Silicon carbide (SiC) is a core material of third-generation compound semiconductors. With physical properties such as high breakdown electric field, high thermal conductivity, and wide bandgap, it breaks through the performance ceiling of traditional silicon-based power devices and is a core component for high-voltage and high-power scenarios such as new energy vehicles, AI computing power supplies, photovoltaic energy storage, and rail transit. In 2025, the industry experienced overcapacity and deep clearing through price wars. Starting from the fourth quarter of 2025, the industry ushers in a structural inflection point. Automotive-grade and AI power-grade high-quality substrates and devices enter a shortage cycle, while supply of low-end general-purpose products remains loose, showing obvious structural differentiation in the industry. The global market maintains a high compound growth rate in the long term, with new energy vehicles as the fundamental base and AI computing infrastructure becoming the second growth curve. The domestic industrial chain has achieved leapfrog breakthroughs in the substrate segment, leading globally in 8-inch substrate market share, while the epitaxy, device, and module segments continue to advance domestic substitution. However, automotive-grade certification, yield, cost, and patent barriers remain core shortcomings.
I. Industry Overview: Basic Principles of Silicon Carbide (SiC) and Industrial Chain Structure
1.1 Material Properties and Application Value of Silicon Carbide
Silicon carbide (SiC) is a compound semiconductor material composed of silicon and carbon, belonging to third-generation wide-bandgap semiconductors. Compared with traditional silicon (Si) materials, it has three core advantages: breakdown electric field strength is approximately 3 times that of silicon, supporting higher voltage levels; thermal conductivity is approximately 3 times that of silicon, significantly improving heat dissipation capability; and switching losses are lower with stronger high-temperature resistance. In high-voltage and high-power power conversion scenarios, SiC power devices can reduce system losses, shrink heat dissipation and volume, and improve energy conversion efficiency, making them the underlying material for energy transition and computing infrastructure.
Traditional silicon-based IGBTs (Insulated Gate Bipolar Transistors) have prominent shortcomings in losses and heat dissipation in 800V and above high-voltage platforms, while SiC MOSFETs (Silicon Carbide Metal-Oxide-Semiconductor Field-Effect Transistors) are naturally suited to 800V high-voltage platforms, thus becoming the core choice for new energy vehicle high-voltage platforms. In addition to power devices, SiC can also be used in emerging directions such as RF devices and advanced packaging heat dissipation substrates, opening up greater market space.
1.2 Industrial Chain Structure
The SiC industrial chain is divided into three major segments: upstream substrates and epitaxy; midstream chip manufacturing, devices, and power modules; and downstream applications. Substrates are the segment with the highest value and strongest technical barriers in the entire industrial chain, accounting for approximately 47% of total device cost, and are the core focus of industry competition.
1. Upstream: Substrates + Epitaxy
SiC substrates: divided into conductive type (power devices, new energy vehicles, energy storage, AI power supplies) and semi-insulating type (RF devices). The mainstream size is 6 inches, the industry is iterating toward 8 inches, and 12 inches has entered the sample verification stage. Larger sizes are the core path to cost reduction. The area of 8-inch substrates is 78% larger than 6-inch substrates, the number of die cut per wafer increases by approximately 1.8 to 2 times, edge loss is reduced, and per-wafer cost decreases by approximately 35%. The core difficulties in substrate preparation lie in crystal growth yield, defect control, and size scaling. Crystal growth furnaces, cutting, grinding, and polishing equipment are core capital expenditures.
Epitaxial wafers: Epitaxial layers are grown on substrates through MOCVD (Metal-Organic Chemical Vapor Deposition) equipment. Device performance is determined by epitaxial layer quality, and this is a key link connecting substrates and chip manufacturing.
2. Midstream: Chip Manufacturing, Devices, Power Modules
On SiC epitaxial wafers, semiconductor processes such as lithography, etching, and ion implantation are completed to prepare chips such as MOSFETs and SBDs (Schottky Barrier Diodes); these are then packaged into power modules and supplied directly to downstream end manufacturers. Overseas IDM (Integrated Device Manufacturer) companies (STMicroelectronics, Infineon, Wolfspeed) mostly cover the full chain from substrate to epitaxy to devices to modules; domestic enterprises more often adopt a division-of-labor model, with substrate companies, epitaxy plants, and device design and packaging companies developing independently, while some enterprises extend upward into upstream布局.
