[News] Samsung and Naver Jointly Invest in New AI Chip, as Naver Reportedly Claiming an Eightfold Performance Boost Compared to NVIDIA H100

2023-12-25 Semiconductors tomaslin

Samsung Electronics and South Korean internet giant Naver have joined forces to invest in an artificial intelligence semiconductor solution. According to BusinessKorea’s report, the energy efficiency of the first solution chip from the two companies is expected to be roughly eight times higher than competitors like NVIDIA H100.

This new solution is based on a Field-Programmable Gate Array (FPGA) customized for Naver’s HyperCLOVA X large language model.

Per Tom’s Hardware cited from Naver, it indicated that this AI chip is eight times more power efficient than NVIDIA’s AI GPUs H100 thanks to the usage of LPDDR memory. However, specific details remain undisclosed, and the timeline for product development by the two companies is yet to be clarified.

Samsung and Naver began their collaboration at the end of 2022, utilizing Samsung’s advanced process technology, expertise in memory technologies like computational storage, processing-in-memory (PIM) and processing-near-memory (PNM), as well as Compute Express Link (CXL). Naver’s strengths in software and AI algorithms are also leveraged in this collaboration.

Samsung has already produced and sold various types of memory and storage technologies for AI applications, including SmartSSD, HBM-PIM, and memory expansion modules with CXL interfaces, all crucial for the upcoming AI chips.

“Through our collaboration with NAVER, we will develop cutting-edge semiconductor solutions to solve the memory bottleneck in large-scale AI systems,” said Jinman Han, Executive Vice President of Memory Global Sales & Marketing at Samsung Electronics.

Read more

(Photo credit: Samsung)

Please note that this article cites information from Samsungtomshardware and BusinessKorea