Samsung Electronics has announced its intention to significantly increase production of its high-bandwidth memory (HBM) and DRAM chips next year, responding to the surging demand from AI applications. Analysts anticipate that this expansion will help resolve the supply bottlenecks that have hindered upgrades in cloud and enterprise networks. The company noted a forecasted rise in demand for HBM4, planning to enhance its 1c (nanometer) DRAM production capabilities while strengthening its offerings of high-value products including DDR5 and LPDDR5x.
As Samsung prepares for HBM4 mass production, the company, which is already a crucial supplier of HBM3E chips, will stabilize supply for major AI semiconductor manufacturers such as NVIDIA. Experts suggest that this boost in memory production capabilities will prevent delays in infrastructure upgrades for hyperscalers and enterprises. However, some caution that even with increased capacity, meeting the explosive AI demand may still prove challenging due to ongoing manufacturing constraints, particularly in HBM production, which currently occupies significant DRAM wafer capacity. Gartner analysts emphasize that memory continues to be a silent driving force behind AI data center performance and scalability.
There is widespread optimism regarding the potential benefits of Samsung’s production increase, likely impacting various sectors of the data center ecosystem unevenly. While immediate relief from supply constraints is not expected, experts foresee the first visible effects in AI-centered data centers, with further ramifications in AI networking infrastructure down the line, particularly as DRAM and NAND price stabilization unfolds.
👉 Pročitaj original: CIO Magazine
