CXMT and YMTC Push Forward IPOs, Targeting AI Memory Market
nashnova research
China's two largest memory-chip makers — CXMT and YMTC — are advancing toward public listings simultaneously. CXMT has cleared its STAR Market review; YMTC has completed IPO coaching. The parallel push signals a shift from catch-up R&D to scaled-up expansion, driven by AI server demand.
Why are both companies racing to IPO at the same time?
CXMT (长鑫存储) has passed the Shanghai STAR Market listing review. YMTC (长江存储) has completed its IPO coaching and filing procedures.
This means → both firms are at the capital-market door almost simultaneously — China's domestic memory industry is entering a "seats at the table" phase.
In plain terms = they used to chase technology in the lab; now they're bringing scorecards to raise money for expansion.
How far has the technology come?
YMTC's flagship is its Xtacking architecture — fabricating the storage array and the control circuitry separately, then bonding them together for higher density and speed. It has completed 294-layer 3D NAND development, with mass-production yield reportedly above 90%.
CXMT is mass-producing 16 nm-class DDR5 and investing in next-generation DRAM and HBM — high-bandwidth memory, the large-capacity, high-speed memory designed to sit next to AI chips.
This means → on their respective tracks — NAND flash and DRAM — both have reached the "shipping in volume" stage, no longer just showing samples.
Where do they rank globally?
YMTC's global NAND market share rose to roughly 13% in Q1 2026, approaching SanDisk and Micron levels. Total global NAND revenue that quarter was about $46 billion.
CXMT held roughly 8% of global DRAM in the same period — up year-on-year, but still far behind Samsung, SK Hynix, and Micron.
In plain terms = in NAND, they have squeezed onto the main table; in DRAM, they are still catching up — DRAM's technical bar is inherently higher.
How is AI demand reshaping revenue?
Smartphone LPDDR products still account for over 60% of CXMT's revenue, but AI-server DDR5 and related products now exceed 30%, and the enterprise segment keeps growing.
YMTC's shipment growth is driven mainly by enterprise SSDs, data-center, and AI-related demand.
This means → both companies' growth engines are pivoting from consumer electronics to AI infrastructure — whoever captures AI-server orders faster will have the stronger IPO valuation story.
How far along is equipment localization?
Under ongoing U.S. export controls, CXMT's overall domestic-equipment ratio has risen to roughly 40%–50%. YMTC's Wuhan Phase 3 fab sources over 50% domestically, with etch and thin-film deposition reportedly above 60%.
Advanced lithography, metrology, and certain critical process tools remain dominated by overseas suppliers. Full localization faces a significant gap in the near term.
In plain terms = more than half the equipment is now homegrown, but the hardest steps — especially lithography — still cannot be replaced.
What is the biggest risk?
Memory chips are deeply cyclical. If AI-driven demand slows or capacity expansion tips the industry back into oversupply, prices will come under pressure.
The core test for both companies is the same one every global peer faces: staying profitable while sustaining heavy R&D spending.
This reflects the key variable the capital market will weigh at IPO — not today's market share, but whether these firms can survive the next cycle trough and keep investing in R&D.
市场有风险，内容仅供研究参考，不构成投资建议。