In early November, news emerged that TCL Zhonghuan was planning to acquire a minority stake in its controlling subsidiary, Zhonghuan Leading. This timing coincides with the semiconductor silicon wafer industry standing at the starting point of a new cycle.
According to SEMI data, global silicon wafer shipment area reached 12.973 billion square inches in 2025, a year-on-year increase of 5.8%, returning to a growth trajectory.
Unlike previous cycles driven by consumer electronics generational shifts or automotive electronics penetration, this round is fueled by AI, and its transmission path is broader than ever before. AI's pull on semiconductors extends beyond computing chips and HBM. For data centers to truly land, power management ICs are needed for power conversion, power devices for electrical energy transformation, analog devices for signal processing and sensing, and optoelectronic devices for data transmission. Demand is expanding from the two core segments of computing and storage to broader semiconductor categories including power, analog, optoelectronics, and sensors.
These categories ultimately converge on one thing: silicon wafers. Silicon wafers account for approximately 30% of wafer manufacturing material costs and almost determine the yield ceiling of the entire production line. While demand spreads across all categories, the expansion cycle and certification barriers on the supply side mean capacity cannot keep pace synchronously. This "foundation" segment is undergoing a repricing driven by changes in demand structure. The question worth asking is: when demand shifts from single points to full-industry expansion, who has the capability to absorb it?
AI-Driven Full-Industry Demand Expansion
1.1 Transmission from Core Segments to All Categories
The pull of AI on silicon wafers can be seen through a set of data. According to SUMCO's estimates, a single AI server requires approximately 3.8 times more 12-inch silicon wafers than a general-purpose server; for equivalent storage capacity, HBM consumes three times more silicon wafers than mainstream DRAM. In 2026, monthly demand for advanced process silicon wafers for AI-related applications is expected to exceed 1 million pieces, accounting for over 10% of global total 12-inch demand.
Demand does not fall only on computing and storage but follows a longer transmission chain: "Building and operating data centers requires supporting components such as analog devices, power chips, and power management, while subsequent data transmission further drives optoelectronic devices." SMIC stated: "Taking a rack equipped with 72 GPUs as an example, approximately 16,000 units are required just for power supply and power management devices." From computing chips to storage, from server power supplies to power devices, from signal chains to optical interconnects, AI is spreading demand across all semiconductor categories.
Downstream prosperity also shows a multi-point bloom pattern: storage is driven by generative AI and HBM, logic by AI infrastructure investment, analog benefits from automotive intelligence and industrial control recovery, power absorbs demand from data centers, energy storage, and robotics, while optoelectronics and sensors follow multi-camera upgrades and ADAS adoption. Prosperity cycles of different terminals are not entirely synchronized, forming a "demand relay," which reduces the impact of volatility in a single track on silicon wafer companies and broadens the overall demand landscape.
1.2 Expansion Cycles and Overseas Restraint
The mismatch between the speed of demand expansion and the rhythm of supply release is also a cause of the gap. Silicon wafer expansion cycles typically require 18–24 months. According to CITIC Securities research reports, among the top five overseas silicon wafer manufacturers, only GlobalWafers has significant expansion plans. Considering construction and validation cycles, incremental supply from overseas will remain limited before 2028. Even under optimistic forecasts, overseas 12-inch heavily doped silicon wafer capacity is projected at 750,000 pieces/month in 2028, meaning industry supply will remain tight.
Global 12-inch silicon wafer monthly demand in 2026 is approximately 11 million pieces, with effective supply below 10 million pieces; demand may rise to 13 million pieces/month in 2027. The reaction speed of the supply side determines how long this gap persists. The restraint in expansion by leading overseas manufacturers, combined with entry barriers formed by certification cycles, results in the release of new effective capacity lagging behind the pace of demand expansion.
1.3 Top Concentration Leading and Window for Domestic Share Increase
Looking at the global landscape, the current silicon wafer market still exhibits clear top-concentration characteristics. CITIC Securities estimates: Shin-Etsu, SUMCO, GlobalWafers, Siltronic, and SK Siltron together hold a 76% market share. Changes are occurring, with domestic participation increasing. The localization rate of 12-inch silicon wafers in mainland China was approximately 15%-20% in 2025 and is expected to rise to 25%-30% in 2026.
Growing demand, customers' desire to increase local sourcing, and limited overseas increments combine to create a window for domestic manufacturers to increase their share. However, the speed at which this window opens depends on certification progress, not just the determination to expand capacity. Customer verification for silicon wafers typically takes one to two years; customers invest time, engineering resources, and costs to certify a supplier and will not indefinitely add similar suppliers.
A Decade of Advancement by an Established Material Manufacturer
Industry-level data outlines a window of comprehensive demand expansion, but the pace of supply-side expansion and certification cannot keep up synchronously. Who can truly capture this wave of demand depends on a company's accumulated depth in mass production, certification, and customer collaboration.
Zooming in from the industry to the enterprise level, Zhonghuan Leading Semiconductor Technology Co., Ltd. (hereinafter referred to as "Zhonghuan Leading") is a sample worth dissecting. Its path—from independently building 12-inch production lines, to filling gaps in memory and high-end logic capabilities through M&A integration, to achieving quarterly operational profitability—corresponds step-by-step to the difficult thresholds in the silicon wafer industry. More critically, what has settled in is not just capacity, but a product system covering all sizes, dual processes, and multiple applications, along with matching customer certification foundations and mass production management capabilities as leading advantages.
