Nomura initiates coverage of ChangXin Memory Technologies with a Buy rating and a target price of CNY116
AI summary card
Nomura initiates coverage of ChangXin Memory Technologies with a Buy rating and a target price of CNY116
The report believes that AI and agentic AI will drive significant growth in global memory demand, while supply expansion is constrained by capacity, equipment, materials, and talent; CXMT is expected to increase its global DRAM share through capacity expansion, process migration, and domestic substitution.
- The target price of CNY116 is based on 20x 2028F EPS of CNY5.8, corresponding to an initiation Buy rating.
- The report forecasts CXMT's 2026-28F revenue and attributable net profit CAGR at 63% and 74%, respectively.
- Nomura expects agentic AI to drive more than 7x growth in global memory usage during 2026-30F, implying bit CAGR of over 60%.
- The report expects CXMT's global DRAM market share to rise from about 10% currently to about 18% by end-2028F.
- Major downside risks include weakening end demand, intensified domestic competition, insufficient supply of components such as ICs and CPUs, and escalating US-China geopolitical tensions.
Report interpretation
Overview
This is Nomura's initiation coverage company research on ChangXin Memory Technologies. The report positions CXMT as the largest DRAM manufacturer in China and the fourth largest globally. Its core judgment is that tight global memory supply-demand conditions will persist, and that demand from AI inference, RAG, agentic AI, HBM, and SSD will together create a memory supercycle, giving CXMT room to gain share against the backdrop of China's low self-sufficiency rate.
Core views
The report's core views include: first, agentic AI will significantly amplify demand for KV cache, context integration, multi-turn tool calling, and persistent storage; even after factoring in a 4x storage-efficiency discount, memory demand could still grow more than 7x during 2026-30F. Second, global DRAM supply bit CAGR is about 30-40%, below demand growth of more than 60%, and tight supply-demand conditions are favorable for CXMT. Third, CXMT can gain share through capacity expansion, migration from 16-17nm to 14-15nm, advancement in HBM3 capability, and demand from local customers. Fourth, the MATCH Act, Entity List, or other US sanctions could affect overseas equipment and high-end photoresist supply, making them key monitoring variables for the investment thesis.
Analysis framework
The report uses an initiation coverage framework, combining top-down AI memory demand estimates, global DRAM supply-demand judgment, China's domestic substitution opportunity, CXMT's capacity and process roadmap, earnings forecasts, relative valuation, and regulatory risks. Valuation uses 2028F EPS and a target P/E multiple, while industry analysis focuses on comparisons of demand bit CAGR, supply bit CAGR, capacity expansion pace, and inventory pressure.
Methodology notes
Target P/E valuation
The target price of CNY116 is based on 20x 2028F EPS of CNY5.8; the report believes that, due to market-share gains and higher valuation multiples in the China market, CXMT may trade at about 2x Micron's roughly 10x valuation over the past five years.
DRAM bit supply-demand gap
The report assumes memory demand CAGR of over 60% during 2026-30F, while industry supply CAGR is about 30-40%, with supply expansion constrained by physical limitations such as cleanrooms, equipment, materials, and talent.
Process migration and capacity expansion
The report analyzes CXMT's current 1x-1y nodes, DDR5/DDR4 yields, progress toward 1z-1a nodes, HBM3 samples, and the wafer-capacity expansion path during 2026-28F.
Export controls and sanctions risk
The report incorporates the MATCH Act, Entity List, overseas equipment, DUV, EUV, high-end photoresist, and localization rate into its risk assessment, with particular focus on Shanghai capacity expansion and sub-15nm process advancement.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- ChangXin Memory Technologies / 688825.SS / 688825 CHCore coverage target
- Strengths
- The largest DRAM manufacturer in China, with room for domestic substitution; the report expects capacity expansion, node migration, and ASP improvement to drive high growth; AI memory demand and China's low self-sufficiency rate support market-share gains.
- Weaknesses
- There is still a gap in process technology versus leading overseas manufacturers; the report estimates mainstream technology at about 16-17nm; high-end photoresist and some equipment depend on overseas supply; expansion and yields are sensitive to supply-chain stability.
- Comparison
- The report uses Micron as the main overseas comparable and believes CXMT may enjoy a higher target P/E due to market-share gains and higher valuation multiples in the China market.
- Risks
- The MATCH Act, Entity List, or other US sanctions; intensified domestic competition; weakening end demand; insufficient supply of components such as ICs and CPUs; escalating US-China geopolitical tensions.
- Micron / MU USValuation comparable
- Strengths
- A major global DRAM manufacturer, with historical valuation used as a reference for CXMT's target multiple.
- Weaknesses
- The report does not assign a rating to Micron and uses it only as a valuation reference.
- Comparison
- Micron traded at about 10x P/E over the past five years, while Nomura uses 20x 2028F P/E for CXMT.
- Risks
- Not rated; the report does not elaborate on company-level risks.
- YMTCDomestic competition reference
- Strengths
- An unlisted domestic semiconductor memory-related company, representing potential local competition.
- Weaknesses
- The report does not provide detailed financial or valuation analysis.
- Comparison
- Listed as one of the downside risks that could intensify domestic competition and affect CXMT.
