SK Telecom is planning to build gigawatt-scale artificial intelligence (AI) data centers in "key regions" throughout South Korea.
Announced at the SK AI Summit 2024, the company is planning to develop an "AI Infrastructure Superhighway."
In addition to the AI data centers, SK Telecom plans to launch a GPU as a Service (GPUaaS) solution and an Edge AI offering.
SK Telecom first plans to develop an AI data center "testbed" in Pango Korea, in December. The facility will trial direct liquid cooling, immersion cooling, and precision cooling, and will feature AI semiconductors including SK Hynix's HBM and GPU virtualization solutions and AI energy optimization technology to advise future AI data center developments.
Following that, the South Korean telco aims to build 100MW+ data centers in local regions, with plans to scale to a gigawatt or more in the future.
Plans for the AI cloud offering were first revealed in August of this year.
The GPUaaS offering will be housed at the Gasan data center, which SK Telecom plans to convert into an AI facility. Work on this will kick off in December, and SK Telecom will offer access to Nvidia H100 GPUs through a partnership with Lambda.
The GPUaaS will add Nvidia H200s in March 2025.
The Gasan data center launched 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.
Finally, the "Edge AI" offering aims to combine mobile communication networks and AI computing to reduce latency compared to large-scale AI data centers.
The company is currently researching the possibility of building AI data center-utilizing communication infrastructure and customized servers, and is carrying out proof of concept projects to see where this will be suited, including in healthcare, AI robots, and AI CCTV.
SK Telecom CEO, Ryu Young-sang noted that upcoming 6G technology will enable this "next-generation AI infrastructure where communication and AI are integrated.”
Earlier this year, SK Telecom invested $200 million in AI and high-performance computing (HPC) infrastructure business Smart Global Holdings (SGH). The companies plan to work together to “leverage their complementary capabilities to enhance customer offerings in the development of differentiated global end-to-end AI factory and data center solutions and services, advanced memory market products and services, and NPU-based AI Edge servers.”
