China’s two biggest homegrown memory chip makers are about to start competing on each other’s turf. ChangXin Memory Technologies (CXMT), the country’s leading DRAM producer, is preparing a research-and-development line for NAND flash memory at a new plant in Beijing, according to a report dated September 18, 2026 and confirmed by outlets including Wccftech. At the same time, Yangtze Memory Technologies (YMTC), China’s dominant NAND flash maker, has already sent low-power DRAM samples to customers and is weighing a full move into CXMT’s core DRAM business. Three people with knowledge of the matter described CXMT’s plans to reporters, marking one of the more significant strategic pivots in China’s semiconductor industry this year.
Neither company has confirmed a commercial launch date, production capacity, or customer list. But the direction is clear: China’s memory sector, long split between a DRAM specialist and a NAND specialist, is heading toward direct overlap. That shift lands in the middle of a global memory chip shortage that has already squeezed everything from AI servers to budget phones.
CXMT Prepares to Enter the NAND Flash Market
CXMT built its reputation on DRAM, the fast, volatile memory that handles active workloads in PCs, phones and servers. Now it’s reportedly setting up a research-and-development production line dedicated to NAND flash, the non-volatile storage technology used in SSDs, phones and increasingly in AI infrastructure. The new line is tied to a plant CXMT is building in Beijing, separate from its existing DRAM manufacturing base.
The move would put CXMT in direct competition with YMTC, the company that has effectively owned China’s domestic NAND supply chain since it emerged as a serious 3D NAND producer in the late 2010s. It would also put CXMT closer to global NAND leaders such as Samsung Electronics, which the report says still dominates the flash-memory market alongside other foreign manufacturers.
What a Customer Win Would Signal
CXMT is reportedly in early discussions with potential customers, including an unnamed startup that wants NAND flash for storage used in AI systems and supercomputers. That detail matters more than it looks. AI infrastructure is exactly the segment where global buyers are currently rationing supply, and it’s the same segment driving the price spikes covered in our look at how China’s AI chip prices jumped as the HBM shortage bit into supply. If CXMT can land even one credible AI-storage customer, it changes the calculus for how seriously the rest of the industry treats its NAND ambitions.
YMTC’s Quiet Move Toward DRAM Production
YMTC’s side of the story moved first, and moved more quietly. An earlier Reuters report cited in the September coverage says YMTC has already sent low-power DRAM samples to customers, a step that typically precedes a formal qualification process rather than a marketing stunt. DRAM has been described in the reporting as CXMT’s core market, which makes YMTC’s sampling a direct challenge rather than a side project.
Nothing in the available reporting confirms a commercial DRAM launch date, a production capacity target, or which customers received the low-power samples. What’s confirmed is narrower: samples went out, and YMTC is considering entry into DRAM as a business line. That’s a meaningful signal on its own, since sampling requires real engineering resources, but it stops well short of a market entry.
Why Low-Power DRAM First
Low-power DRAM, the class of memory built for phones, tablets and increasingly for power-constrained edge AI devices, is a logical entry point for a company without DRAM manufacturing history. It’s a smaller, faster qualification cycle than high-performance server DRAM or HBM, and it lets a new entrant build a customer relationship before attempting the harder segments. Whether YMTC follows that path toward server-grade DRAM, or stays focused on mobile and edge use cases, is one of the open questions this story leaves unanswered.
Inside CXMT’s New Beijing Research Institute
CXMT has set up a research institute in Beijing, and NAND development is listed among its projects there, according to the report. That institute appears to be the organizational home for the NAND R&D effort described above, separate from CXMT’s existing DRAM fabs elsewhere in China. Setting up a dedicated research arm, rather than folding NAND work into an existing DRAM team, suggests CXMT is treating this as a genuine second product line rather than a side experiment.