3. Downstream Applications
New energy vehicles are the main track (main inverter, onboard OBC, DC-DC, onboard charging piles); followed by photovoltaic inverters, energy storage converters, industrial power supplies, AI server high-voltage power supplies, rail transit, high-voltage grids, and fast-charging infrastructure; the frontier direction is advanced packaging heat dissipation substrates for AI computing chips.
II. Global and Domestic SiC Industry Status
2.1 Market Size and Growth Overview
The global SiC power device market continues to grow rapidly. In 2025, the global SiC power device market size is approximately RMB 40.691 billion. Institutions predict a compound growth rate of 20.5% from 2026 to 2032, with the market size reaching RMB 147.968 billion by 2032. Open Source Securities predicts that the global SiC market size will reach USD 12.4 billion by 2030, with AI infrastructure contributing nearly half of new demand, breaking the past growth structure that relied solely on new energy vehicles.
From the demand structure perspective, new energy vehicles currently account for approximately 61% of SiC device revenue and are the largest application market. The increase in penetration of 800V high-voltage platform models directly drives up SiC content per vehicle; AI server power supplies and photovoltaic-storage-charging are the second growth curve. In the past, the market generally defined SiC as a material for the new energy vehicle track. The current industry logic has upgraded to the three-driver model of "new energy vehicles + AI computing power + new energy storage."
2.2 Supply-Demand Pattern: Deep Clearing in 2025, Structural Shortage in 2026
From 2021 to 2024, large amounts of domestic capital poured into the SiC track, and a large number of 6-inch substrate projects were launched in concentrated fashion. Industry capacity expanded rapidly. Combined with periodic demand falling short of expectations in the new energy vehicle industry in 2025, the industry entered fierce price wars. Prices of low-end substrates and general-purpose devices fell sharply, many small and medium-sized manufacturers suffered losses, and the industry entered a capacity clearing cycle. Utilization of low-end capacity declined significantly, and some manufacturers exited the market.
Starting from the fourth quarter of 2025, the supply-demand pattern reversed, but this was not a comprehensive supply shortage, rather a structural shortage: (1) Supply of automotive-grade and AI power-grade high-quality conductive substrates, epitaxy, and MOSFET modules is tight. Capacity utilization of leading substrate manufacturers has recovered to above 90%, 6-inch substrate prices have bottomed out and rebounded, and 8-inch substrate prices have stopped falling and edged upward; overseas IDM manufacturers Infineon and STMicroelectronics, as well as domestic companies such as Yangjie Technology, have successively raised device quotations. Delivery times for SiC devices for AI server power supplies have lengthened to approximately 30 weeks, with prices up more than 15% compared with the end of 2025. (2) Supply of low-end substrates with low defect indicators that cannot pass automotive-grade certification and industrial general-purpose specification devices remains in excess, and prices continue to come under pressure.
The core result of this round of industry reshuffling: industry competition has shifted from simply competing on capacity scale to comprehensive capability competition in yield, defect control, automotive-grade certification, stable delivery, and cost control. Industry concentration continues to rise, and leading enterprises benefit.
2.3 Global Competitive Landscape
Traditional overseas giants have first-mover advantages and mainly adopt the IDM model: Wolfspeed (United States, formerly Cree), ROHM (Japan), STMicroelectronics (ST), Infineon (Infineon), and onsemi. Overseas manufacturers have deep accumulation in substrate patents, automotive-grade customer certification, device processes, and module packaging, and have long occupied the main share of the global power device market.
Domestic enterprises have achieved major breakthroughs in the substrate segment. In 2025, domestic enterprises achieved global leadership in the conductive substrate field. Tianyue Advanced's global market share of 8-inch SiC substrates exceeded 50%, and its global share of 6-inch conductive substrates ranked first. Tianyue Advanced and Tianke Heda together hold a conductive substrate market share exceeding that of Wolfspeed. China's SiC substrate supply accounts for approximately 50% of the global total and has entered the global first tier. Domestic epitaxy, device, and module segments are still in the catch-up stage. Domestic device manufacturers still lag behind overseas leaders in automotive-grade certification, reliability, and stable batch delivery capability. The top five manufacturers in the domestic power device market together occupy 70.4% market share, with the top three all being overseas enterprises, while domestic manufacturers still have relatively low market share.
Representative enterprises in each segment of the domestic industrial chain: (1) Substrates: Tianyue Advanced, Tianke Heda, Luxshare Technology; (2) Epitaxy: Tianyue Advanced, San'an Optoelectronics, Changfei Advanced; (3) Devices/Modules: StarPower Semiconductor, Times Electric, Silan Microelectronics, Yangjie Technology, Basic Semiconductor; (4) Upstream equipment: Jingsheng Mechanical & Electrical (crystal growth furnaces), Jingsheng Equipment, AMEC (MOCVD epitaxy equipment).
