Samsung to begin large-scale HBM4 production later this month
Published: 08 Feb. 2026, 12:38
Updated: 08 Feb. 2026, 18:27
Samsung Electronics HBM4, a sixth-generation high-bandwidth memory solution for AI and high-performance computing applications, on display during the 2025 Korea Tech Festival in southern Seoul on Dec. 4, 2025. [REUTERS/YONHAP]
Samsung Electronics will begin mass production of the world's first sixth-generation high-bandwidth memory, known as HBM4, later this month, industry sources said Sunday.
The South Korean tech giant plans to start shipping HBM4 chips as early as next week, following the Lunar New Year holiday, for use in graphics processing units (GPUs) made by Nvidia, the sources said. Nvidia's GPUs are widely used in generative AI systems.
"Samsung, which has the world's largest production capacity and the broadest product lineup, has demonstrated a recovery in its technological competitiveness by becoming the first to mass-produce the highest-performing HBM4," said an industry source, who declined to be named.
The global high-bandwidth market is currently dominated by fifth-generation HBM3E chips, but industry observers expect HBM4 to emerge as a key technology. Nvidia plans to adopt HBM4 in its next-generation AI accelerator, named Vera Rubin.
Samsung has passed Nvidia's quality certification process and secured purchase orders, the sources said, noting that the production schedule was finalized to align with Nvidia's launch plans for Vera Rubin.
The volume of HBM4 samples supplied by Samsung for customer-side module testing has also increased significantly under the latest purchase order, the sources added.
Yonhap





with the Korea JoongAng Daily
To write comments, please log in to one of the accounts.
Standards Board Policy (0/250자)