- Risks
- If domestic competition intensifies, it could compress CXMT's market share or profitability.
- TOK / 4186 JP, JSR, Shin-etsu / 4063 JPRelevant parties in the key materials supply chain
- Strengths
- Major suppliers of high-end photoresist, with high technology and materials barriers.
- Weaknesses
- For CXMT, they constitute an external supply dependency.
- Comparison
- The report points out that China lacks alternative suppliers in high-end photoresist.
- Risks
- If CXMT is prohibited from importing Japanese high-end photoresist, migration to sub-15nm nodes could be hindered.
- GigaDeviceRelevant party in DRAM-on-logic WoW collaboration
- Strengths
- The report mentions that CXMT and GigaDevice cooperate in DRAM-on-logic production and design to meet China's edge AI and physical AI demand.
- Weaknesses
- The report does not provide the financial contribution of this collaboration or certainty of mass production.
- Comparison
- This direction is used to support edge AI scenarios such as smart cockpits, premium smartphones/PCs, robots, and physical AI.
- Risks
- Momentum in WoW technology is expected to strengthen from late 2027F, and there is still uncertainty around the pace of mass production and customer adoption.
Key data
- RatingBuyNomura's initiation rating.
- Target priceCNY116.00Based on 20x 2028F EPS of CNY5.8.
- IPO priceCNY8.66The starting price shown in the report table.
- Implied upside+1239.5%Calculated based on the target price and IPO price.
- 2026-28F revenue CAGR63%Supported by capacity expansion, process migration, and ASP increases.
- 2026-28F attributable net profit CAGR74%The report's core forecast for CXMT's earnings growth.
- Global DRAM market shareabout 10% currently to about 18% by end-2028FThe report expects CXMT's market share to keep rising.
- 2026-30F agentic AI memory usagemore than 7x growthAlready factors in a 4x discount from memory-efficiency technologies.
- 2026-30F memory demand CAGRmore than 60%The report's AI-driven demand assumption under a scenario of no growth in non-AI applications.
- Industry supply bit CAGRabout 30-40%Below the report's assumed demand growth rate.
- CXMT bit CAGRabout 40-45%Based on capacity forecasts and the assumption of process upgrades about every two years.
- CXMT capacityabout 280kwpm by end-2025, about 350kwpm by end-2026F, about 550kwpm by end-2028FThe report expects about 100kwpm of new capacity in each of 2027F and 2028F.
- IPO proceedsCNY57.9bn, up to CNY66.6bn after greenshoeEquivalent to about USD8.55bn, or up to about USD9.83bn after greenshoe.
- Domestic equipment localization rateabout 40%Nomura estimates that if forced to raise the localization rate quickly, yields could be affected.
- China DRAM self-sufficiency rateabout 30% in 2025China accounts for about 25% of the global DRAM market, but the revenue share produced by Chinese domestic manufacturers is about 10%.
Impact & implications
The report's investment implication is positive overall: in an AI-driven memory demand supercycle, CXMT may benefit from rising China's DRAM self-sufficiency, procurement support from domestic customers, capacity expansion, and process migration. However, the investment thesis is highly sensitive to external constraints, especially whether overseas equipment, materials, and high-end photoresist supply are affected by sanctions; if Shanghai expansion is delayed or progress at sub-15nm nodes is hindered, valuation and earnings forecasts could come under pressure.
Risks
- Weakening end demand, causing DRAM prices, shipments, or customer procurement to come in below expectations.
- Intensified competition from domestic peers such as YMTC, affecting CXMT's market share and profitability.
- Insufficient supply of key components such as ICs and CPUs, limiting downstream system shipments and realization of memory demand.
- Escalating US-China geopolitical tensions, affecting expectations for equipment, materials, customers, or capital markets.
- The MATCH Act, Entity List, or other US sanctions restricting CXMT's access to overseas equipment and materials.
- Inability to replace key materials such as Japanese high-end photoresist, potentially hindering progress at sub-15nm nodes.
- If Shanghai capacity expansion is delayed due to sanctions or supply-chain impacts, 2028F capacity and earnings forecasts may be affected.
- Agentic AI adoption lower than expected, or greater-than-assumed savings from memory-efficiency technologies, could weaken demand growth.
What to watch
- Whether CXMT's 2026-28F capacity materializes along the path from 350kwpm to 550kwpm.
- Progress in R&D, yields, and customer validation from 16-17nm to 14-15nm, 10-15nm, and HBM3/HBM3e.
- The actual scope of implementation of the MATCH Act, Entity List, and export controls by the US and its allies.
- Whether supply of DUV equipment, non-EUV routes, overseas equipment services, and high-end photoresist remains stable.
- The pace of improvement in China's DRAM self-sufficiency rate and the degree of procurement support from domestic customers.
- Changes in global DRAM prices, LTA contracts, inventory weeks, and industry capex-to-sales ratios.
- The actual evolution of agentic AI task volume, AI bot traffic, KV cache demand, and memory-efficiency technologies.
- The rollout pace of DRAM-on-logic WoW in smart cockpits, premium smartphones/PCs, robots, and physical AI scenarios.