How to value Changxin?

How to value Changxin?

CXMT Technology (688825) is about to be listed today, becoming the largest IPO on the STAR Market. The IPO issue price is 8.66 yuan, conservatively estimated to correspond to a market value of about 520 billion yuan. But the market is clearly not planning to stay at this price.

Li Jiu, an analyst at Northeast Securities, valued CXMT from three independent perspectives in a research report, with conclusions converging in the range of 3.2–5.7 trillion yuan. On the same day, Nomura Securities initiated coverage with a “buy” rating, target price 116 yuan, implying an increase of 1239%, corresponding to a market value of about 7.76 trillion yuan—1.4 times the upper limit of Northeast Securities. The core of the divergence between the two institutions lies in their judgment of CXMT’s future market share ceiling—Northeast Securities assumes a base case of 17%, while Nomura bets on a larger market share and higher growth premium.

These valuations may not be exaggerated. CXMT is a unique presence in the A-share market to date: a pure DRAM IDM leader with complete design and manufacturing capabilities, currently enjoying a performance explosion in a “cycle reversal + share gain” phase. The company covers DDR4/5 and LPDDR4X/5/5X, with products entering the supply chains of Alibaba, Tencent, ByteDance, and mainstream smartphone makers. According to Omdia data, in Q4 2025, the company’s global market share is 7.67%, first in China and fourth globally. Benefiting from rising memory prices and high-end product volume, the company’s earnings elasticity is accelerating.

DRAM supply-demand gap persists, and CXMT enjoys “rising volume and price + domestic substitution” double windfall. The real question is not whether it is valuable, but which yardstick to use.

Perspective One: Relative Valuation by Market Share—Target Market Value About 3.49 Trillion Yuan

Logic: Since DRAM is a global unified market, the market value of listed overseas memory companies reflects the market’s pricing for “each percentage point of share.” Using the market values of comparable US listed companies, reverse-engineer “how much market value corresponds to each 1% of future global share,” then multiply by CXMT’s future share.

Operation: Micron and SanDisk’s NAND share is equal (both 13%), so Micron’s (a DRAM and NAND company) market value minus SanDisk (a pure NAND company) market value equals Micron’s DRAM business market value—1022 billion USD - 230.8 billion USD = 791.2 billion USD. Dividing by Micron’s 19.85% future DRAM share yields about 39.86 billion USD per 1% future DRAM share.

Conclusion: As a pure DRAM company, CXMT’s future share is 17% (currently about 8%), corresponding to about 677.676 billion USD in market value, equivalent to about 4.58 trillion yuan (exchange rate 6.77). Excluding about 24% minority shareholder interest, attributable market value is about 3.49 trillion yuan.

Back-test validation: SK Hynix’s reverse market value is about 9.44% higher than actual, Kioxia only 0.66% higher; results basically match actual market values.

Perspective Two: Earnings Split PE Valuation—Target Market Value 2.85–4.27 Trillion Yuan

The second method is more fundamental: without external anchors, directly project CXMT’s own profit. Memory manufacturers have highly standardized cost structures, with fixed costs mainly depreciation determined by capital expenditure, and variable costs changing linearly with shipments. As the prospectus does not disclose actual wafer capacity data, fixed asset original value is used as a proxy for capacity, multiplied by utilization and sales ratios to estimate shipments, then combined with ASP for revenue.

Logic: Split revenue (capacity × utilization × sales ratio × ASP) and costs (fixed depreciation + variable), predict net profit, then apply a PE multiple.

Key forecasts:

  • 2027 revenue 471.6 billion yuan, gross margin 86.96%, net profit 374.7 billion yuan (overall)
  • Exclude minority shareholders (assume 24%), attributable net profit about 284.8 billion yuan

Valuation: Northeast Securities estimates Micron and SK Hynix PE at 7.51x and 7.94x in 2027, but as CXMT’s share is rapidly rising (expected to reach 30%), it deserves a growth premium, assigning 10–15x PE; after adjusting for minority shareholders, the target market value is about 2.85–4.27 trillion yuan.

Perspective Three: Relative Valuation by Unit Capacity—Target Market Value 3.22–3.99 Trillion Yuan

Logic: Divide overseas memory company market value by monthly capacity to get "market value per 10,000 wafers/month," then multiply by CXMT’s capacity planning.

Reference: The main three manufacturers’ market value per 10,000 wafers/month is concentrated in the range of USD 15.8–19.8 billion—SK Hynix USD 16.045 billion, Micron USD 19.78 billion, Samsung USD 15.891 billion.

Conclusion: CXMT’s capacity in 2027 is 450,000 wafers/month, corresponding to market value:

  • Optimistic scenario (average for main three manufacturers: USD 17.2 billion/10,000 wafers): 525.18 billion yuan
  • Neutral scenario (including Taiwan manufacturers: USD 13.9 billion/10,000 wafers): 423.27 billion yuan

After adjusting for minority shareholders, corresponding to about 3.22–3.99 trillion yuan.

Summary of Three Methods: Converging at 3.2–5.7 Trillion Yuan

Northeast Securities points out that CXMT’s minority shareholder interest accounted for as high as 73.76% in 2025, much higher than Samsung, SK Hynix, Micron (all less than 1%), so valuation must exclude this portion.

Assuming minority shareholder interest remains at 24% in 2026 and 2027, the three methods conclude: after adjustment, reasonable valuation is 3.2–5.7 trillion yuan.

The three perspectives use different data and logic chains, but the final attributable range all falls around 3–4.3 trillion. This convergence itself is a signal: under the current share and capacity assumptions, this pricing level is highly self-consistent.

