Can CXMT and Kimi K3 Break the Memory Supercycle?

The memory chip market is experiencing significant volatility as investors weigh the sustainability of massive data center spending against the current AI-driven supercycle. While high-bandwidth memory (HBM) demand remains robust, the complexity of its production and the resulting supply constraints create a structural shift in the industry. Chinese competitors like CXMT are expanding capacity, though they face technological hurdles and limited access to advanced equipment like EUV. Meanwhile, the emergence of low-cost, open-source LLMs like Moonshot’s Kimi raises questions about long-term AI capital expenditure returns, yet memory remains essential for both training and inferencing workloads. NAND flash has recently outperformed DRAM due to tighter supply-demand dynamics, but DRAM retains a healthier long-term outlook as HBM production continues to limit overall wafer availability for standard memory components.