In 2018, the U.S. Department of Commerce placed Synopsys, Cadence, and Mentor (subsequently acquired by Siemens) on a list of potential risk tools subject to export controls against China. The term EDA entered mainstream Chinese public consciousness for the first time. Many people discovered that these three tools, essential to chip design, collectively held over 70% of the global EDA market—and that China's domestic alternatives were virtually nonexistent at the time.
Then seven years passed.
In 2025, Empyrean Technology (Huada Jiutian) achieved revenue of RMB 1.325 billion; its analog EDA tools had received international certification at multiple advanced process nodes. Primarius Technologies' semiconductor device model extraction tools reached parity with international leaders in certain niche segments. Broadway Technology's yield enhancement tools surged nearly 50%. Meanwhile, across the broader industrial software battleground, Yonyou Network and Kingdee International jointly held over 60% of China's ERP system market; CISDI Software's manufacturing execution system held the top global market share in steel for multiple consecutive years; and ZWSOFT's CAD software captured near-complete domestic replacement in 2D design and became the leading domestic option in 3D design. This scorecard would have been unimaginable five years ago.
Before 2020, China's overall industrial software domestic substitution rate was below 25%. CAD/CAE design tools were almost entirely dominated by Dassault, Autodesk, Siemens, and Ansys. In EDA, Synopsys, Cadence, and Siemens (Mentor) collectively held around 90% of global share, and domestic products were starting from zero. Even in the supposedly more competitive ERP market, SAP and Oracle remained irreplaceable for large Chinese enterprises—especially manufacturing industry leaders. China's industrial software, in global terms, was no more than a learner and follower.
Five years later, that landscape is accelerating its transformation. Domestic substitution has crossed the 50% threshold. Strong policy backing, sustained capital investment, and the cumulative depth of technical buildup are turning "domestic substitution" from a policy slogan into industrial reality. Yet look beneath the numbers, and the divergence is equally striking: domestic ERP is mainstream among SMEs, but replacement rates in core ERP modules at large high-end manufacturers remain below 30%; domestic 2D CAD has won, but high-end 3D CAD/CAE for aerospace engine design and automotive body-in-white remains a foreign monopoly; for EDA, three rounds of the "02 Special Fund" and large-scale third-phase National IC Fund investment still leave digital EDA domestic substitution below 15%—with no engineering-ready domestic alternative yet for the most critical logic synthesis and physical implementation tools.
This is China's industrial software reality: half the mountain climbed, half still being scaled.
Executive Summary
This report is a systematic, in-depth survey of China's industrial software value chain, covering five major tracks: CAD/CAE R&D design software, EDA electronic design automation software, ERP/MES/DCS production control software, industrial internet platforms, and industrial AI large models. The time dimension focuses on FY2025 financial data and H1 2026 developments.
Key findings are summarized as follows.
I. Overall market: China's industrial software market reached approximately RMB 350–400 billion in 2025, growing at 6–8%, with overall domestic substitution crossing the 50% threshold—a historic milestone. ERP, DCS/SCADA, and industrial internet lead the domestic substitution race; R&D design tools (3D CAD/CAE) and EDA lag significantly.
II. EDA track: Empyrean Technology, Primarius Technologies, and Broadway Technology generated combined FY2025 revenue of approximately RMB 2 billion+, with extremely high R&D expense ratios (Empyrean exceeding 65%). Domestic analog EDA is commercially deployable at mature process nodes (28nm and above); digital EDA critical links (logic synthesis, physical implementation) remain the deepest bottleneck. The third-phase National IC Fund and "02 Special Fund" continuation provide support, while U.S. export controls continue forcing accelerated commercialization.
III. ERP track: Yonyou Network FY2025 revenue RMB 9.2 billion; Kingdee International approximately RMB 6.3 billion; combined over RMB 15 billion, representing approximately 35–40% of the domestic ERP market. SAP continues to dominate core systems at large manufacturing enterprises, but cloud migration is shortening the domestic substitution window. Kingdee Cosmic and Yonyou BIP5 are frontally attacking the large enterprise market.
IV. MES/DCS track: CISDI Software (steel MES) and SUPCON Technology (chemical DCS) are the top global market share leaders in their respective Chinese segments. Their FY2025 revenues were RMB 10.97 billion and RMB 8.07 billion respectively (both declined due to industry cycles), but their core technical moats are extremely deep.
V. CAD track: ZWSOFT's 2D CAD has largely completed domestic substitution of AutoCAD; domestic 3D CAD penetration in premium scenes (aerospace, automotive) remains 10–20%, with ZW3D being the closest domestic option but still significantly behind CATIA and NX.
VI. Industrial AI large models: Huawei Pangu 5.5 has been deployed across 30+ industries and 500+ scenarios; Baidu ERNIE industrial version, SenseTime, and Zhipu follow. AI integration paths at industrial software vendors (Yonyou YonGPT, Kingdee Cosmic AI, SUPCON Panshi AI) are accelerating in practice, though large model deployment in "direct control" scenarios remains extremely cautious.
VII. Industry structure forecast: China's overall industrial software domestic substitution rate is expected to exceed 70% by 2030; domestic digital EDA partial commercialization at mature process nodes is projected for 2027–2029; industrial SaaS international expansion will become a significant growth increment for leading vendors.
Deepened Research Findings
Several structural findings worth emphasizing emerged during this research, as they may alter conventional analytical frameworks regarding China's industrial software industry:
Finding 1: The "moisture" in domestic substitution rate statistics. The widely cited "China's overall industrial software domestic substitution rate exceeds 50%" includes many cases of "breadth coverage rather than deep substitution." A manufacturer simultaneously using Yonyou (for some modules) and SAP (for core financials) may be counted in domestic substitution statistics, yet SAP remains the main system. Measured by "core business function" rather than "whether domestic software is installed," the actual domestic substitution rate at large manufacturers is significantly below the nominal figure—conservatively estimated at no more than 30–35%.
Finding 2: The essential difference between "domestic substitution" and "domestic excellence." Truly competitive domestic industrial software should not take "replacing foreign software" as its ultimate goal, but rather "providing the best solution in the industry." SUPCON Technology's ability to defeat Honeywell and Emerson in chemical DCS came not from "Xinchuang policy" but from genuinely understanding Chinese chemical plants' specific operating habits, process parameter distributions, and maintenance team capabilities better than foreign competitors.
Finding 3: The "last mile" is implementation service, not product functionality. Research repeatedly found that domestic industrial software products scoring close to foreign counterparts in feature assessments still showed significant gaps in "implementation success rate" and "customer satisfaction." The root cause lies not in product functionality but in the industry experience depth of implementation teams.
Finding 4: Data assets are the hidden competitive moat. Industrial software accumulates valuable training data during operation—equipment fault records, correlations between process parameters and product quality, energy-productivity optimization patterns. CISDI's twenty years of steelmaking data, SUPCON's twenty years of chemical process data, and Yonyou's financial process data from thousands of enterprises are competitive moats that cannot be purchased or replicated.
Chapter 1 Definitions, Classification, and Value Chain Overview
"Industrial software" in the Chinese context is often discussed as a monolithic whole, but it actually encompasses multiple subcategories that differ fundamentally in technical architecture, commercial logic, and application scenarios. Before diving deeper, it is necessary to clarify this classification map, because understanding it is prerequisite to understanding why some segments have achieved rapid domestic substitution while others have struggled for twenty years with little progress.
R&D Design Software (CAD/CAE/EDA/PLM/CAM)
This is the highest-barrier, most foreign-dominated, most difficult-to-substitute major category.
CAD (Computer-Aided Design) handles 3D modeling and engineering drawings—the digital starting point for all manufacturing activity. An automobile body, an aircraft wing, a chip layout: all take shape in CAD before any physical material appears. 2D CAD centers on precise management of lines and layers; 3D CAD centers on solid modeling and surface modeling algorithms based on mathematics such as NURBS (Non-Uniform Rational B-Splines).
CAE (Computer-Aided Engineering) handles simulation analysis. After design drawings are complete but before manufacturing begins, engineers must use simulation to verify whether the design can withstand real operating conditions—structural analysis (will the part break under load?), computational fluid dynamics (is engine intake flow reasonable?), thermal analysis (is the PCB temperature distribution safe under high load?), electromagnetic simulation (does the antenna radiation pattern meet requirements?). CAE software's core is finite element methods (FEM), finite volume methods (FVM), finite difference methods (FDM), and corresponding high-performance solvers. CAE simulation accuracy directly affects engineering decisions; the tolerance for error is extremely small.
EDA (Electronic Design Automation) is specialized software for integrated circuit (chip) and printed circuit board (PCB) design. A modern chip may contain billions of transistors; the design process covers the full chain from logical function description (RTL code) to physical layout (GDS2 files), with each step (logic synthesis, simulation verification, timing analysis, place-and-route, layout verification, design for manufacturability, etc.) having dedicated specialized tools. EDA tool complexity arises from needing deep integration with the process design kit (PDK) provided by the foundry, ensuring the design can be successfully realized in manufacturing.
