Ray Wang on How AI Is Causing DRAM Prices to Surge

The conversation centers on the surging demand for memory chips, particularly DRAM, driven by the rapid expansion of AI technologies. Ray Wang, an analyst at SemiAnalysis, explains how the imbalance between supply and demand is due to underinvestment in DRAM capacity during the COVID-19 pandemic and the shift towards High Bandwidth Memory (HBM) for AI accelerators, which is more wafer-intensive. This shift is causing shortages in commodity DRAM, impacting consumer electronics like PCs and smartphones, with companies potentially raising prices or delaying product launches. The discussion also covers how AI's training and inference processes require significant memory, leading to increased token consumption and demand for both HBM and CPU DRAM. While increased production capacity and non-migration efforts could alleviate the shortage, significant shortages are still expected.