Gf Securities has released a research report stating that China is accelerating the domestic substitution of semiconductors through multi-dimensional financial and policy support. The pace of domestic storage and advanced process node substitution is quickening, leading to a new cycle of high growth in domestic cleanroom construction. Cleanroom engineering services are characterized by low error tolerance, strong customer loyalty, and stable supply of high-end services. The firm anticipates three major shifts in the domestic cleanroom engineering services market: 1. Growth in orders and revenue. 2. Improved gross margins driven by reduced competitive intensity. 3. Regional differentiation and decreased competition, with domestic cleanroom service providers holding significant first-mover advantages. Key recommendations include Bclean Co., Ltd. (601133.SH), a long-term supplier to ChangXin Memory Technologies and Yangtze Memory Technologies Corp., and Shenzhen SED Industry Co., Ltd. (000032.SZ), a leading domestic cleanroom service provider. The main points from Gf Securities are as follows.
With the escalation of technological restrictions, the substitution of domestic storage and advanced process nodes is accelerating, ushering in a new period of high growth for cleanroom construction in China. On April 2, 2026, bipartisan legislators in the U.S. Congress formally introduced the Multilateral Alignment of Technology Control Hardware (MATCH) Act, requiring allied nations to implement semiconductor equipment export controls on China matching those of the U.S. within 150 days. Since 2018, U.S. restrictions on China's semiconductor industry have progressively escalated from sanctions on individual companies to systematic containment involving equipment, EDA software, advanced process nodes, and investment. In response, China is accelerating domestic substitution through multi-faceted financial and policy support. A review shows that demand for semiconductor cleanroom construction in China has experienced three cycles of growth.
The first cycle occurred from 2016 to 2020, marked by the establishment of the "two memory" companies and Semiconductor Manufacturing International Corporation's (SMIC) continued capacity expansion, initiating a wave of factory construction for memory and advanced foundry processes "from zero to one." Yangtze Memory Technologies Corp. was established in 2016 and began a three-phase construction of the National Memory Base project with a total investment of $24 billion. ChangXin Memory Technologies was established in 2016 and announced a 72-billion-yuan investment in 2017 to build a 12-inch wafer fab.
The second cycle spanned from 2020 to 2023, characterized by high capital expenditure in local wafer fabrication following restrictions on advanced manufacturing. In 2020, SMIC's CAPEX surged significantly to 37.2 billion yuan, reaching 53.9 billion yuan by 2023, fueling the construction of multiple projects including SMIC Beijing, SMIC Shenzhen, and SMIC South.
The third cycle began in 2024, driven by AI-fueled demand for High Bandwidth Memory (HBM). A global shortage of memory chips has gradually emerged, leading to a new round of capacity expansion cycles in both domestic and international memory markets.
High-grade cleanroom engineering services feature low error tolerance, strong customer loyalty, and stable supply of high-end services. The acceleration and expansion of domestic substitution are expected to drive increased volume and profitability in high-grade cleanroom construction demand. As a core component of semiconductor industry CAPEX, cleanroom engineering possesses three key characteristics. First, low error tolerance: chip manufacturing imposes strict requirements on cleanliness, temperature and humidity control, Airborne Molecular Contamination (AMC) control, and micro-vibration control, leading to high owner demands for cleanroom engineering quality and speed. Second, strong customer loyalty: given the high engineering difficulty, owners typically select long-term partners with mature technology, extensive experience, and mutual understanding. Third, stable supply of high-end services: the core of cleanroom engineering lies in experienced professional engineers and accumulated on-site knowledge, which have long training cycles and slow capacity growth.
Amid the current domestic memory expansion cycle, Gf Securities expects three changes in the domestic cleanroom engineering services market. Firstly, order and revenue growth: cleanroom projects have short timelines and quick order turnover; the current memory chip expansion is expected to translate into revenue growth for cleanroom companies with about a one-year lag. Secondly, improved gross margins driven by reduced competitive intensity: with short-term supply difficult to scale up rapidly and a preference for domestic engineering service providers, the improved competitive landscape is expected to boost the overall gross margins of cleanroom firms. Thirdly, regional differentiation and reduced competition intensity, with domestic cleanroom service providers holding significant first-mover advantages: as overseas orders surge, Taiwanese companies with parent firms rich in overseas client resources may increasingly shift focus to higher-margin international projects. Consequently, a larger share of the current domestic memory expansion projects may be captured by companies like Shenzhen SED Industry Co., Ltd. and Bclean Co., Ltd.
Market estimates suggest that achieving complete domestic substitution would require over 200 billion yuan in additional cleanroom investment. The investment required per 10,000 wafers per month of capacity is approximately 3.7 billion yuan for NAND, 5.1 billion yuan for DRAM (including HBM), and 40 billion yuan for advanced process nodes below 7nm. The current supply gap is estimated at 690,000 wafers/month for NAND, 520,000 wafers/month for DRAM (including HBM), and 230,000 wafers/month for advanced logic chips. Calculating based on these supply gaps, achieving 100% domestic substitution would require a total investment of about 1,439.4 billion yuan: 254.6 billion yuan for NAND, 264.8 billion yuan for DRAM (including HBM), and 920 billion yuan for advanced logic chips. Cleanrooms account for roughly 15% of wafer fab investment, implying a potential market space of approximately 215.9 billion yuan for cleanrooms if full domestic substitution is realized.
Potential risks include investment levels falling short of expectations, gross margin improvements not meeting forecasts, and challenges in expanding overseas operations.
Comments