Skip to content
  • KOSPI 2824.94 +30.93 +1.11%
  • KOSDAQ 840.81 +4.71 +0.56%
  • KOSPI200 388.53 +5.30 +1.38%
  • USD/KRW 1390.5 +2.5 +0.18%
  • JPY100/KRW 871.81 +2.05 +0.24%
  • EUR/KRW 1491.66 +1.71 +0.11%
  • CNH/KRW 190.87 +0.28 +0.15%
View Market Snapshot
Artificial intelligence

Korea's LG Uplus aims for $1.4 bn in B2B AI sales by 2028

It plans to unveil its first AI chip later this year, stepping up efforts to increase AI services for corporations

By Jul 02, 2024 (Gmt+09:00)

3 Min read

LG Uplus corporate division head Kwon Yong-hyun explains about B2B AI services on July 2 (Courtesy of LG)

South Korea’s leading mobile carrier LG Uplus Corp., which is set to unveil its first artificial intelligence chip by the end of this year, aims for 2 trillion won ($1.4 billion) in revenue from the business-to-business (B2B) AI sector by 2028.

The telecommunications company, which is shifting its focus to the non-telecom sector such as AI for mid- to long-term growth, said on Tuesday that it will expand its AI infrastructure, platform and corporate client services.

“The AI market is expanding far faster than expected,” said LG Uplus corporate division head Kwon Yong-hyun at a press conference in Seoul on July 2. “We will use our entire capability for AI and apply AI technologies to all services to help the growth of corporate clients,” he added.

The company’s B2B AI strategy will focus on three businesses – AI data centers and on-device AI, AI platforms using smaller large language models (sLLMs) and AI-powered services for corporate clients such as business operations and communications.

LG Uplus' press conference on B2B AI services (Courtesy of LG)

AI INFRASTRUCTURE

LG Uplus, which operates two hyperscale data centers in Korea, is planning to build an additional one by 2027.

The company aims to maximize the operational efficiency of AI servers through its generative AI model ixi-GEN, which officially launched last month, and improve the reliability of servers by using cooling technology.

The mobile carrier will unveil an AI chip later this year in a partnership with Korea’s neural network and AI chip developer DeepX. DeepX specializes in neural processing units (NPUs), key components for on-device AI as they boast higher efficiency and consume less power than graphic processing units (GPUs) in data processing.

LG Uplus and DeepX plan to use the AI chip for operations of various sectors including telecommunications equipment, customer care, robotics and mobility.

LG Uplus' headquarters in Seoul (Courtesy of LG)


SECTOR-SPECIFIC AI PLATFORM

The company is set to provide ixi-GEN to finance, education and security industries this year as it is expected to learn sector-specific knowledge to improve the efficiency of business management.

The sLLM, with fewer parameters than a large language model (LLM), can build AI service relatively quickly and with less cost. LG Uplus has trained the model with its communication and platform data on hyperscale multimodal AI Exaone, developed by tech lab LG AI Research.

The parameters of the sLLM can be 8.8 billion, 25 billion or other numbers depending on the client’s needs, said LG Uplus.

LG Uplus has also launched three platforms for AI transformation: ixi Solution, which enables creating and managing AI services; Uplus Data Lake for optimizing data quality; and Violet for automation of AI development and learning processes.

FOUR KEY SERVICES

The company will also use AI for its four major services for corporations – customer care, communications, small and medium-sized enterprise (SME) operations and mobility.

The customer care service will provide information relevant to the industry and summarize and categorize chats.

For SMEs, LG Uplus will use on-device AI for hardware such as self-service kiosks, closed-circuit televisions (CCTVs) and point of sales (POS) as well as AI-powered phone calls and booking services.

The company will also introduce AI to its freight transport services such as reservation, dispatching and settlement functions.

Write to Ju-Hyun Lee at deep@hankyung.com

Jihyun Kim edited this article.
More to Read
Comment 0
0/300