[MWC26] LG Uplus and AI Research Institute Join Forces to Create AI That Solves Customer Problems

K-Exaone to Become the World's Top-Performing Model in the Second Phase
Exaone 4.5 to Be Developed as a VLM Model
200MW-Class AIDC to Be Established in Paju, the Largest in the Metropolitan Area
"AI Direction Focused on Solving Real Customer Problems"

LG Uplus and LG AI Research Institute unveiled their technology roadmap and announced the development roadmap for 'K-Exaone,' their proprietary artificial intelligence (AI) foundation model. The two companies plan to advance next-generation agentic AI technologies based on K-Exaone, aiming to help solve customers' real-world problems. LG Uplus also plans to build its AI data center in Paju, Gyeonggi Province, which is set to become the largest in the Seoul metropolitan area.

On March 1 (local time), the two companies held a press conference at the Inside by Melia Barcelona Apollo Hotel in Barcelona, Spain, where they revealed their AI technology roadmap. The press conference took place one day before the opening of Mobile World Congress (MWC) 2026, the world's largest mobile exhibition.

On the 1st (local time), at a press conference held at the Inside by Melia Barcelona Apollo Hotel in Barcelona, Spain, Lee Sangyup, Chief Technology Officer (CTO) of LG Uplus, and Lim Woohyung, Co-Research Director of LG AI Research Institute (right), answered questions from reporters. Provided by LG Uplus

The event was attended by Sangyeop Lee, Chief Technology Officer (CTO, Executive Vice President) of LG Uplus, and Woohyung Lim, Co-Research Director of LG AI Research Institute.

From the National AI 'K-Exaone' to Multimodal 'Exaone 4.5'... Aiming for the Best Performing Model

On the 1st (local time), Im Woohyung, Co-Research Director at LG AI Research Institute, is presenting at a press conference held at the Inside by Meli? Barcelona Apolo Hotel in Barcelona, Spain. Provided by LG Uplus

Woohyung Lim, Co-Research Director of LG AI Research Institute, emphasized, "The AI that LG pursues is not about competing for higher intelligence, but about creating a partner that helps humans in the real world and solves problems." He added, "We will focus on developing AI that creates tangible value in the physical world beyond the stage of artificial intelligence experience (AX)."

He also disclosed the future development roadmap for 'K-Exaone,' which passed the first phase evaluation of the government-led proprietary AI foundation model project. Previously, K-Exaone received the highest score in performance during the project's first phase evaluation.

LG AI Research Institute plans to develop the highest-performing language model among all existing global open-weight models during the second phase of the proprietary AI foundation model project, which will take place in the first half of this year. Specifically, the development of Exaone will focus on four main strategies: securing leadership in AI foundation model development, aiming for expert-level AI, expanding practical applications in industrial fields, and ensuring trust and safety for sustainable AI.

To that end, LG AI Research Institute has also announced plans to unveil 'Exaone 4.5,' a visual language model (VLM). A VLM is a multimodal model capable of comprehensively understanding not only text but also visual information such as images. Lim explained, "LG already developed Korea's first multimodal AI model, Exaone 1.0, five years ago in 2021," adding, "We are independently designing the model's learning and vision encoder structure internally."

LG AI Research Institute explained that 'Exaone 4.5' will enhance the model's usability by expanding its modalities, and that it will be the highest-performing model among open-weight models of its class worldwide. The institute plans to release the nearly completed 'Exaone 4.5' soon as an open-weight model.

200MW-Class AI Data Center Under 'One LG'... Agentic AI That Learns by Itself

On the 1st (local time), at a press conference held at the Inside by Melia Barcelona Apollo Hotel in Barcelona, Spain, Sangyeop Lee, Chief Technology Officer (CTO) of LG Uplus, is making a presentation. Provided by LG Uplus

LG Uplus revealed the scale of its upcoming Paju AI Data Center (DC) and outlined its infrastructure integration plans. The Paju AI Data Center, scheduled for completion in 2027, will have a capacity of 200 megawatts (MW) and can accommodate up to 120,000 graphics processing units (GPUs). According to LG Uplus, this will make it the largest data center in the Seoul metropolitan area. For the operation of the AI Data Center, including cooling and power supply, the capabilities of LG affiliates such as LG Electronics, LG Energy Solution, and LG CNS will be brought together.

Sangyeop Lee, CTO of LG Uplus, stated, "As K-Exaone moves towards completion next year, the 'One Team LG,' including LG Uplus, is preparing the largest AI Data Center in the metropolitan area," adding, "We are proceeding without a hitch to deliver the best experience with globally top-performing AI."

LG Uplus plans to offer customized AI services based on K-Exaone and the Paju AI Data Center. In addition, the company will introduce an integrated infrastructure combining LG AI Research Institute's Exaone, LG Uplus's enterprise AI platform, and FuriosaAI's neural processing unit (NPU).

On this day, LG Uplus also introduced its next-generation agentic AI strategy, which will be developed in collaboration with LG AI Research Institute. CTO Lee emphasized, "Along with improving the performance of the AI foundation model, it is crucial to establish a systematic evolutionary structure that can solve real customer problems," adding, "Securing an 'agentic architecture' for this purpose is at the core of LG Uplus's agentic AI strategy."

LG Uplus plans to implement a multi-level agentic AI structure that learns and evolves on its own, overcoming the limitations of conventional systems that merely respond to user queries. CTO Lee introduced the four core agentic architectures developed for next-generation agentic AI: self-advancement, model-data foundry, trustworthy integrated control, and hybrid AI infrastructure. The goal is to combine these architectures with LG AI Research Institute's AI foundation model to achieve optimal performance.

The company also expressed its ambition to address customers' challenges through agentic AI. CTO Lee stressed, "Our (AI) direction is to solve real customer problems," and added, "By collaborating with LG AI Research Institute, we aim to set the standard for agentic AI that provides convenience tailored to individuals."

This content was produced with the assistance of AI translation services.