The recent stock market performance of hyperscalers like Amazon, Alphabet, Microsoft, and Meta Platforms has been lackluster, with their stocks declining over the past month while the tech-heavy Nasdaq has risen. A key factor contributing to this underperformance is the shortage of high-bandwidth memory (HBM) chips, a specialized type of dynamic random access memory (DRAM) essential for AI computing. The limited supply of HBM chips, dominated by players like SK Hynix, Samsung, and Micron, has led to increased costs and reduced profitability for the hyperscalers. In response, these companies have attempted to co-design custom AI chips with partners like Marvell Technology and Broadcom, aiming to reduce their dependence on Nvidia. However, the memory shortage persists, and the hyperscalers' growth plans are being hindered. Meanwhile, memory and storage companies like Sandisk, Western Digital, and Seagate have seen their stocks surge, with a basket of memory stocks experiencing a 41% increase over the past month. The capital equipment companies, such as Applied Materials, Lam Research, and KLA Corp, which supply the necessary machinery for chip fabrication, are also benefiting from the strong demand. As the hyperscalers continue to invest heavily in AI, the memory shortage is likely to remain a significant challenge, and their stocks may underperform those of the memory and storage companies. To overcome this hurdle, the hyperscalers may need to rethink their strategies, such as Meta's plan to build a cloud business, or Microsoft's potential merger with OpenAI. Ultimately, the companies that can effectively navigate the memory shortage and develop innovative solutions will be better positioned for success in the AI-driven market.

Source: CNBC