Samsung begins mass production of sixth-generation HBM4 memory, set for Nvidia GPUs and AI accelerators, enhancing performance for generative AI systems.
© A. Krivonosov
Samsung Electronics is gearing up for a major technological breakthrough, with mass production of its sixth-generation high-speed HBM4 memory set to begin in late February. According to industry sources, the first chip shipments are expected next week, primarily destined for Nvidia to use in its next-generation graphics processors.
The new HBM4 memory will be a key component for AI accelerators, delivering exceptional computational power for generative AI systems. The current dominant HBM3E technology is gradually making way for HBM4, which promises significantly enhanced performance.
Industry experts note that the mass production of HBM4 confirms Samsung's technological leadership in the high-performance memory sector. The company has already passed Nvidia's quality certification and received orders to supply modules for Vera Rubin accelerators. Additionally, Samsung has increased the number of samples provided to customers for testing, aiming to accelerate the adoption of the new memory in real-world systems.
As a result, the high-performance memory market is preparing for its next phase, with HBM4 poised to become the new standard technology for AI accelerators and graphics solutions.