The centre of gravity in the telecommunications industry is rapidly shifting to artificial intelligence (AI). All three major telecom operators are reshaping their business structures after declaring plans to transform into "AI companies". Network coverage and speed used to be the core of competition, but AI has now emerged as the yardstick that will determine business success or failure.
Industry officials describe this as a restructuring process. Amid the shift, SK Telecom, KT and LG Uplus are offering different solutions. With AI positioned as a core growth engine, the market is taking shape as a "three-colour competition" among telecom operators, industry watchers say.
SKT takes head-on approach with "full-stack AI" as it expands infrastructure, models and services
SKT is promoting full-stack AI that spans AI infrastructure, models and services. It is building large-scale computing infrastructure based on graphics processing units (GPUs), including data centres, and expanding a GPUaaS (GPU as a Service) business that provides the infrastructure as a service. It is putting in place a structure that covers the entire AI value chain.
Ryu Jong-heon (정재헌), SKT's CEO, said in a letter sent to shareholders ahead of last month's annual general meeting, "If our AI business so far was about incubating various areas, we will now focus more on businesses where SKT can be competitive and secure sustainability in AI competition that is expanding without limit."
SKT is also focusing full efforts on its "sovereign AI" strategy. Sovereign AI refers to building customised infrastructure that enables a country or company to use AI without relying on an external cloud for data. It can also help create new markets as data sovereignty issues expand. SKT is seeking to create new markets with a "sovereign AI package" that provides, in an integrated way, an AI data centre backed by group capabilities, AI services for industry and enterprises, and A.X K1.
Partnerships are also a key pillar supporting SKT's AI business. The company recently signed a strategic partnership with Arm and Rebellions for next-generation AI infrastructure innovation and agreed to cooperate on AI server development. Earlier, at MWC26 held last month in Barcelona, Spain, it signed agreements with global server maker Supermicro and energy management company Schneider Electric to cooperate on AI data centres. Through such cooperation, SKT aims to strengthen its full-stack AI strategy and provide all solutions needed to expand the AI ecosystem.
SKT advanced to a second-stage evaluation of the government's "independent foundation model project". Its goal is to be selected as a final elite team through its self-developed "A.X K1". A.X K1 is the country's first hyper-scale AI model in the 500B class and showed performance similar to or better than global AI models such as DeepSeek V3.1 in major benchmarks. Starting this year, it plans to add multimodal functions in stages and continue follow-up development to expand to parameters in the trillion range. Industry watchers broadly say market dominance would also rise once consumer and enterprise services based on the upgraded A.X K1 are launched in earnest.
KT speeds up transformation into an "AX company" targeting B2B
KT is pursuing an "AX (AI transformation) company" strategy centred on the business-to-business (B2B) market. The core is an attempt to shift into a platform operator that changes overall corporate operations around AI, rather than simply supplying AI solutions. An organisational reshuffle carried out immediately after CEO Yun-young Park (박윤영) took office last month also focused on the AX company transition. It reorganised its research and development (R&D) organisation into the "AX Future Technology Institute" and set up an "AX Business Division" to strengthen competitiveness in the B2B AX field.
KT has been assessed as somewhat behind rivals in in-house AI capabilities. Its proprietary large language model (LLM), "mi-deum", did not generate as much impact as expected, and it also suffered a setback in the independent AI foundation model project. Its push for a partnership with Microsoft also drew an assessment that it chose a "detour" rather than compete on its own AI technological capabilities.
KT adjusted its strategy to target the B2B market through the new organisation. The company says setting up the AX Future Technology Institute was also a choice to raise technological leadership and expertise. The market views KT as choosing practical gains. An industry official analysed, "It appears to be focusing on the B2B market, where monetisation is likely to be high, by leveraging its existing base of corporate customers."
Ultimately, the goal is to support AI-related solutions, including data infrastructure and operating systems, and link them to earnings. Park recently presented "a leap to become an AX platform company" as a key business direction after a meeting between the deputy prime minister and minister of science and ICT, Kyung-hoon Bae (배경훈), and the CEOs of the three telecom operators. Park explained, "If AI services are actors on a theatre stage, we are an AX platform company that builds that stage."
LG Uplus seeks leap into AI software company with ixi-O
LG Uplus, under CEO Beom-sik Hong (홍범식), is putting "safety" and "security" forward as core keywords. The approach is to turn telecom security issues, which have emerged as a risk across the industry, into competitiveness. Hong, who took office at the end of 2024, has shored up the company's fundamentals over the past year or so while stressing "basics". More recently, the company unveiled a strategy to develop its voice AI capabilities built through the AI agent "ixi-O" into a range of areas.
Hong said at a press briefing at MWC26 last month, "We will become an AI-centred software (SW) company that leads solutions in telecommunications and AX technology," and declared plans to enter the global market through a two-track approach of supplying the ixi-O service itself or providing an AI technology stack. It also seeks financial strengthening in a direction that maximises profit growth rather than revenue growth, based on the high-margin structure of the SW business.
LG Uplus also plans to introduce "ixi-O Pro", an enhanced version of its existing service, before moving into global expansion in earnest. Ixi-O Pro is an AI agent that analyses not only speaker identification but also tone, conversational flow and emotional state, and proactively suggests customised information to customers. Through this, it is pushing to build a service ecosystem that connects physical AI and various devices.
"Three firms, three strategies" as profitability fight remains
Industry officials are focusing on the fact that telecom operators' roles themselves are changing. They say this year's biggest task is to reshape business structures around AI and discover new growth drivers. The broad view in the industry is that companies must now focus on monetisation to recoup large-scale investments.
An industry official said, "The key is how to graft telecommunications network technology built up so far onto AI services," adding, "All three have finished setting specific roadmaps. Now is the time to prove it with results."
With global big tech companies simultaneously dominating AI models and cloud infrastructure, how telecom operators secure differentiated competitiveness is also seen as an important variable. Another industry official said, "This is currently a transitional period where investment and strategic competition overlap," adding, "In the end, the operator that first builds a sustainable profit model will take market leadership."