The choice of Beijing is also notable in the context of China’s broader push for domestic chip capacity. Beijing has become a hub for semiconductor R&D investment as the country works to reduce reliance on foreign suppliers, a dynamic that also shows up in reporting on Huawei’s own moves to curb AI chip exports while scaling its Atlas compute platform. Memory and compute are increasingly treated as parts of the same strategic problem inside China’s chip policy.
Why Now: The Memory Shortage Driving China’s Push
The timing isn’t a coincidence. Global memory prices have climbed sharply through 2026 as AI data center buildouts soak up DRAM and HBM capacity that would otherwise go to PCs, phones and general-purpose servers, a trend memory-market researcher TrendForce has tracked closely across its pricing reports. That crunch has already reshaped pricing across consumer hardware, a trend our earlier coverage tracked when memory costs ate into the margins on budget phones and laptops, and again when DRAM shortages left billions in chips stranded at TSMC ahead of iPhone 18 production.
For Chinese buyers specifically, the shortage carries extra weight. Beijing has spent years pushing domestic alternatives to Nvidia’s AI accelerators, and those accelerators need memory to function. A domestic chip strategy that stops at compute and ignores memory leaves a gap that foreign suppliers can exploit, or worse, restrict. CXMT and YMTC expanding into each other’s markets reads as an attempt to close that gap from both directions at once, rather than waiting for one company to master both DRAM and NAND on its own timeline.
The Global Memory Market CXMT and YMTC Are Chasing
Samsung Electronics and other foreign manufacturers still dominate the global flash-memory market, according to the report, and that dominance extends to DRAM and HBM as well. Getting a rough sense of who else sits at that table helps explain why CXMT and YMTC’s moves matter beyond China’s borders.
| Company | Historical core business | Headquarters | 2026 strategic move |
|---|---|---|---|
| Samsung Electronics | DRAM, NAND, HBM | South Korea | Maintains leading share across memory segments, per the report |
| SK hynix | DRAM, NAND, HBM | South Korea | Named among HBM market leaders in the reporting |
| Micron Technology | DRAM, NAND | United States | Named among HBM market leaders in the reporting |
| ChangXin Memory Technologies (CXMT) | DRAM | China | Preparing NAND flash R&D line at new Beijing plant |
| Yangtze Memory Technologies (YMTC) | NAND flash | China | Sampling low-power DRAM with customers |
That table is a simplification. It doesn’t capture exact market share, since the report doesn’t publish current percentage figures for either company, and any specific share number attributed to CXMT or YMTC in 2026 should be treated as unconfirmed until a company or a named research firm publishes it directly.
CXMT vs YMTC: Comparing China’s Two Memory Chip Giants
Put side by side, CXMT and YMTC are mirror images of each other right now, each one strong in the segment the other is trying to enter.
| Metric | CXMT | YMTC |
|---|---|---|
| Established strength | DRAM | NAND flash |
| New target market | NAND flash | DRAM |
| Reported new facility | New plant in Beijing with dedicated NAND R&D line | Not specified in current reporting |
| Current market stage | R&D and research institute stage, early customer talks | Sampling low-power DRAM with customers |
| Named potential customer type | Unnamed startup building AI/supercomputer storage | Not disclosed in current reporting |
| Confirmed production timeline | Not confirmed | Not confirmed |
The symmetry is the story here. Two companies that spent close to a decade specializing in opposite halves of the memory business are now moving toward the same middle ground, at roughly the same time, without either one publicly framing it as a rivalry. Whether that’s coordinated industrial policy or simple competitive instinct is something neither company has addressed on the record.
Samsung, SK Hynix and Micron: How the Incumbents Are Positioned
None of this happens in a vacuum. Samsung, SK hynix and Micron have spent decades building manufacturing scale, yield expertise and customer relationships across DRAM, NAND and now HBM, the stacked memory format that hardware outlets like Tom’s Hardware have flagged as the bottleneck ingredient for AI accelerators. That HBM bottleneck is exactly what has driven the broader shortage discussed in our coverage of Intel’s warning that memory supply won’t ease until 2028.
