YOU ARE AT:AI InfrastructureSK Telecom to pilot Korean NPU for AI services

SK Telecom to pilot Korean NPU for AI services

SK Telecom said it will expand deployment of Rebellions’ NPU-powered servers depending on the outcomes of the ongoing testing phase

In sum – what to know:

SKT pilots domestic AI chip – SK Telecom is testing Rebellions’ ATOM NPU across key AI services, including call summarization and diagnostics, in a move toward AI hardware-software self-reliance.

ATOM-Max aims for scale – A next-gen chip, ATOM-Max is targeted for large-scale AI inference and will enter commercial use by year-end, strengthening Korea’s sovereign AI infrastructure.

Korea eyes AI independence – This collaboration marks a milestone in building a fully domestic AI tech stack—from LLM to semiconductor—positioning Korea to compete globally in AI innovation.

Korean operator SK Telecom said it is piloting compatriot firm Rebellions’ AI semiconductor technology, local newspaper The Korea Times reported.

The Asian carrier is currently testing Rebellions’ ATOM chip, a neural processing unit (NPU) built for AI data centers, across several core services powered by its in-house large language model (LLM), dubbed A.X.

The report noted that the trial involves evaluating the performance of Rebellions’ ATOM in applications such as AI-powered call summarization via SK Telecom’s digital assistant A., spam filtering, the PASS financial assistant and X Caliber, a veterinary X-ray diagnostic service.

Based on initial results, the two Korean companies plan to integrate ATOM-Max, Rebellions’ next-generation chip designed for large-scale AI inference, into commercial services later this year. SK Telecom also confirmed that it will expand deployment of Rebellions’ NPU-powered servers depending on the outcomes of the ongoing testing phase.

The report also highlighted that the partnership aims to localize not only AI hardware but also software infrastructure, creating a self-reliant AI technology stack entirely developed and operated within Korea. If successful, the deployment would mark the operation of an LLM on a fully domestic NPU—an achievement toward achieving sovereign AI, the report added.

“This is not just a technical integration, but a strategic move to reinforce Korea’s AI independence,” said Lee Sang-min, head of SK Telecom’s access network development. “Self-reliance in AI semiconductors, LLMs and data centers is vital for national competitiveness as global AI competition accelerates.”

Rebellions’ ATOM-Max is being developed as a high-efficiency chip suitable for scaling up AI inference workloads. It is expected to provide cost and performance benefits compared to general-purpose GPUs in certain tasks, the report added.

“The fact that a domestic AI chip is being applied to SK Telecom’s services operating on a domestic LLM is a milestone beyond technological maturity to signal real progress toward a self-sufficient AI ecosystem,” Rebellions CEO Park Sung-hyun said.

“Rebellions will continue to contribute to Korea’s leap toward becoming one of the world’s top three AI powerhouses by providing stable and energy-efficient AI infrastructure,” the executive added.

Looking ahead, both companies plan to deepen their partnership and expand adoption of homegrown AI solutions, including infrastructure and chips, the report added.

In 2023, SK Telecom had announced its ambition to become a global artificial intelligence company by strengthening its own AI competitiveness and cooperating with partners globally.

The telco had previously reorganized its operations into seven key divisions, with a strong focus on AI. This includes specialized units such as AI Transformation (AIX), AIDC, A. (A-Dot), and Global Personal AI Agent (GPAA), all dedicated to enabling differentiated services and AI leadership.

To reinforce its AI data center business, the Korean telco has made strategic investments in Lambda, a global GPU cloud provider, and signed an investment agreement with Penguin Solutions, a specialist in AI-driven data center solutions. In December 2024, SK Telecom launched the Gasan AI data center in partnership with Lambda and introduced ‘SKT GPUaaS’ to accelerate AI infrastructure expansion in both domestic and international markets.

ABOUT AUTHOR

Juan Pedro Tomás
Juan Pedro Tomás
Juan Pedro covers Global Carriers and Global Enterprise IoT. Prior to RCR, Juan Pedro worked for Business News Americas, covering telecoms and IT news in the Latin American markets. He also worked for Telecompaper as their Regional Editor for Latin America and Asia/Pacific. Juan Pedro has also contributed to Latin Trade magazine as the publication's correspondent in Argentina and with political risk consultancy firm Exclusive Analysis, writing reports and providing political and economic information from certain Latin American markets. He has a degree in International Relations and a master in Journalism and is married with two kids.