PLM (Product Lifecycle Management) unifies management of all product data from concept design through end-of-life, serving as the spine of digital collaboration in large manufacturing enterprises. Its data exchange interfaces with CAD, CAE, and ERP form the skeleton of enterprise digitalization.
CAM (Computer-Aided Manufacturing) translates CAD designs into G-code (NC programs) that CNC machine tools can directly execute—the last mile connecting design to manufacturing.
These five categories share the characteristics of: heavy dependence on mathematical kernels (geometry engines, FEA solver algorithms, simulation software core solvers), decade-long development cycles, iterative validation requiring continuous real-project feedback, and extreme customer stickiness once established. They are also the industrial software types most stably controlled by a handful of global superpowers.
Industrial software differs fundamentally from consumer software in its "fault tolerance" and "certification requirements." A consumer software bug causes degraded user experience; an industrial software bug may mean scrapped production worth tens of millions of RMB, a production line shut down for days, or even a safety incident. This extreme low-tolerance attribute makes industrial software validation cycles measured in years, commercial barriers far higher than consumer software, and "trust from proven track record" decisive in procurement decisions.
From an economics perspective, industrial software exhibits significant "learning curve effects": the more a tool is used across diverse engineering projects, the stronger its capabilities become, and stronger capabilities attract more users, creating a positive flywheel. Synopsys and Cadence have maintained over 70% combined EDA market share for over twenty years—not because they launch disruptive innovations each year, but because their tools are continuously validated on the world's largest volume of chip design projects, accumulating engineering experience data assets that no challenger can replicate. For China's domestic industrial software, this means: only through real-project "volume accumulation" can the gap with leaders be genuinely narrowed.
The "standard format" of industrial software is another significant moat. Autodesk's DWG format is the de facto standard for 2D design globally; CATIA's formats are the de facto standard for aerospace supply chain collaboration; GDSII/OASIS in IC design are global layout data exchange standards. These format standards are essentially information exchange protocols; once they become industry standards, they become the data infrastructure of the entire industry.
From a talent supply perspective, truly capable developers of top industrial software core algorithms are extremely scarce globally. Developing finite element solvers requires mastery of PDE numerical methods, high-performance parallel computing architecture, and structural mechanics engineering knowledge simultaneously. This highly composite talent type numbers only in the thousands globally, concentrated mainly in U.S., European, and Japanese top industrial software companies. China's industrial software industry has shown markedly improved attraction of senior algorithm engineers in recent years, but the scarcity of core talent will remain a key constraint on development pace for a considerable period.
Production Control Software (ERP/MES/DCS/SCADA)
ERP (Enterprise Resource Planning) is the central nervous system of enterprise operations management, covering financial management, procurement management, sales management, inventory management, manufacturing management, and human resources across six core modules. The essence of ERP is linking all business processes inside an enterprise with a unified data model for real-time information sharing and standardized process execution.
MES (Manufacturing Execution System) is the factory floor production control system, connecting ERP's planning layer with shop floor automation. MES tracks work order progress, material consumption, equipment status, and quality data in real time. Its complexity lies in needing real-time communication with PLC, SCADA, robots, and sensors on the shop floor while simultaneously interfacing upward with ERP—this "real-time on both ends" requirement makes MES technically far more difficult than ERP.
DCS (Distributed Control System) is the automation core for process industries (chemical, petrochemical, refining, power, pharmaceutical), controlling continuous production processes in real time. A chemical plant DCS may include thousands of control loops monitoring hundreds of process variables, executing emergency shutdown logic (ESD), ensuring the process operates within safety boundaries. This extreme reliability requirement makes DCS replacement decisions extremely conservative.
SCADA (Supervisory Control and Data Acquisition) is used for infrastructure (power grid, water treatment, oil and gas pipelines) remote monitoring and discrete manufacturing, distinguished from DCS by broader geographic coverage and relatively lower real-time control density.
Production control software depends intensely on industry know-how—a good chemical DCS embeds not just software logic but decades of petrochemical process pattern recognition, control strategy refinement for specific plants, and codified abnormal condition handling. This is why SUPCON Technology's DCS leads globally in chemicals—it's not just software, it's a thirty-year container of Chinese chemical industry process data.
Industrial Internet and Platform Software
This is the youngest industrial software category. Industrial internet platforms connect all factory machinery, devices, sensors, control systems, MES, and ERP to the cloud, enabling data aggregation, real-time analytics, and cross-factory coordination. Representative platforms include Haier COSMOPlat (national-level industrial internet demonstration platform), TreeRoot (Sany-incubated), Alibaba Cloud supET, Midea's Meiqi, Sany's Root Cloud, and XCMG's Hangyun.
Industrial AI Large Models
This is the most active technical direction of 2024–2025. General large models (Huawei Pangu, Baidu ERNIE, SenseTime Nova, Zhipu GLM, ByteDance Doubao) intersect with industrial software specialized vendors (Yonyou YonGPT, Kingdee Cosmic AI, SUPCON Panshi AI, CISDI AI MES) in this direction. The core value of industrial AI large models is transforming vast unstructured industrial knowledge (equipment manuals, process specifications, quality standards, historical fault records) into knowledge bases queryable and callable via natural language.
The "value chain" of industrial software differs fundamentally from manufacturing—it connects through knowledge transfer, not logistics; through algorithm and data long-term accumulation, not component assembly. Downstream use cases span nearly all modern manufacturing: automotive parts (high CAD/CAE/ERP density), integrated circuits (core EDA scenario), petrochemical industry (DCS-dominated), aerospace (high-end CAD/CAE/PLM), construction machinery (CAD/PLM/MES), building materials (BIM software), medical devices (PLM/quality management), and lithium batteries (MES/automation).
Industrial software does not produce physical products, but it is the digital infrastructure through which all physical products can be precisely designed, stably produced, and efficiently managed. Whoever controls industrial software controls the capability ceiling of manufacturing.
Chapter 2 Global Competitive Landscape and Overseas Giants FY2025
The global industrial software market has long been dominated by a small number of super-players. Understanding this landscape is a prerequisite for understanding where domestic industrial software can break through—and why breakthrough is so difficult.
SAP: Global ERP Leader, Accelerating Cloud Transformation
Germany's SAP is the world's largest enterprise application software company and the undisputed global leader in ERP. FY2025 total revenue was EUR 36.8 billion, up 8% YoY (11% at constant currency). Cloud services revenue was EUR 21 billion, up 23%; its flagship cloud ERP suite (S/4HANA Cloud) grew 28%, reaching EUR 18.1 billion. SAP's market positioning is completing the transition from traditional license to cloud subscription model. In China, SAP has chosen to deepen cooperation with Alibaba Cloud, deploying S/4HANA on Alibaba Cloud infrastructure as "local cloud" to navigate Xinchuang policy pressure while maintaining relationships with large Chinese enterprise customers.
Oracle: ERP and Database Dual Engines, China Market Under Pressure
Oracle's industrial ERP products (Oracle Cloud ERP Manufacturing, Oracle Supply Chain Management Cloud) target large discrete manufacturers, but Oracle's localization service resources in China fall far short of SAP's. Oracle FY2025 total revenue was approximately USD 60 billion, with cloud services growing 30%+, but its presence in China's manufacturing industry continues to be eroded by Yonyou and Kingdee.
Autodesk: Platform Pioneer in CAD/CAM
Autodesk's AutoCAD is the de facto global standard for 2D CAD. FY2025 total revenue was approximately USD 6.6 billion, up ~12%, with SaaS subscriptions dominating (one-time licenses nearly phased out). Autodesk faces direct competition from ZWSOFT in China, especially in SME and government procurement segments where significant share has already been lost. Autodesk's transition from perpetual license to annual subscription has created a window for domestic substitution—Chinese SMEs accustomed to "buy-out" software perceive the annual fee much more expensively.
Dassault Systèmes: King of Integrated CAD/CAE/PLM
France's Dassault Systèmes has arguably the most complete industrial software portfolio. CATIA is nearly irreplaceable in global aerospace and automotive body design. SIMULIA (including Abaqus FEA solver), ENOVIA (PLM collaboration), SOLIDWORKS (SME 3D CAD), and Dymola (multi-physics system simulation) round out its suite. FY2025 total revenue was approximately EUR 6 billion. Its core competitive moat is "digital continuity"—from product concept (CAD), to manufacturing process planning (DELMIA), to product validation (SIMULIA), to lifecycle data management (ENOVIA), all data stored in the unified 3DEXPERIENCE platform format, creating a highly locked ecosystem.
PTC: Pioneer in CAD to Industrial IoT Transformation
PTC started with Creo 3D CAD and has transformed aggressively toward industrial IoT. ThingWorx is a globally recognized industrial IoT platform; Vuforia leads industrial augmented reality. PTC FY2025 revenue was approximately USD 2.4 billion. In China, PTC has deep industrial internet cooperation with Haier COSMOPlat, and Creo has some installed user base in automotive and machinery.
