Overview
- LG AI Research and Nvidia announced an expanded collaboration Wednesday after executives met in Seoul on Tuesday to pair LG’s EXAONE with Nvidia’s Nemotron for sector-specific systems in areas like manufacturing and biotechnology.
- EXAONE, LG’s multimodal foundation model, progressed from version 3.0 to the new 4.5 release using Nvidia GPUs along with the NeMo development framework and TensorRT-LLM to speed training and inference.
- Nemotron bundles open models, datasets, and tools that help teams build specialized AI, and LG says it has already used Nemotron datasets to improve EXAONE’s performance.
- The Seoul meeting brought together LG’s Lim Woo-hyung and Lee Jin-sik with Nvidia’s Bryan Catanzaro and Jeong So-young to map out closer R&D and ecosystem work.
- The companies frame the push as part of Korea’s sovereign AI strategy, noting South Korea’s third-place rank in Stanford’s AI Index model count led by four EXAONE systems, while concrete rollout timelines remain undisclosed.