CXMT’s reported move into NAND flash is an early-stage development plan, not a new source of commercial chip supply. Reuters reported on September 18, 2026, citing three people familiar with the matter, that the Chinese DRAM maker plans a NAND research-and-development line at its new Beijing site.
1
What has been reported?
The proposed line would support NAND development and trial production. CXMT has also established a research institute in Beijing with NAND among its projects, according to coverage of the Reuters report. The company has reportedly discussed its plans with potential customers, including a startup interested in chips for storage products used in AI systems and supercomputers. Those discussions do not establish that CXMT has secured orders or has NAND chips ready to ship.
4
11
Why would a DRAM maker pursue NAND?
DRAM and NAND serve different memory needs, so developing flash could broaden CXMT’s product range and potential customer base. Reuters linked the reported plan to a global memory shortage amid strong demand from AI servers. If CXMT eventually produces NAND commercially, it could compete with China’s YMTC as well as overseas suppliers. For now, that is a potential market outcome—not an established manufacturing capability.
1
11
How is this different from CXMT’s other memory work?
The reported NAND line should not be folded into accounts of CXMT’s separate Beijing DRAM expansion. Nor does the company’s fifth-generation DRAM platform entering mass production demonstrate progress toward commercial NAND production. Earlier HBM work concerns a form of DRAM, while XStacking is associated with YMTC’s flash technology; neither identifies a disclosed process for CXMT’s proposed NAND line.
3
6
16
What remains unknown?
No firm start date has been established for the NAND line. CXMT has not disclosed a NAND process specification or layer count, a wafer-capacity target, or a decision and timetable for moving from trials to mass production. A pilot line could help develop a product, but it is not a forecast of commercial output. The key question is whether—and when—the reported research effort becomes a viable NAND manufacturing business.
1
6
16