Siemens Digital Industries Software: Full-Chain Industrial Digitalization
Siemens DI Software spans NX (flagship 3D CAD/CAM), STAR-CCM+ (CFD), Simcenter (multi-physics simulation), Opcenter (MES, strong in semiconductors and pharma), Mendix (low-code), Teamcenter (PLM), and EDA tools from the Mentor acquisition (Questa, Calibre, HyperLynx), making Siemens the world's third-largest EDA company. DI Software accounts for approximately 15% of Siemens Group revenue.
Ansys (absorbed by Synopsys): CAE-EDA Cross-Sector Fusion
Ansys was the global leader in CAE simulation with products covering structural (Mechanical), fluids (Fluent/CFX), electromagnetics (HFSS/Maxwell), and electronics reliability (Sherlock). The 2025 Synopsys acquisition of Ansys at ~USD 35 billion created a "super industrial software platform" spanning EDA and CAE, providing a unified tool chain from chip design through chip package simulation, PCB electromagnetic simulation, and system thermal analysis.
Synopsys: EDA Global Leader After Ansys Acquisition
Synopsys is the world's largest EDA software company. Its Fusion Compiler, PrimeTime, VCS, and Design Compiler are core infrastructure for chip designers globally. FY2025 total revenue was USD 7.054 billion, up ~15%. Synopsys' moat: its tools are deeply integrated with process design kits (PDKs) from all major foundries—any chip design company using a specific process node must use EDA tools certified against that foundry's PDK, a relationship taking years to establish.
Cadence: EDA #2, King of Analog Design Ecosystem
Cadence forms a duopoly with Synopsys in EDA. FY2025 total revenue was approximately USD 5.7 billion, growing steadily. Cadence's Virtuoso platform is the preferred tool for global analog IC design—the primary competitive field for Empyrean's analog EDA. Cadence's Allegro PCB Designer leads in PCB layout.
Global FY2025 Financial Data Summary
Key financials across major global industrial software companies FY2025 (each company's most recent complete fiscal year):
SAP: EUR 36.8B revenue, ~22% operating margin, ~EUR 2,500B market cap. Oracle: ~USD 60B revenue. Autodesk: ~USD 6.6B revenue, ~37% non-GAAP operating margin. Dassault: ~EUR 6B revenue. PTC: ~USD 2.4B revenue. Synopsys+Ansys: ~USD 7.1B revenue, ~31% non-GAAP operating margin. Cadence: ~USD 5.7B revenue, ~35% non-GAAP operating margin. Siemens DI Software: ~EUR 6B (Siemens Group segment data). Bentley Systems: ~USD 1.5B revenue. Hexagon AB: ~EUR 5B total revenue, ~EUR 3B from industrial software.
This data reveals common features of the global industrial software market: high gross margins (typically 70–80%+), strong recurring revenue (maintenance/subscription increasingly dominant), and high R&D intensity (typically 20–25% of revenue). These features make global industrial software businesses financially excellent—and make entry extremely capital-intensive. Reaching commercially viable technical capability in any single segment typically requires 10+ years and hundreds of millions of RMB.
The Synopsys-Ansys merger's deeper implication: global top industrial software companies are evolving from "specialized tool providers" to "vertical industry full-stack digital platforms." Competing based purely on feature parity will make it increasingly difficult for domestic industrial software to find a foothold in premium markets. The truly effective breakthrough path is building integrated competitiveness from industry know-how to tool in specific vertical segments—the model of SUPCON Technology's approach to chemical DCS.
India is becoming an important base for global industrial software R&D outsourcing. Synopsys, Cadence, Ansys, and Autodesk have all established large R&D centers in Bangalore, Hyderabad, and Pune. This positions China's industrial software companies to face pressure not just from U.S. and European competitors but also from India's continuously rising R&D capabilities.
Global SaaS migration continues to accelerate outside China. Autodesk FY2025 SaaS subscription revenue exceeds 90%; SAP S/4HANA Cloud growth (28%) significantly outpaces overall business growth. SaaS industrial software has clear advantages in iteration speed, AI integration, and multi-site collaboration versus traditional on-premise systems, potentially widening the gap between "global advanced level" and "China on-premise installed base."
Other Major Global Players
Hexagon (Sweden): largest measurement technology and industrial software conglomerate, with MSC Software (acquired) providing CAE capabilities strong in quality management. Bentley Systems: focused on infrastructure digitalization—roads, bridges, water, power—with MicroStation and OpenRoads competing with Autodesk. AVEVA (Schneider Electric): major in process industry digitalization with AVEVA Process Simulation and AVEVA System Platform for chemical and power industries, competing with SUPCON's DCS/MES.
Chapter 3 Policy, Regulation, and Industrial Environment (PEST)
Policy Environment: Systematic Push at National Strategic Level
China's industrial software policy has evolved from project-level support to systemic construction over the past five years. The "02 Special Fund" (National Science and Technology Major Project on Very Large Scale Integrated Circuit Manufacturing and Advanced Packaging Technology) has entered its third round, focusing on EDA and semiconductor equipment. The third-phase National IC Fund (RMB 344 billion) explicitly lists EDA tools and industrial control software among key investment directions.
Procurement Preference Policy
The "dual evaluation" system—requiring government procurement and state-owned enterprise procurement to include "domestic brand preference" and "domestic software priority"—creates a structural tailwind for domestic industrial software vendors in certain segments.
Industrial Software First-Purchase Support
The "first article" (首台套) policy has been extended to industrial software, providing financial subsidies or loan interest discounts for "first large-scale commercial deployment of domestic industrial software" projects. This reduces the risk perception of first-time domestic adoption—providing a risk hedge mechanism for factories considering switching to domestic tools for the first time.
National Industrial Software Evaluation System
The Ministry of Industry and Information Technology has commissioned professional evaluation institutions to conduct multi-dimensional third-party evaluations of domestic industrial software, publishing results as reference for government procurement and SOE procurement decisions. This addresses the information asymmetry problem of "how to objectively evaluate domestic tool capabilities."
Software Intellectual Property Protection
A hidden constraint on China's industrial software development has been low software licensing compliance rates. The improvement in licensing compliance awareness and enforcement, plus Autodesk's migration to subscription models increasing piracy difficulty, has objectively created a more level playing field for legitimate domestic industrial software. When enterprises can no longer "freely" use pirated AutoCAD, ZWSOFT's cost advantage becomes genuinely perceived by decision-makers.
"Belt and Road" International Cooperation Opportunity
Vietnam, Indonesia, Thailand, Mexico, and other countries experiencing accelerated manufacturing development have factories often built by Chinese enterprises that naturally need to bring domestic industrial software with them. Saudi Arabia and UAE's large-scale industrialization plans ("Vision 2030") are creating significant market increments for infrastructure digitalization software (BIM, industrial internet). China's industrial software "going global" initial market follows the path of "wherever Chinese manufacturing goes, Chinese industrial software follows."
Carbon Neutrality Goals and Industrial Software's New Intersection
The "dual carbon" goal (carbon peak 2030, carbon neutrality 2060) is opening a new subcategory: industrial carbon management software. The domestic market in this segment is estimated at RMB 30–50 billion in 2025, growing at over 30%. Local vendors have a clear advantage in "Chinese carbon market rules and verification standards" domain knowledge—this is one of the few industrial software segments where local policy knowledge barriers exceed international technical knowledge barriers.
Xinchuang Policy's Industry-by-Industry Implementation Rhythm
"Xinchuang" (information technology application innovation) policy impact varies dramatically by industry. In "strong Xinchuang" industries (party/government agencies, financial institutions, telecom carriers), domestic requirements are specific and time-bound. In "medium Xinchuang" industries (state-owned manufacturers, energy enterprises, large SOEs), requirements extend progressively from office systems to industrial control and ERP, with relatively lenient implementation rhythm. In "weak Xinchuang" industries (foreign-invested enterprises, private SMEs), Xinchuang is primarily guidance rather than mandate.
This industry-by-industry rhythm determines fundamentally different substitution speeds in different customer segments. The long-term health of domestic industrial software development ultimately depends on whether it can win on merit in "weak Xinchuang" competitive environments—not just maintain share in policy-protected "strong Xinchuang" environments.
Industry Talent Development: National Industrial Software Talent Cultivation Plan
The joint plan by MIIT and Ministry of Education for 2024–2025 includes: establishing dedicated "industrial software" direction cultivation programs at key universities (Beihang, NWPU, HIT, SJTU); supporting enterprise-university joint "outstanding engineer" programs; creating green channels for returning overseas talent in EDA, CAE solver fields. By 2028, China's industrial software core talent pool is projected to more than double from 2023 levels.
Chapter 4 China Market Size, Domestic Substitution Rate, and Competitive Concentration
Overall Market Size
China's industrial software market reached approximately RMB 350–400 billion in 2025, representing 6–8% growth. At purchasing power parity, China accounts for approximately 20–25% of the global industrial software market.
