Defense-Specialized LLM Project: LIG·Samsung SDS Consortium vs. Hanwha·Naver Cloud Consortium Clash
A competition has ignited among South Korea's defense, artificial intelligence (AI), and cloud companies over a project to apply a defense-specialized large language model (LLM) to actual command and control systems. Hanwha, which has strengths in defense, and LIG are leading their respective consortia, while Naver Cloud, Samsung SDS, and others are competing in AI model development and platform-building capabilities.
According to industry sources on the 20th, the Ministry of National Defense has issued a bid for the “Ultra-Large AI-Based Command and Control System Technology” project. Two consortia are competing: one led by LIG Defense & Aerospace (D&A) with Samsung SDS, SK Telecom, and LG AI Research, and another led by Hanwha Systems with Naver Cloud and others. Proposal evaluations are currently underway, with the selection results expected as early as the end of this month.
The project aims to develop a defense-specialized ultra-large AI platform over 42 months (September 2026 to February 2030) with a total investment of 24.2 billion won. It will establish a foundation for fusing, analyzing, searching, and processing multi-modal battlefield information--including text, images, videos, audio, sensors, and radar--to realize Joint All-Domain Command and Control (JADC2). The Ministry plans to apply this platform to the Korean military's Command and Control System (KCCS) in the future.
The project consists of two sub-tasks: “Ultra-Large AI Model Platform Construction Technology” and “Data Fabric and Knowledge Base Construction Technology.”
The first task involves developing a general-purpose defense LLM and a joint battlefield LLM. Based on these, the project will implement real-time battlefield situational awareness, prediction, threat assessment, response strategy generation, weapon system recommendation/allocation, and user-customized common operational picture (COP).
The second task establishes the foundation for providing data and knowledge required for training and inference of the defense general-purpose LLM and joint battlefield LLM developed in the first task. It involves collecting, refining, and standardizing battlefield data provided by the Joint Chiefs of Staff to develop a data fabric and knowledge base for AI learning and inference.
In this bidding war, Hanwha and LIG, as prime contractors, lead the overall project and integration with command and control systems, while Naver Cloud, Samsung SDS, SK Telecom, and others compete in AI models and data/cloud platforms. Both consortia plan to utilize LG AI Research's Exaone during the model platform development process.
Since all participating companies have recently strengthened their defense AI and cloud capabilities, they are expected to emphasize these competencies in the bidding.
An industry insider stated, “Securing experience in creating and operating defense-specialized models is a significant opportunity for companies, so both consortia are fully committed. With major industry players involved, the final selection results will draw great attention.”