A CXMT or YMTC entry into a new memory segment doesn’t threaten that incumbent position overnight. Qualifying a new memory product with major buyers, especially for anything touching AI infrastructure, takes years of reliability testing, not months. What it does change is the long-term supply picture. If China succeeds in building two credible domestic suppliers instead of one specialist each in DRAM and NAND, it reduces how much leverage Samsung, SK hynix and Micron hold over Chinese buyers specifically, even if global market share barely moves.
Market Impact: What a Turf War Could Mean for Chip Prices
It’s tempting to read “China memory chip turf war” as a headline about falling prices. That’s premature. Both companies are described as being at an early, R&D-stage of their new product lines, not ramping commercial volume. Nothing in the current reporting supports a claim that CXMT NAND or YMTC DRAM will hit the open market with pricing that undercuts Samsung, SK hynix or Micron in the near term.
What’s more plausible in the short run is a domestic effect inside China. If CXMT and YMTC each build out a second product line, Chinese OEMs building phones, laptops and AI servers gain an additional local sourcing option even before either company reaches full-scale production, simply because supply contracts and pilot-line allocations can start well ahead of mass manufacturing. That matters given how tight global allocation already is, a squeeze that has shown up directly in device pricing, as covered in our report on budget hardware absorbing rising memory costs.
Longer term, if both companies hit commercial scale, the more interesting effect is competitive pressure on each other rather than on Samsung, SK hynix or Micron. A CXMT that succeeds in NAND stops needing to rely on YMTC or foreign suppliers for flash storage in its own product bundles. A YMTC that succeeds in DRAM does the same in reverse. That’s a China-internal supply chain effect first, and a global pricing effect only if and when either company scales past domestic demand.
Historical Context: How China Built Its Domestic Memory Industry
CXMT and YMTC didn’t emerge overnight. Both companies trace back to a period in the mid-2010s when Beijing began treating memory chip self-sufficiency as a national priority, backing domestic fabs with state-linked investment funds after years of near-total reliance on South Korean, Japanese and American suppliers. YMTC built its name in 3D NAND, closing the technology gap with established flash makers faster than most Western analysts initially expected. CXMT took the DRAM path, a notoriously capital-intensive business dominated for decades by Samsung, SK hynix and Micron.
U.S. export controls aimed at Chinese semiconductor manufacturing, tightened repeatedly since 2022 and tracked in detail by regional business outlets like Nikkei Asia, pushed both companies to lean harder on domestic equipment and domestic R&D rather than importing advanced foreign tools. That pressure is part of the backdrop for why cross-entry into each other’s markets makes strategic sense now: building two independent supply chains from scratch is slower and more expensive than each company borrowing scale and expertise from the other’s existing manufacturing base.
What Remains Unconfirmed
It’s worth being precise about what this story actually confirms versus what it implies, because the gap between the two is wide. Confirmed: CXMT is setting up a NAND R&D production line tied to a new Beijing plant, has a Beijing research institute with NAND among its listed projects, and is in early talks with a potential AI-storage customer. Confirmed: YMTC has sent low-power DRAM samples to customers and is considering a DRAM business.
Not confirmed: final product specifications for either company’s new line, production capacity, a commercial launch date, named customers beyond the one unnamed AI-storage startup, and pricing. Also not confirmed: that either company has formally declared a “turf war” against the other. That framing describes the competitive dynamic the reporting reveals, not a statement either company has made. No on-record quotations from CXMT, YMTC or named company representatives appear in the available reporting on this story.
Competitive Comparison: China’s Memory Ambitions vs Global Leaders
Set against Samsung, SK hynix and Micron, CXMT and YMTC are still playing catch-up on scale, process node maturity and, critically, HBM, the segment analysts at Counterpoint Research point to as commanding the steepest prices and the tightest allocation. Nothing in the current reporting suggests either Chinese company is close to shipping HBM at competitive volumes. Their near-term opportunity is narrower: standard DRAM and NAND for domestic Chinese demand, not the advanced HBM stacks feeding Nvidia-class AI accelerators.