Segmented Market Sizes (FY2025 estimates)
ERP/finance management software: approximately RMB 140–160 billion; domestic substitution rate ~55% (weighted by revenue, not user count). CAD/CAE/PLM R&D design: approximately RMB 70–90 billion; domestic substitution rate ~25–30%. EDA: approximately RMB 8–12 billion total (including foreign vendor China sales); domestic substitution rate ~15–20% (analog EDA approaching 35%, digital EDA below 15%). MES/DCS/SCADA production control: approximately RMB 60–80 billion; domestic substitution rate ~55–60%. Industrial internet platform + IoT software: approximately RMB 60–80 billion; domestic substitution rate ~75–80%. Industrial AI + new applications: approximately RMB 20–30 billion (rapidly growing new category).
Competitive Concentration Analysis
At the overall industrial software industry level, China's top 10 companies by revenue account for approximately 40–45% of total market revenue—a relatively high degree of concentration, but significantly lower than global comparable data (where the top 5 global companies account for over 50% of global revenue, reflecting the different competitive structure between the global market (few superpowers dominating) and the Chinese market (domestic vendors competing fiercely while sharing with foreign vendors).
Effective Substitution Rate vs. Nominal Coverage Rate
More accurate evaluation requires the "effective substitution rate" dimension—the proportion of core business functions borne by domestic software among enterprises that have installed it. Measured by module for large manufacturers (annual revenue above RMB 5 billion): financial accounting ~25% domestic; supply chain management ~30%; manufacturing management ~35–40%; HR and payroll ~55–60%; expense management ~70%; BI/data analytics ~40–50%.
This modular view shows the true depth of domestic ERP substitution—not simply "which software is used" but "where does core business actually run."
The Three-Tier Competitive Landscape
Tier 1 (domestic-dominated segments): DCS in chemicals (63% domestic), MES in steel (CISDI #1 globally), ERP in SMEs (>70% domestic), industrial internet platforms (75–80% domestic). Tier 2 (domestic share rising fast, foreign still leading): ERP in large manufacturers' core modules (30% domestic), 2D CAD (80% domestic but declining piracy-conversion opportunity), MES in automotive and consumer electronics (~30–40% domestic). Tier 3 (domestic penetration still extremely low): high-end 3D CAD in premium scenes (<20%), high-end integrated CAE fluid/structure simulation (<15%), digital EDA critical links (<15%), PLM in major aerospace enterprises (nearly all foreign).
Chapter 5 Deep Value Chain Analysis
EDA Value Chain Technical Deep Dive
EDA's technical complexity is unique among all industrial software subcategories because it serves a highly specialized, extremely precise industry ecosystem—semiconductor manufacturing. Understanding EDA requires starting from the EDA-PDK relationship.
PDK (Process Design Kit) is the set of data files provided by foundries to chip design companies, describing what components (transistors, resistors, capacitors) can be used at that process node, their electrical parameters, layout design rules (DRC rules), and parasitic extraction models. Without PDK, EDA tools cannot simulate chip physical behavior under real foundry process conditions.
EDA tool-PDK binding is the foundation of EDA market moats. TSMC maintains certified PDKs with different EDA tools for each process node (5nm, 7nm, 12nm, 28nm, etc.), meaning EDA tools must pass TSMC's rigorous certification process (typically 1–3 years) to be permitted for use. Empyrean's partial analog EDA tools entering TSMC's certification list at mature process nodes is an important milestone.
The ten core links of the EDA tool chain divide into front-end (RTL design, functional simulation, logic synthesis, formal verification) and back-end (floorplanning, place-and-route, STA, DRC/LVS, parasitic extraction, yield engineering). Of these, logic synthesis and physical implementation—the most critical—remain nearly a blank for domestic vendors; place-and-route is in early commercialization; STA has more mature domestic tools (e.g., Xinhuazhang's STA); analog simulation (SPICE) at Empyrean's ALPS has reached commercial level at mature nodes; layout verification (DRC/LVS) has entry-level domestic tools but enormous gaps at advanced nodes vs. Mentor Calibre; yield engineering is the bright spot with Broadway Technology achieving international commercial validation.
CAE Value Chain: Mathematical Core and Engineering Know-How
CAE simulation software's core competitiveness is "numerical solver accuracy and speed" technically, and "simulation benchmark validation cases accumulated in specific industries" commercially. Both are indispensable.
Structural analysis (FEA/FEM) is the largest CAE subcategory—dominated globally by Ansys Mechanical, MSC Nastran (Hexagon), and Abaqus (Dassault SIMULIA), all with 40+ years of solver development history and tens of millions of lines of code continuously refined through real engineering project validation. Domestic structural simulation software (ANSYS China equivalents: ANSOL, Geostar, IntelliSIM) has made some progress but remains significantly behind in multi-material multi-physics coupled simulation accuracy, large-scale parallel solver efficiency, and integration with mainstream CAD.
CFD (Computational Fluid Dynamics) is dominated by Ansys Fluent globally, followed by STAR-CCM+ (Siemens) and OpenFOAM (open source with commercial variants). Domestic CFD tools (PERA-CFD, Xi'an Jiaotong University spinoffs) can meet some engineering needs in basic functionality but show significant gaps vs. Fluent in high-accuracy large-scale parallel computation (e.g., million-cell LES simulations).
Industrial Internet Platform Standards and Interface Ecosystem
Industrial internet platform competition is substantially about "protocol and interface standards." Key communication protocol standards include: OPC UA (universal industrial device communication, widely used by Siemens/Schneider/ABB); MQTT (lightweight IoT messaging, adopted by Alibaba Cloud and Huawei IoT platforms); Modbus (legacy industrial bus protocol, widespread in existing DCS/PLC equipment); EtherNet/IP (Allen-Bradley PLC); Profinet (Siemens PLC).
China's industrial internet platform standards fragmentation remains a significant barrier to large-scale commercialization. Different platforms' interface standards are highly fragmented—a factory using both Haier COSMOPlat and Huawei FusionPlant requires near-custom development for cross-platform data interchange.
Chapter 6 Key Enterprise Profiles
1. Yonyou Network (600588.SH) — Domestic ERP Absolute Leader
Yonyou Network is China's largest enterprise management software company with FY2025 revenue of approximately RMB 9.2 billion. Its cloud business (subscription revenue) was approximately RMB 3 billion, growing ~30%. BIP5, its next-generation AI-embedded cloud ERP platform, is the core weapon for attacking large enterprise markets. YonGPT—natural language financial assistant, intelligent supply chain prediction, AI-driven factory digital operations—represents the AI differentiation attempt. Yonyou has over 3,000 authorized channel partners nationally, which form the capillary system for market penetration and directly support rapid penetration of medium-large enterprise segments.
2. Kingdee International (HK: 0268) — Cloud ERP and PaaS Pioneer
Kingdee International FY2025 revenue was approximately HKD 6.3 billion, with cloud subscription revenue growing ~25–30%. Kingdee Cosmic's competitive positioning emphasizes enterprise-grade PaaS capability, allowing customers to rapidly customize business applications on the Cosmic platform via low-code—a differentiated approach especially attractive to "high customization requirements, complex processes" customers (large SOEs, financial groups). Kingdee's core customer structure emphasizes finance, large SOEs, and central enterprises, while Yonyou's positioning is more balanced across industry segments.
3. Empyrean Technology / Huada Jiutian (688003.SH) — Domestic EDA Leader in Analog
Empyrean Technology FY2025 revenue was approximately RMB 1.325 billion, up ~57% YoY—the fastest-growing major domestic industrial software listed company. Its R&D expense ratio exceeds 65% (over RMB 800 million in R&D), operating at a loss, primarily supported by government subsidies and the IC Fund. This is the "heavy R&D investment" phase typical before EDA market commercialization—Synopsys and Cadence themselves went through long loss-accumulation periods before reaching today's 30%+ operating margins.
Empyrean's strategic choice concentrates superior resources on building completeness in the analog EDA full-process tool chain rather than frontally attacking digital EDA's core links (logic synthesis, physical implementation). This is rational—analog IC EDA is technically challenging but depends less on transistor count, has relatively shorter validation cycles, and China has a large installed base of domestic analog chip design companies (power management IC, RF front-end, sensor interface) forming accessible domestic customer pools.
Empyrean's technical moats: ALPS analog simulator SPICE kernel accuracy validated in multiple production tape-outs; InVar layout verification tool rule set coverage approaching international competition at some mature process nodes; full-flow tool chain completeness allowing customers to complete entire analog IC design processes using Empyrean tools alone.
4. Primarius Technologies (688206.SH) — Differentiated EDA in Foundry-Side Tools
Primarius has taken a different competitive path from Empyrean: not pursuing full-flow EDA tool coverage but building differentiated advantages in specific high-value tools on the "foundry side." Its core products include device model extraction (SPICE model extraction), DFY (Design for Yield) tools, and EDA process automation platforms. The strategic focus on "foundry side" and "OSAT side" rather than "design side" (fabless chips) creates complementarity rather than competition with Empyrean. FY2025 revenue was approximately RMB 450 million, growing ~30–35%.
