Samsung memory chief, Nvidia CEO discuss next-generation HBM cooperation - Yonhap News Agency

Samsung Electronics' memory business head, Jun Young-hyun, met with NVIDIA CEO Jensen Huang in South Korea to discuss long-term cooperation in next-generation high-bandwidth memory (HBM) technologies.... Read original →

Samsung’s deepening HBM4E/HBM5 collaboration with NVIDIA signals a strategic pivot from compute-centric to memory-compute co-optimized AI hardware stacks. Technically, bandwidth bottlenecks in HBM now constrain large-scale AI training; Samsung’s potential inclusion in NVIDIA’s validated ecosystem would accelerate its TSV and hybrid bonding capabilities, spurring demand for silicon interposers and advanced packaging tools. On compliance, looming U.S.-ROK export controls on advanced memory could raise costs for non-U.S. customers, compelling Samsung to erect complex geo-compliance firewalls. With SK hynix commanding over 70% of HBM3E supply, Samsung is betting on co-designed HBM5 to secure architectural influence—prompting SK hynix to likely deepen ties with AMD and foundries in Taiwan, China. Within 18 months, HBM will shift from commodity component to strategic leverage, turning memory makers into co-architects of AI systems.