Samsung is making significant strides in the tech world, particularly in the realm of high-bandwidth memory (HBM). The company recently announced plans to supply its cutting-edge 5th-generation HBM3E memory for NVIDIA’s top-tier AI accelerators, keeping its hope alive of becoming a primary HBM supplier for the graphics juggernaut.
Earlier, Samsung faced challenges in securing NVIDIA as a customer, causing speculation about its status in the supply chain. However, remarks from Kim Jae-jun, VP of Samsung’s Memory Business Division, have suggested that the company is actively working on supplying HBM3E memory to a “major customer,” believed by many analysts to be NVIDIA.
According to Kim, mass production and sales of the HBM3E 8-stack and 12-stack products are underway. The significant testing phases have been completed with this key client, hinting at an expansion in sales as early as the fourth quarter. Although NVIDIA hasn’t been named directly, these developments suggest Samsung is inching closer to solidifying its partnership with the graphics leader.
Achieving this would be monumental for Samsung, particularly as the firm is seeking to establish itself prominently in the AI sector. A partnership with NVIDIA would underscore Samsung’s capabilities in the AI landscape and potentially elevate their status among AI hardware powerhouses.
Looking to the future, Samsung is preparing to embrace its next-gen HBM4 process. There are plans to possibly partner with TSMC for essential semiconductor components, as the company aims to lead early in the anticipated HBM4 market. The next advancement with 6th-gen HBM memory could bring remarkable improvements in AI technology, integrating advanced memory and semiconductors. For Samsung, the future in high-bandwidth memory looks promising, with a clear focus on innovation and strategic partnerships.