Technical accumulation is the deep foundation of this established factory. Zhonghuan Leading's predecessor traces back to the Tianjin Semiconductor Materials Factory established in 1958, one of the earliest domestic enterprises engaged in semiconductor material production. However, for a considerable period, its technical accumulation remained stuck at products of 8 inches and below. The real turning point occurred around 2017. That was an industry trough. Opportunities for the sale and consolidation of some overseas 12-inch silicon wafer assets emerged. The Zhonghuan team seriously evaluated acquisition paths, contacted several international peers, and pushed some projects to later stages, but none ultimately landed. The team shifted to independent construction, advancing the 12-inch factory in Yixing, Jiangsu. As they recalled, the capability for 12-inch silicon wafers was not bought; it was honed through continuous investment and engineering iterations.
Zhonghuan Leading joining the TCL system in 2020 and integrating the Xuzhou business in 2023 were two subsequent milestones. The Xuzhou business supplemented the originally weak capabilities in silicon wafers for memory and high-end logic, and also brought an internationalized technology and sales team.
Technical accumulation ultimately lands on product structure, creating scarcity in full-size coverage. Zhonghuan Leading is one of the few mainland enterprises simultaneously covering all sizes from 4 to 12 inches, both Czochralski and Float Zone processes, all doping types from lightly to heavily doped, and both polished wafers and epitaxial wafers. Relying on a collaborative network of multiple production bases and product lines, they can flexibly allocate capacity between different product lines according to market supply and demand changes. This complete full-category matrix is a highly scarce comprehensive foundation in mainland China.
The width of the product matrix also supports its advancement in overseas markets. In the first three quarters of 2026, Zhonghuan Leading's overseas revenue reached CNY 1.14 billion, a year-on-year increase of 19%, accounting for approximately 23% of total revenue.
Achievements on the customer side are the most direct verification of these accumulated capabilities. It is understood that Zhonghuan Leading has firmly established itself as a key supplier of localized silicon wafers for leading domestic wafer manufacturers and has entered the supply chain systems of top global wafer fabs. In the power device field, it dominates the high-voltage, high-resistance market through its Float Zone process, serving as a key supplier for leading power device manufacturers; in the CIS field, leveraging control over epitaxial layer resistivity and thickness, it covers applications from consumer-grade to automotive-grade; silicon wafers for memory and logic consistently rank in the domestic first tier.
These accumulations ultimately reflect in operational data. In Q3 2026, Zhonghuan Leading's revenue was CNY 1.86 billion, a year-on-year increase of 23%, with net profit of CNY 0.08 billion; in the first three quarters, revenue was CNY 4.90 billion, a year-on-year increase of 15%, with shipments of 1,101 million square inches, a year-on-year increase of 21%, of which 12-inch grew by 24% and 8-inch and below grew by 18%.
Capacity and Capital Supplementation
First, look at the coordinates. According to disclosures, as of the end of Q3 2026, Zhonghuan Leading ranked first domestically and sixth globally in full-size shipment area and operating revenue. Supporting this position is primarily capacity. Its Inner Mongolia base handles crystal supporting facilities, the Tianjin plant focuses on power and 8-inch products, Yixing and Xuzhou are the main 12-inch manufacturing bases, and the Shenzhen 12-inch project is under construction. According to company plans, by the end of 2026, the combined 12-inch capacity of the Yixing and Xuzhou sites will reach 1 million pieces/month; upon full operation, 12-inch capacity will be 2.1 million pieces/month and 8-inch capacity 1.4 million pieces/month.
Beyond capacity, we must see if the supporting system can keep up. It is understood that the localization rate of equipment and raw materials for Zhonghuan Leading is steadily improving, with the domestic supply ratio of polysilicon raw materials now increased to over 60%; R&D investment as a percentage of revenue is expected to continue increasing in 2026, with much of the R&D work conducted jointly with domestic equipment and material suppliers.
Building the supporting system requires capital cooperation. Zhonghuan Leading is a controlling subsidiary of TCL Zhonghuan, which currently holds 32.72%. TCL Zhonghuan is planning to issue shares to purchase minority stakes in Zhonghuan Leading and raise matching funds for Zhonghuan Leading's project construction. After the transaction is completed, the shareholding ratio will significantly increase. This move will help Zhonghuan Leading better conduct R&D investment and talent incentives, supporting the construction and operation of Zhonghuan Leading's factories. The formation and implementation of raised funds also require gradual acceptance and recognition from the market.
Silicon wafers are a long-cycle, heavy-asset industry where every expansion investment must be decided two to three years in advance. Stable capital and incentive mechanisms indeed affect the rhythm. However, equity concentration solves resource organization issues but does not change industry cycles or the essence of competition. What this cycle truly tests is not who expands faster, but who can continuously accumulate in certification, mass production consistency, and customer collaboration.
In Conclusion
Semiconductor silicon wafers, ultimately, involve racing against time. Expansion, certification, and ramp-up each take years. The demand expansion brought by AI is real, and the inability of supply to keep up is also real. However, the gap does not automatically become orders for any specific company; it must first pass through certification, through ramp-up, and through repeated delivery verifications before landing.
Whether these steps can be completed depends on whether resources can keep up. Silicon wafers are a heavy-asset, long-cycle business requiring continuous investment at every step. Resources are the prerequisite; without resources, capability accumulation is out of the question. At the same time, this is also a resource allocation event in the development of the domestic semiconductor industry. High-quality companies that determine yield ceilings deserve resource support matched to their capabilities.
Companies with solid accumulation often go further once resource allocation is in place. Zhonghuan Leading reaching today's position verifies exactly this: the threshold in the silicon wafer industry has never been about capacity numbers, but lies in those links that require years to build and cannot be directly accelerated by capital. In this round of AI-driven localization process, Zhonghuan Leading has already stood in a position capable of contributing more strength.