Jungchul Kim, head of product strategy group, networks business at Samsung Electronics, told RCR that physical AI applications will place specific demands on networks, including ultra-low latency, high reliability, consistent throughput, greater uplink capacity, and autonomous network operations
In sum – what to know
Beyond connectivity – Samsung says physical AI will require networks to provide intelligent, distributed infrastructure rather than simply connecting AI applications.
Workloads become dynamic – Future networks may need to determine whether AI workloads should run on devices, at the edge or in the cloud.
Trials inform 6G – Samsung’s industrial AI RAN projects with KT and SK Telecom will help the company understand physical AI requirements and inform AI-native 6G R&D.
Korean vendor Samsung believes the emergence of physical AI will place new demands on networks, requiring them to move beyond connectivity and play a more active role in coordinating distributed computing.
That is one of the key lessons Samsung expects to draw from industrial AI RAN projects with Korean carriers KT and SK Telecom, according to Jungchul Kim, head of product strategy group, networks business at Samsung Electronics.
The projects, scheduled to begin this month under South Korea’s Ministry of Science and ICT “Hyper AI Network” initiative, will deploy 5G standalone private networks in industrial environments including shipyards and petrochemical facilities.
For Samsung, however, the significance of the trials extends beyond validating individual AI RAN use cases. Kim told RCR Wireless News they will help identify the network requirements of physical AI in real-world environments and inform Samsung’s R&D work on AI-native 6G.
Kim said physical AI applications will place specific demands on networks, including ultra-low latency, high reliability, consistent throughput, greater uplink capacity and autonomous network operations.
The requirements reflect the nature of physical AI applications being tested in the industrial projects. At the Yeongam Shipyard of HD Hyundai Samho, KT will test physical AI services including welding and painting robots and autonomous robots for telecom facility operations.
At SK Incheon Petrochem, SK Telecom will validate physical AI services including an autonomous patrol robot and CCTV-based monitoring. The patrol robot transfers high-definition video in real time while moving through the facility, enabling AI analysis to identify potential hazards and support integrated monitoring.
Samsung is providing its Network in a Server (NIS) architecture for the projects. Kim described NIS as an integrated platform combining virtualized RAN, AI Core, and AI applications on a commercial off-the-shelf server.
Kim said the combination of connectivity and computing will become increasingly important as physical AI applications generate continuous video and sensor data.
He said future networks will need to determine dynamically where workloads should run — on devices, at the edge or in the cloud — while optimizing radio and compute resources.
That points to a broader role for the network as AI applications increasingly interact with the physical environment. Rather than treating connectivity and computing as separate functions, the architecture will need to coordinate them according to the requirements of individual applications.
“The network will need to do more than connect AI applications; it will need to provide the intelligent, distributed infrastructure that enables them to operate reliably in the physical world,” Kim said.
The industrial trials will provide Samsung with an environment in which to examine those requirements. Kim said the projects are intended to help the company understand how physical AI changes network needs and how those requirements should feed into future network development.
Kim said broader deployment will ultimately depend on each operator’s priorities, requirements and evaluation of the trial results.
The projects therefore give Samsung an opportunity to examine physical AI not simply as another application running over a network, but as a use case that can influence how connectivity, computing and AI capabilities are distributed across the network. For Samsung, those requirements are also part of the longer-term development path toward AI-native 6G.