2.4 Domestic Policy Environment
As a key track of third-generation semiconductors and new quality productive forces, silicon carbide belongs to the strategic direction of domestic semiconductor autonomy and controllability. The national "14th Five-Year Plan" and "15th Five-Year Plan" continue to support the industrialization of third-generation semiconductor materials and devices, and many localities have established special industrial funds to provide subsidies for production line construction, R&D investment, and industrialization projects. At the same time, domestic industrial policies for new energy vehicles, photovoltaic energy storage, and computing infrastructure indirectly drive downstream SiC demand. Policy dividends accelerate the verification and introduction of domestic supply chains, but industry subsidies are gradually shifting from capacity subsidies to high-quality indicators such as R&D, yield improvement, and automotive-grade certification.
III. Downstream Demand Analysis: Three Major Driving Tracks, Growth Logic Restructured
3.1 New Energy Vehicles: Industry Fundamental Base, 800V Platforms Increase Per-Vehicle Value
New energy vehicles remain the largest source of SiC demand. Traditional 400V platforms are dominated by silicon-based IGBTs; 800V high-voltage platforms have advantages such as fast charging, weight reduction, and improved range, and SiC MOSFETs have become the mainstream solution. SiC devices are used in components such as the main inverter, onboard OBC, DC-DC, and onboard charger, and per-vehicle SiC value is significantly higher than that of 400V silicon-based solutions.
Global mainstream automakers such as Tesla, BYD, NIO, XPeng, Volkswagen, and Mercedes-Benz continue to introduce SiC solutions in high-end models. As 800V platform models sink into mid-range models, SiC penetration will continue to rise. In the short term, the new energy vehicle industry faces sales fluctuation pressure, but the medium- and long-term electrification trend is certain. Rising penetration of 800V platforms is a确定性 main line that continuously drives SiC device demand.
3.2 AI Data Center Power Supplies: Second Growth Curve, Reshaping the Industry Ceiling
AI large-model computing clusters require massive server power supplies, and the upgrading of AI server high-voltage power supply architectures creates substantial SiC demand. Traditional server power supplies are dominated by silicon-based devices. AI high-power, high-density power supplies impose higher requirements on losses, heat dissipation, and power density. SiC devices can significantly reduce power supply losses, reduce power supply volume, and lower heat dissipation pressure. Open Source Securities predicts that by 2030, AI infrastructure will contribute nearly half of SiC market demand, becoming the industry's most important incremental source.
Frontier direction: SiC as a high-thermal-conductivity substrate for advanced packaging CoWoS (Chip-on-Wafer-on-Substrate), used for heat dissipation of HBM (High Bandwidth Memory) supporting computing chips. If industrialization is realized, SiC will extend from a power device material to a functional material for advanced packaging, greatly increasing per-chip value and opening up an entirely new market space. However, this direction is still in the customer sample verification stage, and industrialization carries uncertainty.
3.3 Photovoltaic Energy Storage, Industry, and Rail Transit: Stable Increment
Global photovoltaic and energy storage installations continue to grow. Centralized inverters, string inverters, and energy storage PCS are gradually iterating from silicon-based solutions to SiC solutions, improving inverter power density and conversion efficiency. Industrial motors, high-voltage power supplies, rail transit, and high-voltage grid transmission are medium- and long-term stable demands characterized by long certification cycles, stable orders, and long life cycles, with extremely high requirements for device reliability.
IV. Technology Iteration Path and Medium- and Long-Term Development Forecast
4.1 Short Term (2026-2028): Large-Scale 8-Inch Implementation, Yield Improvement + Cost Reduction, Accelerated Automotive-Grade Certification
In the next 2-3 years, the core industry main line will be optimization of 6-inch capacity and large-scale ramp-up of 8-inch substrate, epitaxy, and device production lines. 8 inches is the industry's mainstream size for the next 3-5 years. Leading domestic substrate manufacturers will continue to expand 8-inch production lines to reduce per-wafer cost.
The core contradiction in the industry has shifted from "whether there is capacity" to "yield improvement, defect control, and landing of automotive-grade customer designations." A large number of domestic device enterprises are accelerating AEC-Q100 automotive-grade certification for automakers, energy storage, and AI power supply customers. As yield continues to improve, unit costs of SiC devices continue to decline, driving SiC penetration from high-end models to mid-range models and from high-end power supplies to general industrial power supplies.
