en.Wedoany.com Reported - SK Hynix is accelerating the upgrade of its smart manufacturing system for high bandwidth memory (HBM) production lines, focusing on enhancing the wafer transfer fault prediction system and the real-time equipment location tracking system. The company's AI-based monitoring system "AMOS," first applied to the M14 Fab last year, will add AI agent functionality based on a large language model (LLM) this year. Meanwhile, the equipment location tracking system "MAPS," officially launched at the Icheon and Cheongju fabs in September last year, will be expanded to HBM processes and back-end processes. This move aims to address the growing demand for HBM by reducing production delays and improving quality and delivery stability to strengthen competitiveness. Detailed information has been disclosed in SK Hynix's "2026 Sustainability Management Report," which also includes AI transformation strategies for manufacturing and worksites, such as the generative AI service "LLM Chat" based on an internal security network.
AMOS is an AI-based monitoring system developed in-house by SK Hynix, with the core task of ensuring the stable operation of the automated material handling system (AMHS) responsible for wafer transport within the fab. The AMHS is a core logistics system that moves wafers between various process equipment on the semiconductor production line. As fab scales expand and processes become more complex, AMHS failures can lead to decreased productivity. Particularly amid the rapid growth in demand for high-value-added products like HBM, early detection of bottlenecks or equipment anomalies in wafer transport directly impacts production efficiency. Through AMOS, SK Hynix applies big data generated by logistics equipment and related systems to machine learning algorithms, enabling early detection of abnormal signs and providing real-time alerts and handling guidelines to engineers, thereby preventing failures and shortening recovery times.
According to the report, AMOS was first applied to the M14 Fab in 2025. SK Hynix plans to add LLM-based AI agent functionality to AMOS starting in 2026, expanding the AI service domain through interactive command execution and intelligent anomaly cause analysis, aiming to minimize production losses caused by AMHS failures.
In equipment management, SK Hynix is collaborating with SK Telecom to upgrade the real-time equipment location tracking and asset management system "MAPS (Manufacturing Asset Positioning System)." MAPS monitors the location information of thousands of production equipment in real time, reducing inconsistencies between IT system information and actual equipment locations, and improves asset management efficiency by integrating equipment status data with location information. SK Hynix completed technology validation in 2021 and officially launched MAPS at the Icheon and Cheongju fabs in September 2025. In 2026, the system's application scope is planned to expand to sub-components, HBM processes, back-end processes, and the Wuxi factory. The company expects this move to connect the real-time equipment location information required for implementing digital twins with related systems, accelerating the digital transformation of the production site, and serving as part of a future-oriented smart factory operation system.
AI applications are also expanding in internal business areas. While maintaining a closed network environment to prevent industrial technology leaks, SK Hynix has built an smart AI dialogue service "LLM Chat" based on an internal security network to leverage generative AI. LLM Chat supports various tasks such as report writing, data summarization, multilingual translation, image generation, and code development. Its application scope ranges from internal regulation searches to semiconductor manufacturing data analysis, and it has become a tool to reduce employee work trial-and-error and improve productivity. Since its official launch in October 2025, the service has accumulated over 24,000 users in about four months. SK Hynix plans to add agent-type AI functionality in the future, innovating work methods centered on AI to enhance overall productivity.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
