[Barcelona, Spain = DigitalToday reporter Jin-ho Lee] "We are past the stage of thinking about what we can do with AI. A major transformation will happen in coding agents, hardware and data centres."

SK Telecom is making a decisive push into AI with a “full-stack” strategy. By pooling capabilities across group affiliates, it is building a portfolio that spans infrastructure, models and services and is accelerating its shift into an AI-leading company.

Seokkeun Jeong (정석근), SKT's chief technology officer and head of its AI CIC, said in an on-site interview at MWC26 in Barcelona, Spain, that change in the past 3 years has been greater than in the previous 20 years. He said the company is discussing optimisation that ties together power plants, data centre buildings, servers, chips and software.

◆Pooling group capabilities... building an end-to-end AI data centre

Jeong said he is confident the company can deliver the broad range of services needed for AI business based on group capabilities. AI data centres that combine SK hynix’s semiconductor capabilities, SK Innovation’s energy capabilities, SK ecoplant’s construction and SK Broadband’s fixed-line technology are an area SKT is particularly confident about.

Jeong expects AI data centres to be split into two tracks: ultra-large hyperscale and local edge. He stressed that, as a telecom operator, SKT can absorb diverse demand because it has capabilities to build large data centres as well as edge data centres, which are similar to operating base stations.

Jeong said it is difficult to find a company in South Korea that can deliver end-to-end with such a portfolio. He said that globally it has a structure similar to big tech such as Google, Amazon and Microsoft.

He also said data centres will likely be split into ultra-large hyperscale and local edge. He stressed that telecom operators have capabilities for large data centres, and edge data centres require capabilities similar to operating base stations, allowing them to balance both axes.

◆Targeting a national AI with A.Dot K-1... "high industrial usability"

Jeong also shared his thoughts on the government’s “independent AI foundation model” project. SKT has advanced to the second-stage evaluation with its “A.X (A.Dot) K-1” model. A.Dot K-1, the country’s largest 519B (519 billion-parameter) model, is seeking to evolve beyond existing large language models into multimodal capabilities.

Jeong stressed that A.Dot K-1 will have high usability across various industries going forward. He said the best strategy is to build a very large model and then optimise it in pieces for each specific domain. He said the company is pursuing a full-stack strategy by designing infrastructure optimised for the model as well.

The AI agent “A.Dot” is expected to roll out a paid model within the year. A.Dot has a broad user base, with more than 10 million people using it. SKT is reviewing monetisation in the form of subscriptions or bundled products centred on killer services within A.Dot. Jeong said it is first necessary to find use cases that make customers willing to pay happily, and that the company is continuing its review.

Jeong took a cautious view of predictions that 6G will become core infrastructure in the AI era. Qualcomm, which delivered a keynote speech at MWC this year, presented 6G as a puzzle piece that will complete the AI era, saying faster communications can absorb data uplink demand and improve service stability. Jeong said his honest view is that there still do not seem to be clear use cases, and that it is telecom operators’ job to work hard to find them.

Jeong, meanwhile, explained his dual role as head of the AI CIC and CTO as a matter of “balancing technology and business.” He said the company’s AI business ultimately commercialises technology. He said money is not earned simply because one wants to earn it, but is created in the process of commercialising technology. He added that the company will make efforts to go deep into AI technology and build differentiated business models, saying it needs to understand both technology and business.