5. Broadway Technology / Guangli Wei (688719.SH) — Yield Engineering EDA with Global Reach
Broadway Technology focuses on semiconductor yield enhancement EDA tools and is one of the most internationalized domestic EDA companies—overseas revenue exceeds half of total, with customers including several globally recognized foundries and OSATs. Its core tool value: helping foundries identify key process parameters affecting chip yield from massive inline measurement data and providing parameter optimization recommendations. Every percentage point of yield improvement can mean hundreds of millions of RMB in direct value for a large foundry. FY2025 revenue was approximately RMB 300 million, growing ~40–50%.
6. Xinhuazhang (Xinhua Zhang, unlisted) — Pioneer in Domestic Digital Verification EDA
Founded in 2020, Xinhuazhang focuses on digital EDA in formal verification and hardware-assisted simulation, having achieved early commercial validation in digital EDA's most difficult directions. Its AI-driven formal verification tool shows quantifiable improvement in reducing manual verification workload and has commercial contracts at domestic chip design companies. Its approach of embedding AI capabilities in traditional verification tools—providing value from a new dimension rather than simply replicating Synopsys VCS feature sets—exemplifies a potentially viable "technology leapfrogging" path.
7. ZWSOFT / Zhongwang Software (688083.SH) — 2D CAD National Champion
ZWSOFT FY2025 revenue was approximately RMB 1.77 billion, up ~13%, with R&D expense ratio ~30%. International revenue accounts for ~15–20% of total, primarily from SEA and European SME customers. ZWCAD's near-complete DWG compatibility exceeds 99% against AutoCAD. ZW3D is the most functionally complete domestic 3D CAD/CAM product, with commercial deployments in light industry (mold design, consumer product design) and moderate-complexity mechanical design. However, gaps vs. CATIA/NX in complex surface modeling precision, large assembly performance, and CAE integration remain significant in premium scenes.
The PLM market remains even more severely underpenetrated domestically than high-end CAE and digital EDA. PLM's core value is not just software functionality but embedding the best practices of product development processes—something accumulated over 30+ years that cannot be replicated through short-term R&D investment.
8. CISDI Software (600845.SH) — Global #1 in Steel MES
CISDI Software FY2025 revenue was approximately RMB 10.97 billion (down from the prior year due to steel industry capital expenditure cycle downturn). Its MES for steel production is the global market share leader, managing production data for steelmakers in China, Southeast Asia, and beyond. CISDI's AI-driven quality prediction model embedded in steelmaking MES can predict slab internal crack probability 20 minutes before tapping, enabling preventive parameter adjustments. This "proactive optimization" MES is a fundamental upgrade from traditional passive recording systems.
9. SUPCON Technology (688777.SH) — Global #1 in Chemical DCS
SUPCON Technology FY2025 revenue was approximately RMB 8.07 billion (also down due to chemical industry capex cycle). Its proprietary HIACS DCS system is the global market share leader in the chemical industry. SUPCON's competitive moat lies in its twenty-plus years of continuous improvement in China's chemical plants—not just software, but codified control strategies for specific plant configurations and abnormal condition handling protocols developed over thirty years of Chinese chemical industry data accumulation. SUPCON's ECS-700 DCS has national safety certifications at the highest levels (SIL3), equivalent to international leaders.
10. SUPCON's Panshi AI and the Industrial AI Integration Path
SUPCON's Panshi AI integrates with ECS-700 DCS to provide AI-driven advanced process control (APC): optimizing operating parameters to reduce energy consumption by 3–5% while improving product quality stability during normal chemical plant operation. This "active optimization" capability represents the fundamental definition upgrade for industrial control software—shifting from passive execution to proactive recommendation.
11. Guanglian Da (002410.SZ) — Construction Industry Software Dragon
Guanglian Da is the absolute leader in China's construction industry software (BIM and cost management), with FY2025 revenue of approximately RMB 7.5 billion, growing ~10%. Its BIM software and construction cost management tools dominate the market with deep integration of China-specific calculation norms (national quantity take-off standards, regional pricing databases). This local knowledge moat is impossible for Autodesk Revit or Dassault's BIM platforms to replicate quickly. Its cloud transformation (Guanglian Da Digital Dimension Cloud) has shown progress, with cloud subscription revenue continuously growing as a share of total—a model case for domestic industrial software SaaS migration.
12. CISDI Software and Stone Chemical: Petrochemical Full-Process Digitalization
SINOPEC's Stone Chemical (Shiyuaying Ke) integrates DCS real-time data, ERP plan data, MES manufacturing data, and HSE safety data across dozens of refineries in China, representing the most comprehensive domestic petrochemical digital platform. Its commercial model of "captive SOE deployment then commercial expansion" complements SUPCON's approach: SUPCON focuses on process control (DCS/APC), Stone Chemical focuses on business informatization (ERP/MES integration).
13. Nergy Technologies, Zhengying, Broadway Integration Enterprises
In the industrial internet and digital integration space, several important players deserve mention: Nengke Technology (603871.SH), one of China's largest Siemens automation solution partners, accumulates valuable industry system integration experience and know-how even while deploying primarily foreign software, representing the "system integrator who accumulates toward self-developed products" path. Glodon (002410.SZ) represents the success model for deep vertical specialization in construction. Hanwei Technology (300007.SZ) represents the "sensor hardware + IoT platform" bottom-up path.
14. Science and Technology Giants Entering Industrial Software
Huawei Enterprise BG, Alibaba Cloud, Tencent Industrial Cloud, and ByteDance's Volcano Engine are all extending deeply into industrial digital software. Their competitive advantages lie not in industry know-how depth but in: cloud infrastructure capabilities, AI capabilities (large models, machine learning, computer vision), and ecosystem resources. The resulting competitive-cooperation dynamic with traditional industrial software vendors will be one of the most uncertain dimensions of industrial software's competitive landscape over the next five years.
Chapter 7 Industrial Belt and Innovation Cluster Map
China's industrial software industry exhibits a clear geographic concentration pattern closely correlated with manufacturing industry density, R&D resource distribution, and policy support intensity.
China's industrial software research and development is most intensely concentrated in Beijing-Tianjin-Hebei, the Yangtze River Delta, and the Pearl River Delta regions. When Tianxiagongchang tracks the geographic distribution of 4.8 million active manufacturing factories, the correlation between industrial software vendor clustering and downstream factory density becomes apparent—industrial software vendors gravitate toward areas with the largest concentrations of their target customers.
Shenzhen-Guangzhou Pearl River Delta EDA Cluster
Shenzhen is China's most important IC design center, hosting Huawei HiSilicon, OPPO Research Institute, BYD Semiconductor, Goodix Technology, and hundreds of other chip design companies. This makes Shenzhen the densest city for EDA tool procurement in China. Empyrean Technology's primary R&D base is in Shenzhen; several domestic EDA startups are also headquartered there. Guangzhou's "Digital Economy Pilot Zone" provides specialized support for industrial software companies including EDA, attracting additional talent concentration.
Shanghai Yangtze River Delta Industrial Software Hub
Shanghai is both an R&D and business center for industrial software and a major procurement base for automotive (SAIC Group), precision manufacturing, and aviation industry. Primarius Technologies (EDA) is headquartered in Shanghai; dozens of MES and ERP implementation service companies cluster around SAIC and its supply chain in Anting. The "National Industrial Internet Development Demonstration Zone" status accelerates Shanghai's role as a hub for industrial internet platform companies.
Beijing R&D Innovation Center
Beijing hosts leading research universities (Tsinghua, Peking University, Beihang, UCAS) that generate the bulk of EDA algorithm research talent; major technology companies (Huawei AI Cloud, Baidu Industrial AI, Zhipu AI) with industrial software strategic directions; and aviation and aerospace industry institutions (COMAC, AVIC) that drive high-end CAD/PLM demand.
Zhejiang-Hangzhou DCS and Industrial Internet Base
Hangzhou-based SUPCON Technology's growth is intimately linked to the Ningbo-Zhoushan petrochemical industrial belt and Jiaxing fine chemical clusters—"industrial software + industrial customer" geographic synergy. The Zhejiang government's specific industrial internet policy support has also attracted dozens of industrial internet startups to the greater Hangzhou area.
Chengdu Aerospace Digital Engineering Center
Chengdu hosts COMAC Sichuan R&D center, Chengdu Aircraft Industry Group (CHENGFEI), and several AVIC subsidiaries, creating high-end 3D CAD/PLM demand density. ZWSOFT has a significant development base in Chengdu. Chengdu's policy support specifically covers high-end CAD and industrial simulation tools for aviation.
Universities and Industrial Software Industry Symbiosis
Key university-industry relationships: Tsinghua CS and automation departments are major sources of EDA algorithm research talent; SJTU ME department has long-term CAE numerical methods and manufacturing systems accumulation; ZJU's deep research-commercial cooperation with SUPCON Technology represents the most important knowledge production and commercialization synergy for process control software in China; NWPU's aviation software research teams bear defense-background CAD/PLM research projects, providing extensive foundational research for domestic aerospace digitalization.