That narrower ambition is still meaningful. Standard DRAM and NAND make up the bulk of memory demand in phones, laptops and mainstream servers, categories where Chinese OEMs have every incentive to diversify away from foreign suppliers if a credible domestic option exists. CXMT entering NAND and YMTC entering DRAM, even at modest early volumes, gives Chinese device makers a hedge against the kind of supply shocks that have already pushed prices higher across the industry this year.
Predictions: Where the CXMT-YMTC Rivalry Goes From Here
- CXMT’s NAND effort likely stays in R&D and pilot-production territory through at least the first half of 2027, given that the current stage is described as a research line rather than a mass-production commitment.
- YMTC’s DRAM sampling will probably expand to more customers before any commercial announcement, following the same cautious qualification path CXMT itself took when it built out DRAM years ago.
- Expect both companies to target domestic Chinese OEMs first, with AI-storage and edge-device customers as the earliest adopters, rather than chasing export markets where Samsung, SK hynix and Micron already hold entrenched relationships.
- HBM will remain the one segment neither company challenges in the near term. Watch Samsung, SK hynix and Micron’s HBM roadmaps, not CXMT or YMTC, for signs of when the broader global shortage eases.
- If either company lands a named, on-record customer for its new product line in the next two quarters, treat that as the clearest signal yet that this shifts from R&D positioning to a real commercial threat.
Frequently Asked Questions
What is CXMT and why is it entering the NAND flash market?
ChangXin Memory Technologies (CXMT) is China’s leading DRAM producer. According to a report dated September 18, 2026, it’s preparing a research-and-development production line for NAND flash memory at a new plant in Beijing, a move that would expand it beyond its traditional DRAM business into territory currently dominated by YMTC and foreign suppliers like Samsung.
Is YMTC actually making DRAM chips now?
YMTC has sent low-power DRAM samples to customers, according to an earlier Reuters report cited in the September coverage. That’s a real engineering and qualification step, but it stops short of confirmed commercial production. No launch date or production capacity has been confirmed.
Does this mean memory chip prices will drop soon?
Not based on current reporting. Both companies are described as being at an early R&D or sampling stage, not ramping commercial volume. Any near-term price relief tied to the broader global memory shortage is more likely to come from Samsung, SK hynix and Micron’s own capacity expansion than from CXMT or YMTC.
Who are CXMT’s potential NAND customers?
The report mentions CXMT is in early discussions with potential customers, including an unnamed startup seeking NAND flash for storage used in AI systems and supercomputers. No named customers have been confirmed publicly.
How does this affect Samsung, SK hynix and Micron?
Not significantly in the short term. Qualifying new memory products with major global buyers takes years, and neither CXMT nor YMTC is reported to be close to competing in HBM, the highest-value memory segment. The bigger near-term effect is likely inside China’s own supply chain, where domestic OEMs gain additional sourcing options.
Is this officially described as a “turf war” by the companies involved?
No. That framing describes the competitive dynamic visible in the reporting, not a claim made by CXMT or YMTC themselves. No on-record quotations from either company or its representatives have been published in connection with this story.
What role does the HBM shortage play in this story?
Samsung, SK hynix and Micron dominate the advanced HBM market, and a global shortage in that segment has pushed prices higher across the memory industry through 2026. That shortage adds urgency to Beijing’s push for domestic memory alternatives, even though neither CXMT nor YMTC is reported to be targeting HBM production yet.
When will CXMT’s NAND flash chips actually ship?
No commercial launch date has been confirmed. The current stage described in reporting is a research-and-development production line, not a mass-production commitment, so a commercial shipping date remains unconfirmed.
