It took just one day to find a new material for hair-loss care. Next, it combed through 8,000 stocks on the Korean and U.S. markets, and now it is designing immersion cooling materials that cool the heat of data centers. This is the story of "EXAONE," the generative artificial intelligence (AI) model developed by LG (003550). Beyond an AI that converses well with people, EXAONE is expanding its presence as a "practical AI" taking on two jobs and even three to solve various problems on industrial sites.
At "ICML 2026," the world's most prestigious machine learning conference held at COEX in Seoul starting on the 6th, LG AI Research unveiled cases of EXAONE's industrial applications, including new material development, finance, and data production automation.
What LG emphasized at this ICML was not EXAONE's performance. It was where it is used and how much work it takes over. This is also why LG presented at once the different cases of finding new materials, analyzing stocks, and creating data. The goal was to show that EXAONE has already become an AI that works on industrial sites.
The most eye-catching case was "Rhamsydil," a new hair-loss care material discovered by EXAONE. Co-developed with LG Household & Health Care, this material is an ingredient that EXAONE selected in just one day out of more than 420,000 candidate substances. It showed an effect in preventing hair loss without steroid-derived ingredients, and its research results were presented at the World Congress for Hair Research. The company is currently pursuing commercialization.
Having proven its practicality by finding a new hair-loss material, EXAONE has now taken on the development of core materials for data centers, a concentration of next-generation innovative technologies. LG AI Research unveiled for the first time at this conference a next-generation immersion cooling oil material co-developed with GS Caltex. At the center of such research is "EXAONE Discovery," a scientist platform in which AI reads and analyzes papers on its own.
EXAONE is also taking on a second job, performing the role of a financial analyst beyond a scientist. On this day, LG AI Research unveiled "EXAONE BI (Business Intelligence)," a finance-specialized AI agent that analyzes about 8,000 stocks listed on the Korean and U.S. markets every day. This model automatically writes not only scores for each stock but also analytical commentary at the level of a financial expert to help investors' decision-making.
EXAONE BI was used earlier this year in the U.S. stock forecasting AI service of the London Stock Exchange Group (LSEG), and this week it signed a contract with Koscom, a subsidiary of the Korea Exchange, supporting an AI service for analyzing the domestic stock market. In effect, it has proven its practicality in the financial domain, beyond science.
EXAONE has now reached the stage of producing data, beyond consuming it. LG AI Research's "EXAONE Data Foundry" automatically generates professional data and builds industry-specific AI models based on it. Through this, data productivity has increased about 1,000-fold.
Behind LG AI Research turning EXAONE into a "practical AI" lies accumulated research capability. The papers presented at this ICML numbered as many as 14. In particular, its new-material generation AI ranked second in the world in a global benchmark. Since its launch in 2020, its papers at top-tier AI conferences have reached 363, and its patent applications 838.
Lim Woo-hyung, co-director of LG AI Research, said, "EXAONE is no longer a laboratory AI but an expert AI that finds answers in the field," adding, "We will seize global AI leadership through a first-mover strategy to solve industrial challenges."
