Samsung Electronics and Naver have embarked on a groundbreaking joint venture to develop an advanced artificial intelligence semiconductor solution. As reported by BusinessKorea, the collaborative effort aims to produce a chip with an energy efficiency approximately eight times superior to that of NVIDIA’s H100.
This innovative AI solution centers around a Field-Programmable Gate Array (FPGA) specifically tailored for Naver’s HyperCLOVA X large language model. According to Tom’s Hardware, citing Naver, the remarkable power efficiency of this AI chip is attributed to the utilization of LPDDR memory. However, finer details of the chip’s specifications and the timeline for its development remain under wraps.
The collaboration, initiated at the end of 2022, combines Samsung’s prowess in advanced process technology and memory innovations with Naver’s expertise in software and AI algorithms. Samsung brings to the table its experience in computational storage, processing-in-memory (PIM), processing-near-memory (PNM), and Compute Express Link (CXL) technologies. These are integral to the creation of AI chips capable of addressing the increasing demands of large-scale AI systems.
Samsung has already made significant strides in AI applications, having produced and marketed various memory and storage technologies. These include SmartSSD, HBM-PIM, and memory expansion modules featuring CXL interfaces.
Jinman Han, Executive Vice President of Memory Global Sales & Marketing at Samsung Electronics, emphasized the importance of this partnership. “Our collaboration with NAVER is aimed at developing state-of-the-art semiconductor solutions to overcome the memory bottleneck challenges in large-scale AI systems,” Han stated.
This collaboration between Samsung and Naver represents a significant leap in AI semiconductor technology, positioning them as formidable players in a market dominated by giants like NVIDIA. The development of this high-efficiency AI chip not only showcases the synergy between hardware and software expertise but also signifies a pivotal shift towards more energy-efficient and powerful AI computing solutions.