Supply side: after capacity clearing in 2025, the pace of new capacity expansion slows, and the release pace of new capacity is lower than demand growth. Automotive-grade high-quality products maintain a tight pattern, and the industry enters a cycle of rising volume and prices; low-end substrates and general-purpose devices continue to face competition, and small and medium-sized manufacturers lacking yield and customer certification continue to be cleared out, further increasing industry concentration.
4.2 Medium Term (2029-2032): 8 Inches Become the Absolute Mainstream, 12-Inch Industrialization Begins, Domestic Substitution Enters Deep Water Zone
8-inch production line yields continue to optimize, costs fall significantly, and SiC penetration in new energy vehicles, energy storage, and AI power supplies rises substantially. The domestic substrate segment maintains global competitiveness; domestic substitution in epitaxy, device, and module segments accelerates, and domestic IDM manufacturers gradually enter the supply chains of overseas leading automakers and computing power supply manufacturers.
12-inch SiC substrates enter the industrialization verification stage. Companies such as Tianyue Advanced have already sent samples of 12-inch products, targeting advanced packaging heat dissipation substrates and ultra-high-power device scenarios. 12 inches further dilutes per-wafer cost, but equipment investment and process difficulty are higher, and the pace of industrialization is slower than that of 8 inches.
The global market expands rapidly, and the share of incremental contribution from AI computing power supplies continues to rise. SiC market growth no longer depends entirely on new energy vehicle cycle fluctuations, and multiple downstream scenarios hedge single-industry cycle risk.
4.3 Long Term (After 2032): Technology Matures, Costs Fall Sharply, Entering Large-Scale Popularization Stage
SiC device costs continue to decline, gradually replacing silicon-based power devices in more medium- and low-voltage scenarios, with application scenarios expanding to home appliances and consumer power supplies. 12-inch SiC substrates achieve large-scale mass production. The global industrial chain landscape tends toward stability, and China forms a complete, autonomously controllable SiC industrial ecosystem with global competitiveness across the entire chain from substrates, equipment, epitaxy, and chips to modules. At the same time, potential competition exists in technology routes. New-generation wide-bandgap materials such as gallium oxide (Ga₂O₃) pose potential substitution threats in specific high-voltage scenarios, but in the medium and long term they are unlikely to overturn SiC's mainstream position in power semiconductors.
V. Investment Logic and Track Selection (Industry / Equity Investment Perspective)
The SiC track is a capital-intensive, technology-intensive, and long-cycle verification track. Investment cannot simply look at capacity scale; the core lies in yield, defect indicators, customer certification, cash flow, and cost curve. After the 2025 industry reshuffle, investment logic has shifted from "capacity expansion" to "technical barriers + customer barriers."
5.1 Investment Priority Analysis by Industrial Chain Segment
1. Upstream Substrate Segment: Prioritize Automotive-Grade High-Yield Conductive Substrate Enterprises
Substrates are the segment with the highest value and strongest barriers in the entire industrial chain, and the only segment in which China has achieved global leadership. Core investment indicators: conductive substrate yield, micropipe defect density, 8-inch mass production capability, customer certification, and in-house crystal growth equipment capability. Substrate enterprises that can only produce low-defect-indicator products and cannot enter automotive-grade supply chains have weak long-term profitability and are high-risk targets. Upstream crystal growth, cutting, grinding, and polishing equipment offer domestic substitution opportunities, and equipment manufacturers benefit from continued expansion of domestic substrate production lines.
2. Epitaxy Segment: Focus on Epitaxy Enterprises with Stable Batch Supply and Compatibility with Automotive-Grade MOSFETs
Epitaxy quality directly determines device reliability, and automotive-grade epitaxy has high technical barriers. Epitaxy plants need deep collaboration with substrate plants and device plants. Enterprises with stable epitaxial layer thickness and doping control capabilities have long-term value.
3. Midstream Devices and Power Modules: Focus on Automotive-Grade Designations, Reliability Verification, and Packaging Capabilities
The device segment has the deepest moat for overseas giants, with long certification cycles (automaker certification usually takes 2-3 years). Investment should prioritize enterprises that have already obtained designations from leading automakers, AI power supply, and energy storage customers and possess AEC-Q100 certification; power module packaging is a domestic weak point, and high-reliability packaging processes and advanced packaging technologies such as silver sintering offer investment opportunities. Fabless enterprises with pure design and no proprietary production lines face risks from substrate supply and process fluctuations.
