With the recent increase in AI-related workload and traffic within data centers, the need for international standards for AI data center technologies has emerged, and the AI data center technology proposed by SKT has been approved as a new standardization project.

SKT announced that the item 'Interconnection Structure and Method of AI DC Technology,' which it proposed to the International Telecommunication Union Telecommunication Standardization Sector (ITU-T) under the UN to establish global standards for AI data center (AI DC) related technologies, was approved as a new standardization project at the ITU-T international conference in Geneva, Switzerland on the 10th (local time).

ITU-T is an organization within the International Telecommunication Union (ITU) that conducts research and standardization on standards related to electricity and telecommunications, with over 900 organizations, companies, and research institutes from more than 190 member countries participating. This ITU-T international conference was held in Geneva from the 1st to the 10th.

The idea proposed by SKT was approved as a new standard project after undergoing review and eligibility assessment by members of Study Group 11 (SG), which is conducting standardization on the structure and requirements of switching and signaling methods at the ITU-T.

SKT stated that this project aims for the organic interoperability and combination of key technological elements constituting the AI DC, and includes the structure, signaling methods, and usage methods of various technological elements within the data center.

In addition, it was announced that the interoperability structure between key technology elements constituting the AI DC has been defined by classifying them into three modules—AI Infrastructure, Management, and Resource Allocation—according to their functions and roles, and that a blueprint for the interoperability structure and data communication between each module has been included in the standardization plan.

The AI infrastructure module covers technical elements such as AI processors, memory, storage, next-generation cooling technology, energy efficiency solutions, and security, while the management module contains technical elements related to the management of the AI DC infrastructure. Resource allocation is a module for technical elements responsible for resource virtualization, resource allocation, and authentication within the AI DC.

SKT stated, “The adoption of this AI DC technology standardization project is due to SKT’s cooperation with various affiliates within the SK Group and its years of accumulating capabilities in the AI and ICT fields and developing core technologies,” adding, “SKT expects this standardization work to serve as a catalyst to promote the establishment of AI DCs by companies and institutions on a global scale.”

He also stated, “Through standardization, we will provide AI DC-based services and functions capable of meeting the demands of various sectors, including corporate customers, individual users, and the public sector, and it will also help optimize interconnectivity to ensure compatibility between data centers.”

Moving forward, SKT will develop detailed standards with member companies regarding the interoperability structure between each module of the AI DC and the types of data required for interoperability. Once the developed proposal passes the discussion and final adoption process at ITU-T meetings, SKT's standardization proposal will officially become a global standard.

Lee Jong-min, Head of Future R&D at SKT, stated, “The approval of this new standardization project signifies that the International Telecommunication Union (ITU-T), an international standardization organization, not only recognizes the importance of AI DC-related technologies but also acknowledges SKT’s accumulated AI R&D capabilities in the field.” He added, “SKT will complete the AI DC standard specifications through SK Group’s capabilities and global cooperation.”