The general assembly of the global artificial intelligence (AI) network alliance led by Nvidia has been confirmed to take place in Seoul this November. Analysts say that Nvidia, which has chosen Korea as a major hub for physical AI, is laying the groundwork to strengthen cooperation with the domestic tech industry in building AI network infrastructure that can accelerate commercialization. Attention is focused on whether a physical AI system encompassing robotics, manufacturing AI, and AI networks can preempt the global standard.
According to the information technology (IT) industry Tuesday, the "AI RAN Alliance" recently set Seoul and November as the venue and timing of its 2026 annual general assembly. The alliance is a global AI network coalition launched in 2024, and this marks the first time its general assembly will be held in Korea. The coalition, led by Nvidia, also includes domestic and international mobile carriers.
AI RAN, Core Infrastructure for the Physical AI Era
The general assembly to be held in November is expected to become an occasion for advancing cooperation between Nvidia, which is centered on robotics, and Korean industry. It is observed that the domestic mobile telecommunications industry will announce a concrete investment roadmap for AI RAN to build AI networks. The investment scale is expected to take shape in line with physical AI-related projects that Nvidia will pursue in Korea. Nvidia CEO Jensen Huang, who arrives in Korea Wednesday, plans to discuss physical AI cooperation measures such as robotics and manufacturing automation through meetings with chiefs of major companies.
AI RAN is considered core infrastructure for the physical AI era. This intelligent base station goes beyond the role of existing communications infrastructure and operates like a small AI data center. A single base station can simultaneously perform radio access network (RAN) processing and AI computation. AI effectively arranges wireless radio frequency signals, enabling maximum capacity and speed even with limited spectrum. It supports a customized network so that physical AI technology can be implemented through functions such as ultra-low-latency performance, high security, and distributed computing. To implement this technology, Nvidia has chosen a method of mounting its own graphics processing units (GPUs).
In particular, robots that perform tasks such as food delivery or automobile assembly require rapid decision-making, so there is a greater need to process data on a nearby AI network rather than at an AI data center. As networks advance, the battery consumption of robots and the cost of data processing devices can also be greatly reduced. Generative AI services can be accommodated even by existing communications networks, but AI RAN demand is expected to increase significantly as physical AI develops, requiring more sophisticated computation and inference.
Samsung Electronics, Three Telecom Carriers Partner With Nvidia
Cooperation between Nvidia and the Korean IT industry began in earnest with the launch of the Korean AI network cooperation body "AINA" (AI Network Alliance) in March this year. AINA was formed with the goal of building AI networks and fostering the industrial ecosystem, consisting of Korea's three mobile carriers, 13 chair companies from industry, academia, research, and government, and about 20 companies related to the AI network value chain. Nvidia signed a memorandum of understanding (MOU) with AINA and has been seeking cooperation measures for building domestic AI networks. In particular, Samsung Electronics (005930.KS), which manufactures telecommunications equipment, is jointly developing AI RAN technology with Nvidia. An IT industry official said, "I understand that Nvidia is also collaborating jointly with domestic IT companies to develop the software needed to implement AI RAN," adding, "Multifaceted cooperation is being pursued to transform existing networks into AI-based ones."
This cooperative move comes because the interests of Nvidia, which has preempted AI semiconductor and software technology led by GPUs, align with those of Korean industry, which is strong in manufacturing. In its "AI Network Strategy Report" published in September last year, the Electronics and Telecommunications Research Institute (ETRI) explained, "AI RAN equipment does not stop at simple hardware expansion but is developing into advanced network equipment capable of autonomous and intelligent operation by combining with network AI models that operate intelligently throughout the entire process of network design, control, operation, and evolution." This means that high-level technical capabilities must be combined on both the hardware and software sides to deploy an AI network.
The government has also unveiled a strategy to advance AI-based communications networks. In December last year, the Ministry of Science and ICT (MSIT) announced the "Hyper AI Network Strategy" to advance the country's network infrastructure in the AI era and strengthen industrial competitiveness. The strategy includes a plan to install more than 500 AI RAN units at hubs nationwide until the commercialization of sixth-generation (6G) in 2030. Specifically, the policy is to build 200 at carriers' concentrated central offices, 100 at industrial complexes, 100 at airports, ports, and railways, and 100 at education and medical facilities. The policy is to reduce base station power consumption by more than 30% by utilizing domestically produced AI network models in AI RAN. Through this, the plan is to preempt 20% of the global AI network market and achieve a 30% share of the 6G standard.
