Core Insights - Nvidia is facing significant supply chain challenges as it approaches the production timeline for its next-generation AI accelerators, potentially leading to a shift in its procurement strategy to prioritize supply stability over extreme performance [1][5] - The overall market supply dynamics are heavily influenced by Nvidia's final procurement decisions, with expectations that it will adopt a more pragmatic approach by procuring both top-spec and slightly lower-spec versions of HBM4 [1][2] Group 1: Supply Chain Challenges - Major storage chip suppliers are encountering dual obstacles in capacity expansion and yield improvement, which may force Nvidia to relax its technical specifications for HBM4 [1] - Samsung Electronics, despite being ahead in the qualification process for HBM4, may struggle to meet Nvidia's demand for large-scale production due to yield and capacity limitations [3] - SK Hynix is also facing challenges in meeting performance standards for HBM4, with its products struggling to reach the required 11Gbps performance level [4] Group 2: Procurement Strategy - Nvidia is likely to implement a dual-track procurement strategy, purchasing both the highest specification HBM4 and slightly lower-spec versions to ensure stable supply from major suppliers like Samsung, SK Hynix, and Micron [5] - This strategy is seen as a necessary measure to ensure the timely delivery of the next generation of AI infrastructure amid increasing shortages in the storage chip market [5]
报道:英伟达或放宽HBM4规格要求,因三星、SK海力士面临产能和良率限制