4. Downstream Application End: Focus on Power Supply and Electronic Control Enterprises with SiC Solution System Design Capabilities
The value of system manufacturers is not only in purchasing devices, but also in possessing system application, heat dissipation, and control algorithm capabilities for SiC devices, which can maximize SiC efficiency advantages. Prioritize leading enterprises in 800V electronic control, AI server power supplies, and energy storage PCS.
5.2 Investment Strategies
Primary equity investment: Prioritize layout in high-barrier segments such as substrates, upstream equipment, automotive-grade epitaxy, and power modules; avoid low-end substrate projects that rely only on low prices and low yields. Prioritize enterprises that have completed sample delivery and designation verification with leading customers and have sustained R&D investment; be cautious about projects that have not yet entered customer verification and remain only at the sample stage.
Industrial investment: Suitable for vertical integration of the industrial chain. Automakers, energy storage, and computing power supply enterprises can lock up substrate and epitaxy capacity upstream to ensure supply chain security and hedge against overseas supply chain disruption risks.
Valuation focus: The industry is in an expansion cycle, and enterprises generally have high upfront capital expenditures and lagging profitability. PE valuation cannot be used simplistically; focus on tracking yield ramp-up progress, unit wafer cost, customer designation orders, capacity utilization, and cash flow.
VI. Core Industry Risk Warnings
6.1 Risk of Downstream Demand Falling Short of Expectations
If new energy vehicle sales and 800V model penetration fall below expectations, SiC demand will be directly dragged down; if AI server capital expenditure temporarily contracts, AI power supply demand growth will slow. SiC industry demand is deeply tied to the two major capital-expenditure-heavy cyclical industries of new energy and computing power, and there is cyclical fluctuation risk.
6.2 Technology and Yield Risk
SiC crystal growth and epitaxy processes are difficult. Yield improvement progress falling short of expectations will continue to push up costs; ramp-up difficulty of large-size 8-inch and 12-inch production lines may be higher than expected; technological breakthroughs in other wide-bandgap materials such as gallium oxide may create substitution risk.
6.3 Price Competition and Profit Deterioration Risk
Although automotive-grade products are currently in structural shortage, if a large number of 8-inch substrates are put into production in concentrated fashion in the next 2-3 years, the industry may again enter oversupply, triggering a new round of price wars, compressing corporate gross margins, and causing many enterprises to remain loss-making. Industry capital expenditure is enormous and depreciation pressure is high; even with revenue growth, enterprises may still remain loss-making for a long time.
6.4 Patent and International Trade Risk
Overseas leaders hold a large number of core patents for SiC substrates and devices, and domestic enterprises face patent litigation risk; semiconductor export controls and overseas trade barriers affect domestic enterprises' expansion into overseas customers and imports of equipment and materials.
6.5 Certification Risk
Certification cycles for automotive-grade and industrial-grade devices are long. After investing substantial R&D funds, failure in customer certification or delayed designations will directly affect project commercialization.
VII. Conclusion
As a core track of third-generation compound semiconductors, silicon carbide (SiC) has experienced brutal capacity clearing in 2025 and ushers in a structural prosperity inflection point. The growth logic has upgraded from single new energy vehicle drive to multi-track resonance of new energy vehicles + AI computing power supplies + photovoltaic energy storage. In the short term, the industry shows structural supply-demand differentiation: high-quality automotive-grade products are in shortage, while low-end products continue to suffer intense competition; in the medium and long term, large-scale 8-inch cost reduction and domestic substitution are the main lines, and AI computing power supplies open the long-term growth ceiling of the industry.
China's industrial chain has achieved global breakthroughs in the substrate segment, while epitaxy, device, and module segments continue to catch up, but automotive-grade certification, yield, cost, and patents remain core shortcomings. At the investment level, the track's long-term growth space is certain, but the industry is a high-capital-expenditure, long-verification-cycle track and is not suitable for simply chasing capacity concepts. Core criteria for screening investment targets: high yield, low defect indicators, leading customer designations, implementation of automotive-grade certification, and a controllable cost curve. Prioritize layout in high-barrier segments such as upstream substrates, crystal growth equipment, automotive-grade epitaxy, and power modules, and avoid projects without core customers, with low yield, and relying solely on capacity scale. At the same time, it is necessary to continuously track key variables such as new energy vehicle sales, AI capital expenditure, 8-inch yield ramp-up, and overseas patent and trade policies to prevent cyclical fluctuations and risks of industrialization falling short of expectations.