Samsung Electronics Co., Ltd. has announced it has started mass production and supply of its latest-generation HBM4 memory chips to an undisclosed client. This development signifies a strategic advance for the company within the highly competitive high-end AI memory market. The announcement, made on Thursday, represents a significant milestone for the chipmaker. Samsung is currently focusing its efforts on meeting the surging market demand driven by GPUs from NVIDIA, which are the leading accelerators for training and running AI models. Until recently, competitor SK Hynix held a technological advantage in the High Bandwidth Memory (HBM) sector and secured the majority of orders from NVIDIA. HBM is an essential core component for AI accelerators. However, in recent months, Samsung has substantially narrowed the technological gap with its rival. Regarding production capacity, Samsung explicitly stated it is actively expanding its HBM4 production lines and forecasts that its total HBM sales in 2026 will more than triple compared to 2025. With HBM4 now entering commercial mass production, the roadmap for Samsung's next-generation products has also been revealed. The company expects to begin sample shipments of HBM4E in the second half of 2026 and plans to deliver customized, differentiated HBM solutions based on client specifications in 2027.