KT will use about 3,500 telecom facilities it owns across South Korea as artificial intelligence computing hubs, building a distributed AI infrastructure for the era of physical AI. The approach distributes AI computation by linking a central AI data center (AIDC) that handles large-scale training and inference with telecom facilities located near industrial sites. KT also plans to upgrade network technologies including enterprise-dedicated 5G, AI-RAN and post-quantum cryptography (PQC), expanding its existing telecom infrastructure into intelligent infrastructure spanning AI computing, networking and security.
KT said on the 17th that it will build "AI Edge" by using about 3,500 telecom facilities nationwide as AI computing hubs. By linking the central AIDC with telecom facilities near industrial sites, the company aims to create an environment where AI computation can be processed close to where data is generated.
In existing AI services, data generated at a site is sent to a central data center, where AI processes it before the results are sent back to the site. But physical AI requires fast and stable data processing, making it necessary to place AI computing resources near industrial sites. KT plans to use its telecom facilities to organically connect the central AIDC with AI Edge nodes nationwide, providing an ultra-low-latency, high-reliability environment for physical AI services such as robots and autonomous driving that require immediate decisions.
Along with using telecom facilities nationwide as AI Edge nodes, KT will develop modular AI Edge units to speed up deployment at manufacturing sites. The company said the same day that it will take part in a project run by the Ministry of Science and ICT to build a research and development ecosystem for a collaborative-intelligence physical AI software platform. The government is currently pursuing a project in North Jeolla Province to implement a dark factory that automates and unmans its entire process using physical AI. In the project, KT will demonstrate "modular edge AIDC" infrastructure that performs AI computation close to manufacturing sites.
KT will also upgrade the 5G networks used by companies and public institutions. The company is working with Samsung Electronics and Google on technology development and verification, aiming to unveil "App Slicing" for enterprises within this year. The technology allows work applications on a single smartphone to connect to a company's internal network through enterprise-dedicated 5G, while personal apps connect through the general internet. To lower the barrier for companies and public institutions entering 5G work networks, KT has also launched a "5G work network egg rental" service, which lets customers lease egg devices — terminals dedicated to work networks — instead of buying them. Clients can rent only the quantity they need, with KT providing integrated support from activation to device operation and troubleshooting. KT is also expanding its "5G work network hub" service, in which a regional institution and its affiliated organizations share a single work network infrastructure.
KT plans to combine this with post-quantum cryptography and AI-driven autonomous security technology. AI will detect abnormal behavior in networks and devices to isolate compromised terminals, while preparing for the possibility that quantum computers could threaten existing encryption systems. The strategy is to expand its business-to-business (B2B) operations beyond the existing telecom business of connecting data via 5G into physical AI infrastructure that also provides AI computation and security. KT said it will "lead security innovation in the AI transformation (AX) and quantum era based on proven results."
