Meritz Securities' Interpretation of ChangXin Memory's Q3 Results

‌CXMT 3Q26 Preliminary Results: Slowing Growth Rate and Sharp Stock Price Drop]
Today, the stock price of ChangXin Memory (CXMT), a Chinese DRAM company, fell by -7.8%.
The decline in stock price was significant compared to overseas memory companies, which appears to be due to the QoQ growth rate in the preliminary Q3 results announced on the 17th slowing down more significantly than expected.
(Based on 3Q preliminary results)
- Revenue: +15.3%~20.3% QoQ
- Net Profit: +1.8%~4.7% QoQ
- Net Profit Margin: 2Q 53.1% -> 3Q 47%
The issue is the net profit growth rate. The Q3 growth rate was limited to low single-digit growth compared to the previous quarter, and the net profit margin also dropped by 6%p. This stands in stark contrast to the 2nd quarter, when both revenue and net profit increased by 96% and 117% respectively compared to the previous quarter.
Local interpretations suggest the causes of the profitability slowdown are 1) differences in yield and cost structure and 2) differences in product mix.
Until the 2nd quarter, the increase in product prices was larger than costs, masking the issue, but as the price increase rate slowed, cost issues including yield and depreciation were revealed. Furthermore, regarding product structure, while overseas companies produce high-value-added products, China focuses on commodity products and has not yet been able to ship HBM, meaning AI benefits are more limited compared to overseas competitors.
The final business report will be released on 10/31. I will provide further updates then.
For China's ChangXin or Yangtze Memory, there are clear limitations unless US regulations are lifted.
US investment in AI data centers is at least 5 times greater than China's.
China can even reach those figures when including government support.
There is almost no possibility of ChangXin or Yangtze products entering the US market; even if the US wanted to import them,
demand outstrips supply.
If they were to lift the restrictions, the US would face a situation it cannot handle, and doing so would certainly cement them as the number one incompetent ruler in history.
If they keep pushing for more capacity with the small amount of expansion planned through 2027, Samsung Electronics' shortage next year is almost certain.
For memory semiconductor investors, do not overreact to obvious news of Chinese capacity expansion; what matters is how much more the US invests in AI data centers.