With the AI data center interconnection structure proposed by SK Telecom being adopted as an international standard, a common standard has been established for the operation of increasingly complex AI infrastructure. The key to this approval is the structuring of the connection method between internal systems in an AI data center environment where large-scale computing equipment, as well as power, cooling, storage, security, and resource management systems, operate simultaneously.

SK Telecom announced on the 18th that its proposed "Technical Elements and Interconnection Structure for AI Data Centers" was finally approved as an international standard at the SG11 meeting of the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) held earlier this month. The company explained that it achieved this result after approximately two years of research and international consultations following the approval of the proposal as a new standardization project in May 2024.

The core of this standard is to divide AI data centers into three layers—service layer, management layer, and infrastructure layer—and to systematically organize the roles of each layer and the signaling requirements necessary for interoperability. The structure is such that the service layer is responsible for executing and providing AI services, the management layer for operations and control, and the infrastructure layer for actual facility functions such as power, cooling, servers, and storage. It essentially establishes a standard for how systems with different functions exchange state and control information.

The need for such standards stems from the structural complexity of AI data centers. Unlike general data centers, AI data centers must simultaneously consider large-scale computation processing, power supply, heat control, storage resource allocation, security maintenance, and operational automation. Consequently, there has been a persistent demand from the industry to standardize signal exchange and interoperability methods between internal systems.

The scope of application extends beyond telecommunications companies and cloud providers to the entire related industry supplying servers, networks, power, cooling, and security equipment. If the standard becomes widespread, companies and organizations are expected to be able to connect different equipment and management systems within a more consistent structure when building or operating AI data centers.

As the competition for AI infrastructure expands beyond securing semiconductors to operational efficiency and interoperability systems, the industry is paying close attention to the possibility that this standard will become one of the benchmarks for future discussions on the construction and operation of global AI data centers. SK Telecom also stated that this approval establishes a global standard for AI data center operational structures and expressed its intention to continue participating in international standardization discussions in the AI field.