Regional Policy Competition Effects
Shanghai's "Industrial Software Innovation Development Action Plan" (2023–2025) offers specialized subsidies for industrial software companies registered and R&D-focused in Shanghai. Shenzhen's "Software Industry Base" policy provides strong office space and R&D subsidies, attracting Empyrean and Primarius to choose Shenzhen as primary operations cities. Wuhan leverages HUST and WHU engineering strengths with a "Optics Valley Industrial Software Innovation Center" focused on MES and DCS startups. Chengdu leverages CHENGFEI and CETC aerospace industry clustering with targeted support for high-end 3D CAD and industrial simulation.
Chapter 8 Eight Specialized Deep Dives
Specialty 1: EDA Digital Tool Chain — The Deepest Bottleneck
The digital EDA tool chain's five most critical links in order of national security importance: logic synthesis → physical implementation (place-and-route) → static timing analysis → functional simulation → formal verification. Of these, logic synthesis and physical implementation remain the most critical and nearly unsubstituted domestically.
Logic synthesis (transforming RTL code into gate-level netlist) uses complex optimization algorithms (technology mapping, multi-level logic optimization) requiring decades of accumulated optimization rule databases. Synopsys Design Compiler and Fusion Compiler represent 30+ years of such accumulation—current domestic tools are orders of magnitude behind in optimization quality for complex designs.
Physical implementation (place-and-route) is arguably the most computationally intensive step: placing hundreds of millions of gate cells on a chip within area constraints while routing connections to meet timing and power specifications. Global routing, detailed routing, and post-route optimization each have competing optimization objectives that must be balanced. This is a combinatorial optimization problem at massive scale where Synopsys and Cadence have built near-unassailable algorithm portfolios.
Specialty 2: Analog EDA — Where Domestic Has Made Real Progress
Analog IC design EDA tool requirements are fundamentally different from digital: emphasis on accuracy of device physics models, SPICE simulation convergence and accuracy, and careful handling of process variation. Empyrean's ALPS SPICE simulator has reached commercial usability at 28nm and above process nodes—the most important validated domestic EDA success to date. The path to broader analog EDA adoption runs through more foundry PDK certifications (especially SMIC's various nodes) and more production tape-out validations with domestic fabless customers.
Specialty 3: ERP Large Enterprise Market — The Key Battleground
The "ERP cloud war" among Yonyou, Kingdee, and SAP in large manufacturing enterprise markets is fundamentally a race to see "who helps the customer complete core system cloud migration first, locking in the next decade." SAP's China cloud strategy deploys S/4HANA on Alibaba Cloud dedicated resource pools; SAP China cloud business achieved 30%+ growth in FY2025, showing its cloud transformation in Chinese large enterprise market has made meaningful progress.
Yonyou BIP5's AI differentiation (YonGPT natural language financial assistant, AI procurement recommendation, AI sales forecasting) could constitute real competitive differentiation if it translates to verifiable business value. The BIP5 stack-up of enterprise reference cases at 10B+ RMB revenue manufacturers in FY2025 is the critical indicator to watch for genuine large enterprise substitution progress.
Kingdee Cosmic's PaaS-based SaaS architecture—allowing low-code rapid business application customization—is uniquely attractive for "high customization requirements, complex processes" customers. Kingdee Cosmic's base customer structure emphasizes finance, large SOEs, and central enterprises.
Specialty 4: Chemical DCS — How SUPCON Became Global #1
SUPCON's competitive story against Honeywell and Emerson is a masterclass in domestic industrial software competitive strategy. Starting in the 1990s from Chinese university spinoffs, SUPCON initially served smaller, simpler Chinese chemical plants that Honeywell/Emerson didn't prioritize, building reliability track records there. With each decade, SUPCON moved up-market to larger and more complex plants, carrying accumulated know-how. Today SUPCON's ECS-700 handles the largest and most complex Chinese petrochemical plants—refineries processing 10+ million tons/year—matching or exceeding foreign competition on technical capability while providing superior localization.
The lesson: domestic industrial software "win-from-below" strategy works when you choose a segment where the incumbent foreign vendor is not fully focused, build a track record, and progressively move up-market with accumulated data and experience.
Specialty 5: Steel MES — CISDI's Global Leadership Path
CISDI's global MES leadership in steel mirrors SUPCON's in chemicals: deep specialization in one industry vertical, thirty years of continuous accumulation. China is the world's largest steel producer, providing CISDI with the world's largest and most diverse steelmaking customer base for product validation and iteration. AI quality prediction embedded in CISDI's MES—predicting slab internal defects 20 minutes before tapping—is a product innovation that no non-specialist MES vendor can easily replicate.
Specialty 6: Industrial Internet Platforms — Differentiation and Consolidation
The 2017–2020 industrial internet boom created hundreds of platform startups; the 2025 landscape is in deep differentiation. The "large industrial enterprise self-built platform model" (Haier COSMOPlat, TreeRoot, XCMG Hangyun, Foxconn BEACON) has the advantage of "knowing where factory problems really are." The "cloud computing enterprise platform model" (Alibaba Cloud supET, Tencent Industrial Cloud, Huawei HiLens) has strong infrastructure and AI capabilities but weaker industry know-how. The "vertical industry specialized platform model" (CISDI for steel, SUPCON for chemicals) creates irreplicable industry barriers.
Prediction: by 2028, China's industrial internet platform market will complete major consolidation into "a few nationwide general platforms + multiple industry vertical leaders." Industry enterprise-backed platforms will achieve higher survival rates than internet-backed platforms because factory problem-solving requires process understanding, not coding capability.
Specialty 7: Manufacturing Execution System Differentiation by Industry
Semiconductor FAB MES (called WIP system) is the most technically demanding MES segment globally—simultaneously tracking hundreds of thousands of WIP lots, managing hundreds of equipment items worth millions to billions, processing thousands of process steps in sequence, and comparing real-time process measurements against design specifications. Currently in China, Applied Materials' MES system and Siemens Opcenter dominate; domestic MES vendors have virtually no presence here. The reason is clear: wafer fab MES embeds decades of semiconductor manufacturing process knowledge—requiring teams with both software engineering capability and deep semiconductor manufacturing process domain expertise, which remain extremely scarce.
Automotive MES (by VIN full-traceability model): Saisi Information (300687) is the most representative domestic vendor, having entered BYD, Geely, Chery, and SAIC. Its automotive MES technical maturity approaches Siemens Opcenter levels. But at foreign JV auto plants (VW, BMW, Mercedes in-China), SAP ME and Siemens Opcenter still dominate—because foreign OEMs need their Chinese plants to share data standards and process specifications with global headquarters, which naturally favors using the same MES vendor as headquarters.
AI integration in MES is redrawing the product definition. Domestic MES vendors have a clear advantage in AI integration speed due to deeper understanding of Chinese industrial knowledge bases and tighter partnerships with domestic AI large model vendors (Huawei Pangu, Baidu ERNIE), creating local advantage in intelligent manufacturing scenario deployment speed.
ERP Cloud War Deep Analysis
SAP's China cloud strategy achieves "local cloud" status via Alibaba Cloud dedicated pools, maintaining data security compliance while retaining deep control over customer systems. This strategy achieved 30%+ growth in SAP China cloud business in FY2025.
Yonyou BIP5's AI differentiation and Kingdee Cosmic's PaaS architecture each represent distinct competitive approaches targeting large enterprise market share from SAP/Oracle. The core battlefield is "digitalization-mature medium-to-large manufacturers" who have run SAP for 5–10+ years, have partial domestic substitution motivation, but will not accept functional regression or stability risk. Whoever first accumulates credible reference cases in this customer segment holds the key to high-end ERP substitution market entry.
Specialty 8: Industrial Internet Platform Differentiation
See Specialty 6 above for platform model differentiation. Additional dimension: the standards fragmentation problem in China's industrial internet remains severe. Different platform interface standards require custom development for cross-platform data interchange in a single factory—a significant barrier to large-scale commercialization and a core agenda item for future standards development work.
Chapter 9 Technology Evolution Frontier
AI Large Models' Deepening Penetration into Industrial Software
Industrial AI large models' most mature deployment scenarios as of 2025: equipment fault diagnosis (natural language Q&A against equipment manuals and historical fault records), process parameter optimization recommendation (AI recommending operating parameter adjustments based on historical production data correlations), quality anomaly early warning (AI predicting quality issues from process data trends before operators detect them), production planning optimization (AI automatically balancing multiple constraints like equipment availability, material inventory, and order priority).
Scenarios still at early pilot stage: autonomous control parameter adjustment (AI directly writing DCS/MES control commands—extremely cautious due to safety implications), new product process design (AI generating initial process plans—requires extensive validation), cross-plant resource scheduling optimization (AI coordinating production across multiple factories in a group company's network).
