SK Telecom unveils plans for AI Infrastructure at SK AI Summit 2024
Introduction:
During the two-day SK AI Summit 2024 [1.], SK Telecom CEO Ryu Young-sang unveiled the company’s comprehensive strategy which revolves around three core components: AI data centers (AIDCs), a cloud-based GPU service (GPU-as-a-Service, GPUaaS), and Edge AI. SK Telecom is planning to construct hyperscale data centers in key regions across South Korea, with the goal of becoming the AIDC hub in the Asia Pacific region. Additionally, the company will launch a cloud-based GPU service to address the domestic GPU shortage and introducing ‘Edge AI’ to bridge the gap between AIDC and on-device AI. This innovative approach aims to connect national AI infrastructure and expand globally, in collaboration with partners both in South Korea and abroad.
Note 1. The SK AI Summit is an annual event held by the SK Group, where global experts in various AI fields gather to discuss coexistence in the era of artificial general intelligence (AGI) and seek ways to strengthen the ecosystem.
………………………………………………………………………………………………………………………………………………………………………..
Constructing AI Data Centers in South Korea’s key regions:
SK Telecom plans to start with hyperscale AIDCs that require more than 100 megawatts (MW) in local regions, with future plans to expand its scale to gigawatts (GW) or more, to leap forward as the AIDC hub in the Asia Pacific region.
By extending the AIDC to national bases, centers can secure a stable power supply through the utilization of new renewable energy sources such as hydrogen, solar and wind power, and easily expand to global markets through submarine cables. SK Telecom anticipates building AIDC cost-effectively when the data center combines SK Group’s capabilities in high-efficiency next-generation semiconductors, immersion cooling, and other energy solutions, along with its AI cluster operation.
Prior to this, SK Telecom plans to open an AIDC testbed in Pangyo, Korea, in December, which combines the capabilities of the SK Group and various solutions owned by partner companies. This facility, where all three types of next-generation liquid cooling solutions—direct liquid cooling, immersion cooling, and precision liquid cooling—are deployed, will be the first and only testbed in Korea. It will also feature advanced AI semiconductors like SK hynix’s HBM, as well as GPU virtualization solutions and AI energy optimization technology. This testbed will provide opportunities to observe and experience the cutting-edge technologies of a future AIDC.
Supplying GPU via cloud to metropolitan areas:
SK Telecom plans to launch a cloud-based GPU-as-a-Service (GPUaaS) by converting the Gasan data center, located in the metropolitan area, into an AIDC to quickly resolve the domestic GPU shortage.
Starting in December, SK Telecom plans to launch a GPUaaS with NVIDIA H100 Tensor Core GPU through a partnership with U.S.-based Lambda. In March 2025, SK Telecom plans to introduce NVIDIA H200 Tensor Core GPU in Korea, gradually expanding to meet customer demand.
Through the AI cloud services (GPUaaS), SKT aims to enable companies to develop AI services easily and at a lower cost, without needing to purchase their own GPUs, ultimately supporting the vitalization of Korea’s AI ecosystem.
Introducing ‘Edge AI’ to open a new opportunity in telco infrastructure:
SK Telecom plans to introduce ‘Edge AI,’ which can narrow the gap between AIDC and on-device AI, using the nationwide communication infrastructure.
Edge AI is an infrastructure that combines mobile communication networks and AI computing, offering advantages in reduced latency, enhanced security, and improved privacy compared to large-scale AIDCs. Additionally, it enables large-scale AI computing, complementing the existing AI infrastructure, compared to on-device AI.
SKT is currently conducting research on advanced technologies and collaborating with global partners to build AIDC-utilizing communication infrastructure and develop customized servers. The company is also carrying out various proof of concept (PoC) projects across six areas, including healthcare, AI robots, and AI CCTV, to discover specialized Edge AI services.
“So far, the competition in telecommunications infrastructure has been all about connectivity, namely speed and capacity, but now the paradigm of network evolution should be changed,” said Ryu Young-sang, CEO of SK Telecom. “The upcoming 6G will evolve into a next-generation AI infrastructure where communication and AI are integrated.”
Developing a comprehensive AIDC solution to enter global market:
SK Telecom plans to develop a comprehensive AIDC solution that combines AI semiconductors, data centers, and energy solutions through collaboration with AI companies in Korea and abroad, with the aim of entering the global market.SK Telecom aims to lead the global standardization of Edge AI and collaborate on advanced technology research, while working towards the transition to 6G AI infrastructure.
………………………………………………………………………………………………………………….
About SK Telecom:
SK Telecom has been leading the growth of the mobile industry since 1984. Now, it is taking customer experience to new heights by extending beyond connectivity. By placing AI at the core of its business, SK Telecom is rapidly transforming into an AI company with a strong global presence. It is focusing on driving innovations in areas of AI Infrastructure, AI Transformation (AIX) and AI Service to deliver greater value for industry, society, and life.
