SKT가 미국 GPUaaS 기업 ‘람다(Lambda)’와 손잡고 오는 12월 서울에 AI 데이터센터를 연다. 람다가 보유한 엔비디아 GPU 자원을 SK브로드밴드의 서울 가산 데이터센터에 전진 배치하는 것이다. SKT는 AI 데이터센터에 배치할 GPU를 3년 안으로 수천 대 이상까지 확대할 계획이다.
Lambda and AI Cloud Partnership to Expand to Thousands of Units in 3 Years
“Securing AI Core GPU Stability, Expanding Korea’s AI Competitiveness, and Global Expansion”
SKT opens an AI data center based on NVIDIA GPUs in Seoul and seeks to expand Korea's AI competitiveness and enter the global market.
SKT announced on the 21st that it will open an AI data center in Seoul in December in cooperation with American GPUaaS company 'Lambda'.
Lambda will deploy NVIDIA GPU resources to SK Broadband's Seoul Gasan Data Center. SKT plans to expand the number of GPUs deployed in its AI data center to thousands within three years.
Lambda, founded in 2012 by AI engineers, is a GPUaaS company that provides AI cloud services by receiving the latest GPUs from NVIDIA. Global big tech companies such as Intel and Microsoft are Lambda's customers. According to foreign news agencies such as Reuters, Lambda is attracting attention as an emerging AI company, having been valued at $1.5 billion in February and attracting $320 million in investment.
SKT and Lambda will deploy the NVIDIA GPU 'H100' in the existing SK Broadband data center located in Gasan-dong, Geumcheon-gu, Seoul in December.
Considering the rapid increase in domestic GPU demand due to the growth of the AI market, SKT is planning to increase the number of GPUs to thousands of units within three years and is also pushing for early introduction of the latest GPU model, 'H200'. Through this, the goal is to expand the largest 'GPU Farm' in Korea consisting of a single NVIDIA GPU, starting with the Gasan Data Center.
SK Broadband will provide a data colocation environment optimized for high-density GPU server operation environments by leveraging its data center operation know-how. To ensure stable operation of GPU servers, the Gasan Data Center plans to implement a power density per rack of 44kW, the highest in Korea. This is approximately 9 times higher than the average power density per rack of domestic data centers, which is 4.8kW.
With the opening of the AI data center in December, Lambda’s Korea region will also be the first in the Asia Pacific region. Data from domestic companies using Lambda GPU-based AI cloud services will be stored in the Korea region.
SKT also plans to launch 'GPUaaS (GPU-as-a-Service),' a subscription-based AI cloud service based on Lambda GPU resources, in December. 'GPUaaS' is a service that allows corporate customers to rent resources in a virtual environment through the cloud without directly purchasing the GPUs needed for AI service development or utilization. Large corporations, small and medium-sized businesses, and startups that find it burdensome to directly purchase GPUs that are in short supply and expensive can use them at a relatively low cost.
Accordingly, IT companies around the world are launching 'GPUaaS', and the market size is also growing rapidly. According to market research firm 'Fortune Business Insights', the global 'GPUaaS' market is expected to grow from $4.31 billion in 2024 to $49.84 billion in 2032, with an average annual growth rate of 35.8%.
SKT plans to introduce various promotions for domestic startups and SMEs, such as GPU replacement compensation programs, cloud cost optimization consulting, and AI startup support programs, along with the launch of 'GPUaaS' in December. More details can be found on the SK Telecom Enterprise homepage.
Kim Kyung-deok, head of SKT’s Enterprise Business Division, said, “Securing GPUs stably through strategic cooperation with Lambda is significant in terms of expanding the domestic GPU supply.” He added, “We will strive to build the largest GPU farm in the country in the future to enhance the nation’s AI competitiveness and establish it as a bridgehead for advancing into the global market.”