Samsung announced a partnership with South Korean search giant NAVER, hoping to develop a dedicated chip for large-scale AI models. The two sides said that in recent years the amount of AI data has grown exponentially, and current technology has limitations in terms of performance and efficiency to properly handle such a large workload.
For this reason, Samsung and NAVER have decided to work together to create semiconductor chips and other solutions optimized for AI. In this collaboration, the hardware and software will be combined with Samsung’s compute storage (SmartSSD), CXL (Compute Express Link), PIM (Processor In Memory) and PNM (Processing Near Memory) technologies, combined with NAVER’s extensive experience in AI algorithms, to significantly increase AI workloads.
Samsung Electronics will contribute its hardware and manufacturing expertise, while NAVER will bring its expertise in the development and verification of AI-based algorithms, and NAVER will improve HyperCLOVA (an exascale model with over 200 billion parameters) and its compression algorithm to increase efficiency.