For more information, please contact [email protected] or visit our LinkedIn page www.linkedin.com/company/sk-telecom
………………………………………………………………………………………………………………….
References:
SKT-Samsung Electronics to Optimize 5G Base Station Performance using AI
SK Telecom (SKT) and Nokia to work on AI assisted “fiber sensing”
Plans for the AI cloud offering were first revealed in August. SK Telecom will deploy Lambda’s AI Cloud platform in SK Broadband’s Gasan data center later this year, with plans to eventually scale up to support thousands of GPUs. The Gasan data center opened in 2021, with 10 stories above ground and another five basement floors. At the time, it could support 46MW of IT Load across 69,300 square meters (745,940 sq ft). SK Telecom said that the site can currently support up to 44 kilowatts per rack.
“SKT shares in our vision to make GPU compute as ubiquitous as electricity,” said Lambda CEO and co-founder, Stephen Balaban.
“Given the rapid pace of AI innovation happening in South Korea, we’re excited to partner with SKT in their mission to transform their company and country into a global AI powerhouse.”
-SKT announced revenue of KRW 4.5321 trillion, operating income of KRW 533.3 billion, and net income of KRW 280.2 billion on a consolidated basis
-Company is focused on increasing sales in AI data centers, AI B2B and AI B2C businesses that are expected to quickly generate profits
-AI business thrives as ‘A.’ (A-DoT) subscriber count surpasses 5.5 million and continues to receive orders for enterprise AI
………………………………………………………………….
On the second day of the SK AI Summit 2024, SK Telecom unveiled Aster (A*), the company’s new global AI personal agent. Aster, meaning a ‘Guiding Companion’, is an agentic AI that understands user intentions, sets goals, makes plans, and completes tasks on user’s behalf. SK Telecom is committed to delivering innovative experiences through Aster, guided by the three core concepts of ‘Planning Made Easy’, ‘Naver Drop the Ball’, and ‘Timely Answer’
Aster will launch a closed beta in North America later this year, with a full market launch planned for the following year.
SK Telecom is actively seeking partnerships with global search providers, large language model developers, and third-party applications to bring unparalleled value and convenience to the customers.
Nov 17 2024 Update:
SK Telecom to Introduce AI-Powered Customer Service Utilizing Proprietary Telco LLM and LMM
SK Telecom has successfully completed the beta service of its AI Customer Service Support System. This revolutionary system, powered by the company’s proprietary telecommunications-specialized large language model (Telco LLM) and large multimodal model (LMM), is set to transform the customer service landscape.
Developed in collaboration with global LLM leaders, the system introduces three transformative features: AI Knowledge Search Assistant, Intelligent Document Processing, and the Automated Post-Processing System for Consultation Results.
Agents have reported exceptional performance from the Telco LLM during the month-long beta test, noting its ability to streamline information access and reduce workload.
A key innovation in the system is the implementation of Retrieval Augmented Generation (RAG), which significantly enhances response accuracy by retrieving relevant information and leveraging this information in the response generation. The company’s Telco LMM takes this capability further by processing both text and telecommunications-related images, creating a truly comprehensive service solution.
The AI Customer Service Support System introduces three transformative features: the AI Knowledge Search Assistant, Intelligent Document Processing, and the Automated Post-Processing System for Consultation Results.
The AI Knowledge Search Assistant, launched October 21, is a natural language interface that enables agents to instantly access relevant information without manual searching, similar to advanced AI platforms like Perplexity or ChatGPT. The service is being refined through real-world usage and will be fully deployed across all customer service operations in 2025.
Intelligent Document Processing, implemented on October 23, is an LMM-powered system that automatically categorizes and processes various customer-submitted documents received through multiple channels. Future updates will introduce automated content analysis capabilities, significantly reducing document verification time.
The Automated Post-Processing System for Consultation Results, scheduled for December release, will automate many tasks, including categorizing and summarizing consultation outcomes. This automation will reduce the minutes-long post-call documentation process significantly, allowing agents to serve more customers efficiently while maintaining comprehensive consultation records. This enhancement allows agents to serve more customers efficiently while maintaining comprehensive consultation records.
Initial feedback from the month-long beta test has been positive. Agents report exceptional performance from the Telco LLM, particularly noting its ability to streamline information access and reduce workload for newer team members. The automated document classification system has notably improved operational efficiency, allowing agents to focus more on customer interactions.
“Our transformation into an AI contact center represents more than just operational efficiency — it’s about delivering truly personalized, customer-centric service,” said Hong Seung Tae, Vice President and Head of Customer Value Innovation Office at SK Telecom. “Through our Telco LLM-based AI Customer Service Support System, we expect to set new standards in customer service excellence and position ourselves as the industry leader in AI-driven customer support.”