A memory maker can win a shipment ranking and still lose the money race. Yangtze Memory Technologies (长江存储, YMTC) overtook Japan's Kioxia in NAND flash shipments this year. By revenue it remains behind Kioxia and Micron. That gap between bits and dollars is the most useful thing about the company's 2026.
Bits are not dollars
Counterpoint Research's second-quarter 2026 NAND tracker, released in August, put YMTC at 14 percent of global NAND bit shipments — third in the world behind Samsung at 25 percent and SK hynix at 22 percent, and just ahead of Kioxia and Micron at about 13 percent each. YMTC's shipments rose 22 percent year on year and 5 percent from the first quarter.
Then comes the qualification. By revenue, YMTC was still fifth. It sells a mix weighted toward consumer devices — notebooks and phones — rather than the expensive enterprise solid-state drives that data centres buy. Enterprise SSDs accounted for 48 percent of all NAND bits shipped globally in the second quarter, nearly double the 26 percent a year earlier, according to the same research. A company can ship a great many bits and still sit on the cheaper side of that split.
This distinction is often blurred in headlines about Chinese memory catching up. It should not be. Shipment share measures how much silicon a company can move; revenue share measures what customers will pay for it.
What changed in two years
YMTC was founded in Wuhan in July 2016 and is widely described as the Chinese mainland's only integrated device maker for 3D NAND. Its climb was not a single breakthrough. It moved from 32 layers to 64, then to 128 and 232, and in 2026 it began mass production of 267-layer chips built on its proprietary Xtacking (晶栈) 4.0 architecture, with 1Tb (terabit) TLC capacity per die, according to EE Times. It is developing products above 300 layers.
The business result arrived with the technology. China Daily reported first-quarter 2026 revenue above RMB 20 billion (≈US$2.8 billion / HK$22.2 billion), roughly double the year before. On 27 March the company unveiled three storage products aimed at high-performance, high-security and high-capacity AI workloads. In August, a disclosure on the China Securities Regulatory Commission website showed it had completed the pre-listing coaching review for a Shanghai STAR Market (科创板) IPO, with CITIC Securities and China Securities as advisers.
Note the stage carefully: coaching completed is not a filed prospectus, and a prospectus is not a listing. YMTC is preparing to raise money, not raising it yet.
The factory nobody has ordered yet
In early November, Reuters reported that YMTC plans to source as much as 70 percent of the equipment for a planned fourth Wuhan factory from domestic suppliers, up from just over half at its third plant. The planned fab would make its most advanced NAND — more than 300 stacked layers — mainly for data centres and AI, the report said.
Those details come from two people familiar with the matter who spoke anonymously. Reuters noted that YMTC has held only preliminary supplier discussions, has placed no formal orders, and expects construction to begin in 2027. Treat this as a reported plan, not a committed project. An anonymous sourcing report is evidence of intent, not a purchase order.
The same report cited Omdia data showing YMTC's NAND share in China rose to 38.5 percent in the second quarter from 28 percent in 2025, overtaking Samsung domestically. A separate Reuters report in April had described YMTC planning two additional Wuhan fabs.
Why AI changed the memory game
For years, NAND was a commoditised business: more bits, lower prices, thin returns. AI has altered the customer at the other end of the wire.
Training a model is compute-heavy. Running one — inference — is increasingly storage-heavy, because the system must fetch large datasets and key-value caches quickly and repeatedly. That is why enterprise SSDs have become the fastest-growing part of NAND demand, and why YMTC's consumer-heavy mix matters so much. The company has said it plans to raise the enterprise share of its mix in the second half of 2026. Saying so is easier than doing it: enterprise customers validate storage for far longer than phone makers do.
There is also a constraint that no amount of demand removes. Washington added YMTC to its Entity List in December 2022, limiting access to certain advanced tools. Reuters reported that the company still relies on overseas suppliers for lithography and inspection equipment even as domestic makers win more of its cleaning and polishing orders. More stacked layers require more precise etching and deposition. That is the part of the roadmap the 70 percent target is really about.
Honest limitations
- Shipment share and revenue share come from different framings. Counterpoint's bit-shipment ranking is well documented in the sources consulted; the revenue ranking is reported by EE Times without a full table.
- The fourth-factory story is anonymously sourced. Equipment targets, layer count and timing come from Reuters' unnamed sources, not a YMTC announcement. YMTC did not comment.
- The Q1 revenue figure is a media report. The RMB 20 billion (≈US$2.97 billion) number comes from China Daily's summary, not from a company filing I reviewed directly.
- Xtacking's technical details are company-described. Layer counts and per-die capacity are attributed to YMTC via trade press, not independent teardowns.
- No IPO timetable is assumed. Coaching completion does not establish when, or whether, the listing will price.
What readers can do now
- When you read "world's third-largest NAND maker," check the metric. Shipments, revenue and capacity are three different leaderboards, and YMTC sits in different positions on each.
- For anyone buying storage, watch the enterprise mix. A supplier that wins consumer bit share is not yet a substitute for an enterprise SSD vendor.
- Treat reported factory plans as staged claims. Preliminary discussions, formal orders and construction are separate events — and Chinese self-media often collapses all three into one headline.