"Boundary Blurring" Across Industrial Software Categories
ERP and industrial internet platform fusion: AI enables ERP to consume real-time factory floor data for production plan optimization; industrial internet platforms can consume ERP commercial data for equipment scheduling optimization. This bidirectional fusion breaks original hierarchical separation—both ERP vendors (Yonyou, Kingdee) and industrial internet platform vendors (Haier COSMOPlat, TreeRoot) are competing for this fusion market.
CAD and CAE real-time fusion: AI-based "surrogate models" (neural network physical performance predictors) can deliver design-time simulation feedback in milliseconds vs. hours for traditional FEM, enabling engineers to iterate dozens of design variations in minutes. Ansys SimAI, Dassault 3DEXPERIENCE AI Design Assist, and domestic equivalents are all developing in this direction.
EDA and AI-Assisted Chip Design (AICD): large models applied to RTL code generation (natural language to Verilog), circuit topology optimization, and timing convergence acceleration. Google's AlphaChip research demonstrated AI surpassing human engineers in specific place-and-route objectives—this is being rapidly productized by EDA companies. For domestic EDA, this creates potential "lane change" opportunity: strong AI inference capability could reach commercially usable quality in some design scenarios without fully replicating traditional algorithm knowledge bases. Xinhuazhang (formal verification AI acceleration) is the domestic EDA startup furthest along in this direction.
5G/6G Integration with Industrial Software
5G "Ultra-Reliable Low-Latency Communication" (URLLC) extends MES digital management from wire-covered zones to factory-wide wireless coverage. Huawei's cooperation with CISDI Software on 5G private networks for steelmaking plants reduced mobile equipment real-time data acquisition latency to under 10ms, enabling mobile equipment-level real-time production control. By 2027, 5G industrial private network coverage is projected to exceed 40% of above-scale Chinese manufacturers.
6G (commercial window: 2030s) would enable factory "full perception" and "real-time digital twin" becoming technically fully feasible—every part's real-time location and every equipment sensor's data at millisecond acquisition granularity, fundamentally reshaping MES and industrial internet platform product forms.
Industrial Software and Robot Collaboration Fusion
New-generation industrial robots (from ABB, KUKA, FANUC; domestic collaborative robots from Jaka, Aubo) increasingly feature adaptive capabilities—vision-guided workpiece position compensation, force-sensing assembly quality feedback, AI-driven weld quality judgment. "Digital twin-driven robot programming" (engineers plan robot trajectories in virtual environments, then download validated programs to physical robots) is increasingly standard: Siemens Tecnomatix, ABB RobotStudio, and FANUC's ROBOGUIDE all provide this capability; domestic industrial internet platforms (TreeRoot, Haier COSMOPlat) are advancing domestic robot digital twin solutions with domestic robot vendors (Inovance, Estun, Universal Robots).
Digital Twin: From Concept to Practical Application
Digital twin reached practical scaling in 2023–2025. The truly valuable applications are specific scenarios with high "simulation + prediction" value rather than comprehensive coverage. In production line automation, digital twin enables "virtual commissioning"—validating production line operation logic virtually before physical completion, compressing commissioning cycles from months to weeks. ABB, Siemens, and FANUC all offer matching digital twin capabilities in their major product lines.
Quantum Computing's Potential Impact on Industrial Software
Large-scale matrix solving in CAE (FEA) is mathematically one of the most promising quantum computing application scenarios; EDA combinatorial optimization problems (optimal path search in place-and-route) are important potential applications for quantum annealing. If quantum computing enters practical utility in the 2030s, it could fundamentally redefine the core value of industrial software relying on precise numerical algorithms. IBM, Google, Huawei, Alibaba, and Baidu all have quantum computing R&D investments.
Chapter 10 Risk Map
Technical Catch-up Speed vs. Foreign Iteration Speed Risk
If foreign industrial software's AI integration speed (especially AI capability embedding) exceeds domestic industrial software's catch-up speed, relative capability gaps may widen rather than narrow even as domestic software continues to improve. 2024–2025 saw globally accelerated AI integration in industrial software: Synopsys embedded AI inference into main EDA tool chains; Ansys launched SimAI surrogate models; Autodesk introduced AI-assisted modeling in Fusion 360; Siemens added AI anomaly detection to Opcenter. These rapid AI capability integrations significantly increased "effective functional density" of foreign industrial software within 1–2 years.
Geopolitical Escalation: Two-Directional Impact
Geopolitics brings both "supply cut" risks (further EDA export control escalation) and structural opportunities (national procurement policy and SOE Xinchuang opening large enterprise markets previously closed). But these two effects are asynchronous in timing: supply cut risks can be sudden and urgent; domestic substitution commercialization is gradual and long-cycle. The "gap period" when foreign tools become unavailable due to geopolitical risk but domestic alternatives are not yet technically sufficient could cause irreversible production disruption.
An increasing number of core manufacturers are establishing "domestic backup solutions" for key links without fully replacing foreign industrial software, reducing passive supply cut risk exposure. This risk management practice is becoming a standard element of strategic planning in Chinese core manufacturing enterprises.
Commercial Model Innovation Uncertainty
Industrial software commercial model is undergoing structural transformation (perpetual license → subscription SaaS) globally. Chinese factory customers (especially SMEs) have significantly lower subscription acceptance than Western customers—the intuitive preference for "buy-out" software persists, and annual subscription costs feel higher than equivalent buy-out fees in annual expenditure perception. How to balance "Chinese customer habits" with "global commercial model trend" is a commercial model challenge domestic industrial software vendors must solve without ready cross-national reference answers.
Risk Quantification: Probability × Impact Matrix
High probability, high impact (address immediately): U.S. expanding EDA export controls to include cloud-based EDA tool access (~50–60% probability); if it occurs, disruption to Chinese chip design companies still relying on Synopsys/Cadence cloud services at advanced nodes.
Medium probability, high impact (track continuously): macro-economic downturn causing sharp manufacturing enterprise IT budget cuts (~30–40% probability); industrial software market growth potentially declining from current 6–8% to 0–3%.
Low probability, high impact (black swan preparation): major domestic EDA/CAD company experiencing significant security breach or data leak (~5–10% probability), severely damaging trust in domestic industrial software overall, potentially reversing major procurement decisions.
Medium probability, medium impact (periodically evaluate): core industrial software algorithm talent loss (high-end EDA/CAE algorithm engineers recruited by international competitors or moving abroad); consistent ongoing risk.
Chapter 11 2026–2030 Strategic Forecast
Market Scale Prediction
By 2030, China's industrial software market is projected to reach RMB 700–900 billion (base case), growing at 8–12% annually, driven by: continued domestic substitution volume growth; manufacturing enterprise digital transformation deepening expanding the market itself; AI application layer creating new incremental demand; and industrial software vendors expanding into SaaS overseas markets.
Segmented Track 5-Year Forecasts (Extended)
ERP track: Competition focus will shift from "feature parity" to "AI differentiation." Conservative prediction: by 2030, domestic ERP in 10B+ RMB revenue manufacturers' core finance modules will rise from ~25% to ~40–45%. Optimistic scenario: if SAP China cloud service experiences major security or compliance issues, domestic substitution could accelerate dramatically, potentially pushing core finance module market share above 60% in 3–5 years.
EDA track: The most critical five years ahead. Analog EDA: Empyrean may achieve full-process domestic EDA commercialization at 16–28nm nodes by 2027. Digital EDA: conservative estimate of 20–30% domestic substitution at mature nodes by 2030. Key milestone to track: announcement of a domestic fabless company's first production chip completed entirely with domestic EDA—projected earliest in 2027–2028.
CAD/CAE track: New energy vehicle ("new established mature process chain") is the key window for domestic 3D CAD breakthrough—BYD, Li Auto, Xpeng have far less legacy than traditional OEMs and stronger willingness to try domestic 3D CAD. CAE simulation domestic substitution may happen first in "industry specialization" scenarios (wind turbine blade fatigue life analysis, lithium battery thermal runaway simulation, seismic response analysis for buildings) rather than frontal general-purpose FEA competition.
Industrial AI large models: By 2028, expected to complete the critical crossover from "demonstration" to "quantifiable business value." The commercial model will be primarily "industrial software embedded" (industrial software vendors purchasing large model APIs and integrating into products) rather than "standalone industrial AI large model subscription," because factory procurement habits favor "buying solutions" rather than "buying AI capabilities."
International Expansion: By 2030, international revenue for domestic industrial software is projected at 8–12% of total industry revenue (base case), potentially 15–20% (optimistic). The initial overseas market follows "wherever Chinese manufacturing goes, Chinese industrial software follows."
Industrial Software Consolidation Trends
2026–2030 is projected to enter a consolidation period themed around "vertical integration." Possible M&A logic: Yonyou or Kingdee acquiring vertical industry MES vendors (gaining industry know-how and precise customers); CISDI Software deepening synergy with stone chemical industry information platforms; technology giants (Huawei, Alibaba) acquiring or taking strategic stakes in industrial software subcategory leaders (forming "cloud + software + hardware" vertical integration capability).
