Kioxia (285A JT) reportedly pushes the use of NAND to replace part of DRAM functions via a CXL memory-expansion module; module could double memory capacity and raise performance by 30%, and plans to supply SSD samples in 2028

The mechanism here is CXL-attached memory expansion: if NAND can absorb part of the DRAM workload in data-center servers, the addressable market for flash widens at the expense of DRAM makers, so the read-across typically splits the memory complex, positive for NAND-exposed names, competitive pressure for DRAM-heavy peers. Past episodes of storage-class memory commercialization show the pattern clearly: headline performance claims arrive years ahead of volume economics, and the trade has tended to fade unless hyperscaler qualification follows. The figure cited is a company claim on an unlaunched module, and the stated sample timeline is distant, which historically means the near-term revenue impact is nil and the move is sentiment and positioning. What separates signal from noise in this sequence is customer engagement: design-win announcements, CXL ecosystem partners, and any capex or wafer-mix shift toward this product line. Until such follow-ons appear, this is a directional story about memory architecture rather than an earnings event.