Nomura Securities: Target Price 116 Yuan, Implied Growth of 1239%

On July 27, Nomura Securities also initiated coverage of CXMT Technology, giving a more aggressive judgment.

The firm began coverage with a “buy” rating, target price 116 yuan, corresponding to about 20x PE—double the current valuation of Micron (about 10x), and more than double SK Hynix’s current valuation.

Based on the IPO issue price of 8.66 yuan, the 116 yuan target price means an implied increase of 1239.5%, corresponding to a market value of about 7.76 trillion yuan.

This far exceeds Northeast Securities’ upper limit of 5.7 trillion, with about 2 trillion difference. The essence is a different bet on two core variables: where CXMT’s share ceiling is, and what growth premium the market should give this company.

2026 Is Just the Starting Point! Domestic Substitution Plus AI Demand, Nomura Gives CXMT Dual Growth Premium

Nomura’s logic behind the 20x PE premium builds on three judgments.

First, structural supply tightening will persist for years. The key statement: "Global memory supply is unlikely to loosen in the coming years." Samsung, SK Hynix, and Micron have shifted capital expenditure largely to HBM and advanced process, suppressing new supply for generic DRAM. This means the generic DRAM market where CXMT sits will remain in short supply for quite a long time, rather than the traditional memory cycle of "rise two years, fall two years."

Second, CXMT’s share gain logic is “accelerated” rather than “linear". Nomura judges that as CXMT’s capacity continues to expand and the process migrates from fourth to fifth generation, its share gain in the global generic DRAM market will exceed market expectations. Current share is about 8%; Nomura’s model implies a much larger future space than Northeast Securities’ 17% upper assumption. Nomura’s implied future market share, reverse-engineered from the 7.76 trillion target market value, could be 25–30% or even higher.

Third, domestic substitution plus AI demand, dual growth premium. Nomura sees CXMT not only as a memory cycle play, but also as a "domestic substitution" theme. Chinese cloud and smartphone makers continue to increase willingness to buy local DRAM, providing CXMT incremental demand independent of the global cycle. Meanwhile, AI servers’ DRAM demand is growing exponentially, with per-server DRAM much higher than smartphones—nearly 80 times—this structural change will support rising ASP. With both logics combined, Nomura believes CXMT should enjoy a higher premium than overseas peers, not a discount.

That is, Nomura does not think 2026 is the peak, just the starting point.

On financial forecasts, Nomura predicts CXMT’s sales and attributable net profit to grow 63% and 74%, respectively. Drivers include capacity expansion from 270,000 wafers/month in 2025 to 450,000 in 2027, process migration lifting value per wafer, and DRAM average price rising given tight supply. Nomura’s earnings forecast is more aggressive in absolute numbers than Northeast Securities, and the 20x PE further amplifies the final valuation.

Backing this aggressive case is supply-demand data: global generic DRAM capacity calculations show a gap remains in 2027; Q1 2026 DRAM contract price soared 93–98% quarter-on-quarter, far above previous forecasts; CXMT’s Q1 2026 gross margin reached 79.16%, single quarter attributable net profit 24.762 billion yuan.

The height of the price cycle is real-time revising all model assumptions.

From Zero to Fourth Globally, CXMT Took Seven Years

In 2019, CXMT Technology (formerly RuiLi Integration) launched the first Chinese mainland self-developed mass-produced 8Gb DDR4, marking the breakthrough from nothing to something for domestic DRAM.

Seven years later, it is China’s No.1, and fourth globally as a DRAM maker. According to Omdia, in Q4 2025 CXMT’s global market share reaches 7.67%.

In terms of products, CXMT now covers DDR4/5 and LPDDR4X/5/5X generations, and will stop its own DDR4 production by end of 2024, fully shifting capacity to DDR5 and LPDDR5/5X and other high-value products. Its customers include Alibaba, Tencent, ByteDance, and mainstream smartphone supply chains.

Capacity-wise, the company has three 12-inch wafer fabs—two in Hefei and one in Beijing. Northeast Securities expects capacity to expand from 270,000 wafers/month in 2025 to 450,000/month in 2027, global share rising from 14% to 17%.

Financial Turning Point: Shocking Profit Elasticity in Pricing Cycle

CXMT’s financial path is the typical script of a heavy asset memory factory—fixed costs brought forward in the down cycle lead to losses, rapid profit release in the up cycle.

Key nodes:

  • 2025: Attributable net profit turns from a loss of 16.34 billion yuan to a gain of 1.875 billion yuan; gross margin rises to 40.99%, essentially on par with Samsung (39.38%)
  • 2026 Q1: Single quarter revenue 50.8 billion yuan (year-on-year +719%), gross margin 79.16%, attributable net profit 24.762 billion yuan
  • First half 2026: Management expects revenue of 110–120 billion yuan, attributable net profit 50–57 billion yuan

The core driver of profit boom is price. According to TrendForce’s latest survey in June 2026, Q1 2026 generic DRAM contract price soared by about 93–98% quarter-on-quarter, far above previous double-digit forecasts.

Northeast Securities also listed four main risks:

  1. Demand below expectations: Slowdown in AI server construction or weak recovery in consumer electronics.
  2. Price cycle downturn: In 2022–2023, prices fell by as much as 50% from previous cycles.
  3. Capacity and technology iteration lagging expectations: If fifth-gen process platform R&D is delayed, volume and price realization will be affected.
  4. International trade friction and supply chain restrictions: Escalating geopolitics could increase industry chain instability.

 

Risk Warning and DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice and does not consider individual users’ special investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Investment based on this article is at your own risk.