infrastructure provides. Samsung's Network in a Server platform combines wireless connectivity, computing power, and AI tools on single hardware units, allowing factories and industrial sites to operate robots with minimal delay. These Korean trials represent an early step toward understanding how next-generation 6G networks might need to function as intelligent computing platforms rather than mere data pipes.
The outcomes of Samsung's industrial trials could influence how telecommunications networks are designed and operated globally. If Samsung's approach proves viable, manufacturers and industrial operators worldwide might gain access to more reliable autonomous systems with reduced downtime and improved safety capabilities. However, the success of this model depends on networks evolving faster than current timelines suggest, and widespread adoption would require significant investment from carriers and equipment vendors in new infrastructure. The results may also shape regulatory conversations around network infrastructure's role in critical industrial operations.