CXMT Moves Into NAND as YMTC Prepares DRAM, Setting Up a Chinese Memory Rivalry
China’s memory market is entering a new phase as CXMT moves into NAND while YMTC prepares its first DRAM production. YMTC is reportedly targeting LPDDR5 trial production before the end of 2026, while CXMT is developing NAND technology and planning a research production line in Beijing. The shift puts the country’s 2 largest memory manufacturers on a path toward direct competition as AI infrastructure continues driving demand for both memory and storage.
According to a September 22 industry report, YMTC plans to begin LPDDR5 trial production at its Wuhan Phase 3 operation by the end of 2026, with initial DRAM capacity reportedly around 25,000 wafers per month. This would mark a major departure for a company best known for its Xtacking based 3D NAND technology. Trial production should not be confused with mature commercial output, since the initial stage is intended to validate equipment, processes, yields, and product quality before larger scale manufacturing begins.
CXMT is moving in the opposite direction. The company reportedly plans to establish a NAND research and production line at a new Beijing facility and has discussed future supply with potential customers, including an AI focused company developing storage products for servers and supercomputers. Unlike YMTC’s near term DRAM trial schedule, there is currently no confirmed date for CXMT NAND production. Industry estimates suggest commercial manufacturing could begin around 2028 or 2029, but CXMT has not publicly committed to that timeframe.
The timing is notable because both companies are already strengthening their core businesses. CXMT recently placed its fifth generation DRAM platform into mass production and continues moving into server and higher capacity memory. We have also seen CXMT DDR5 enter products from international memory brands. YMTC, meanwhile, has continued increasing NAND output and remains one of China’s most important advanced storage manufacturers.
AI demand is helping create the conditions for this crossover. Supply chain analysis estimates that hyperscale AI infrastructure investment is growing much faster than NAND manufacturing capacity, while training, inference, and enterprise AI systems require increasingly large volumes of high performance storage. That creates an opening for additional suppliers at a time when memory and storage availability remain under pressure.
The transition could also create technical and legal challenges. YMTC has built a substantial patent portfolio around Xtacking and advanced 3D NAND manufacturing, while CXMT would need its own path toward high layer NAND under existing equipment restrictions. Similar challenges could emerge as YMTC moves into DRAM, where CXMT has spent years building process expertise and intellectual property. Reports also suggest CXMT has recruited employees with experience at YMTC, adding another layer to the growing competition between the 2 companies.
Calling this a full memory war is still premature. YMTC is preparing DRAM trial production, while CXMT remains much earlier in its NAND effort. Neither company has yet shown that it can challenge the other at mature commercial scale outside its traditional market.
What has changed is the relationship between them. CXMT and YMTC were previously focused on largely separate segments, but AI demand and high memory pricing are making broader portfolios more attractive. If both strategies succeed, the impact could reach well beyond China by giving PC manufacturers, smartphone brands, server vendors, and storage companies more sourcing options across both DRAM and NAND.
Could CXMT and YMTC eventually become serious global competitors across both DRAM and NAND, or will their existing specialties remain their strongest advantage?