AI Large Models Potentially Disrupting Industrial Software Pricing
Industrial AI large models may disrupt traditional industrial software pricing logic. "Outcome-based pricing" becomes possible—factories pay for "AI-optimized production planning's reduced inventory capital tie-up," for "AI predictive maintenance's reduced unplanned downtime," for "AI quality warning's reduced defect rate." This pricing logic change—from "selling tools" to "selling outcomes"—is a disruptive challenge to industrial software companies' business model and simultaneously a major opportunity to increase ARPU.
Chinese Industrial Software's Three Development Path Options
Facing the next five years, three strategic path options exist: "Divided autonomy" (each subcategory deep-cultivating domestic market without pursuing globalization); "Breakthrough from a point" (concentrating resources in 1–2 areas with genuine global competitiveness potential, such as analog EDA and chemical DCS); "Comprehensive catch-up + selective leadership" (maintaining competitive domestic products across all major subcategories while targeting global leadership in select core tracks). Reality will likely be a mix: breakthrough concentration for EDA and high-end CAE; stable domestic market for ERP/MES/DCS; selective global leadership attempts for industrial internet platforms and industrial AI large models.
Investment Perspective on Industrial Software Track Assessment
Business model quality: high gross margins (70%+), strong customer stickiness, recurring revenue (maintenance/subscription increasingly dominant) represent excellent business model quality. Domestic leading companies (Yonyou, Kingdee, CISDI, SUPCON) are approaching global peer quality levels.
Growth space: China's domestic substitution rate continuing to rise from 50% to 70% represents significant new market increment. The market also grows as manufacturing digitalization deepens overall.
Valuation rationality: A-share industrial software company P/E multiples of 30–60x in late 2025 are not dramatically different from global peers (Autodesk ~50x, Cadence ~50x, SAP ~35x). Some (Empyrean in loss phase) are valued primarily on P/S multiples incorporating expectations of future profitability improvement.
Key risks: EDA company valuations based on expected domestic substitution market space—if digital EDA commercialization progress disappoints, valuation faces significant downward revision. ERP company valuations based on cloud migration and domestic substitution progress—if SAP/Oracle continue dominating large enterprise markets, growth expectations will disappoint.
Chapter 12 Conclusion: The Self-Sufficiency of the Digital Skeleton
Historical Perspective on China's Industrial Software Catch-up
Placing China's industrial software domestication catch-up in the broad historical context of China's manufacturing industry catch-up yields a more balanced judgment. Japan and Korea's automotive and semiconductor industries in the 1980s–1990s, under U.S. export controls and technology embargoes, underwent similar "forced catch-up" processes. Japan cultivated Nikon and Canon in lithography through "government support + long-term enterprise investment" over twenty years; Korea achieved the historic transformation from zero to global #1 in memory chips through thirty years of sustained heavy investment by Samsung and SK Hynix. These precedents demonstrate: forced competitive pressure + national strategic resource focus + extremely long-term commercial patience are the common elements enabling catch-up economies to achieve breakthroughs in complex technology industries.
Three Path Options
See the extended forecast section in Chapter 11 for the three path options and the likely mixed reality.
From Historical View to Path Forward
China's industrial software industry's current state can be described as "half the mountain climbed, half still being scaled": ERP, DCS, and steel MES have reached global-leading levels; high-end 3D CAD, top CAE solvers, and digital EDA critical links remain an arduous journey where major breakthroughs may take years or more than a decade. This duality should not be replaced by either one-sided optimism or pessimism, but understood as the complex industrial reality of "locally leading, overall catching up."
What truly deserves anticipation is not the "comprehensive domestication" timetable in policy documents but the paradigm shift quietly occurring within China's industrial software industry: from "replicating foreign products, competing for domestic market share" toward "building a new generation of industrial software products based on China manufacturing data assets and AI capability advantages." If this transformation genuinely occurs, China's industrial software opportunity will not be limited to replacing existing foreign industrial software, but to defining the standards of the next generation of industrial software—this is the true meaning of "software for a nation."
The development history of China's industrial software proves a simple truth: the core value of software lies in solving real engineering problems in specific scenarios, not in feature parity on a spec sheet. The future winners will be those who have accumulated irreplicable know-how in specific industry and process scenarios and transformed that know-how into software intelligence. This is the essence of industrial software, and the fundamental path through which China's industrial software will ultimately realize the ideal of "software for a nation."
Data Sources and Key References
This report uses publicly available sources only. Key sources for quantitative data:
Tianxiagongchang Industrial Platform: factory density, industry distribution, and factory operation data; provides industrial concentration and factory distribution analysis foundation.
Listed company financial reports: annual reports and investor relations materials for Yonyou Network (600588), Kingdee International (HK:0268), Empyrean Technology (688003), ZWSOFT (688083), CISDI Software (600845), SUPCON Technology (688777), Guanglian Da (002410), Primarius Technologies (688206), Broadway Technology (688719), and other companies.
International company public data: SAP SE FY2025 Annual Report (Frankfurt: SAP.DE), Oracle FY2025 Annual Report (NYSE: ORCL), Autodesk FY2025 Annual Report (NASDAQ: ADSK), Dassault Systèmes FY2025 Annual Report (Paris: DSY.PA), Synopsys FY2025 Annual Report (NASDAQ: SNPS), Cadence Design Systems FY2025 Annual Report (NASDAQ: CDNS), Bentley Systems FY2025 Annual Report (NASDAQ: BSY), PTC Inc. FY2025 Annual Report (NASDAQ: PTC).
Industry research institutions: CCID Consulting (赛迪顾问), CCID Research Institute, IDC China, China Academy of Information and Communications Technology (CAICT) industrial software research series.
Government and industry associations: Ministry of Industry and Information Technology (MIIT) Industrial Software Industry Development Report; AII (Alliance of Industrial Internet) platform development index; SASAC SOE digital transformation annual reports.
Academic: IEEE TCAD (Transactions on Computer-Aided Design), DAC (Design Automation Conference), ICCAD conference proceedings relevant to EDA technology.
Known Research Limitations
Some listed company segment revenues not publicly disclosed require indirect estimation; industrial software implementation service revenues often mixed with product revenues, complicating cross-company comparisons; industrial AI large model commercial data largely non-transparent; geopolitical risk impacts on industrial software are highly uncertain—all predictions assume no major geopolitical conflict escalation.
Recommended Further Reading
Annual reports of each listed company (investor relations materials are the most authoritative commercial data sources); MIIT and CCID Research Institute annual industrial software white papers; Gartner and IDC Magic Quadrant/Marketscape reports for ERP, CAD, EDA segments; AII industry standard and platform white papers; EDA Consortium market statistics data.
Data Collection Timeframe
Primary data collection period: January–June 2026. Listed company financial data is based on annual reports for fiscal years ending December 2025 (or most recent complete fiscal year); companies with different fiscal year-end dates (Autodesk ending January 31) use their most recent complete fiscal year data. All predictive conclusions are built on known information as of this report's publication date; if major policy changes, geopolitical events, or technology breakthroughs occur, predictions will need corresponding adjustment.
Industrial Software Core Glossary
EDA (Electronic Design Automation): Specialized software tool collection for integrated circuit and PCB design, indispensable infrastructure for chip design. CAD (Computer-Aided Design): Software for creating, modifying, analyzing, or optimizing designs; divided into 2D CAD (engineering drawing) and 3D CAD (solid modeling). CAE (Computer-Aided Engineering): Software using numerical computation methods to simulate and analyze engineering designs, including FEA and CFD. PLM (Product Lifecycle Management): Information system managing data and processes throughout a product's full lifecycle from concept design to retirement. MES (Manufacturing Execution System): Real-time information system connecting factory management planning layer with shop floor control layer, tracking production progress and quality data. DCS (Distributed Control System): Automation control core for process industries, using distributed controllers for real-time monitoring and control of continuous production processes. PDK (Process Design Kit): Set of data files provided by foundries to chip design companies, describing process parameters and design rules—the interface between EDA tools and manufacturing processes. ARR (Annual Recurring Revenue): The annualized recurring revenue expected based on current subscription contracts; core metric for evaluating SaaS company growth quality.
Policy Target Reference
According to MIIT's 2025–2027 industrial software special action plan, key quantitative targets by 2027 include: R&D design software (CAD/CAE/EDA/PLM) overall domestic substitution rate exceeding 40% (currently ~25–30%); production control software (ERP/MES/DCS) overall domestic substitution rate exceeding 70% (currently ~55%); domestic platforms accounting for no less than 80% of industrial internet platform domestic market (already close); formation of 3–5 globally competitive leading enterprises in global industrial software subcategories.
Industrial Software Sector Investment Assessment
From a long-term investment perspective, domestic industrial software has high gross margins (70%+), strong customer stickiness, growing recurring revenue, and growth headroom from rising domestic substitution rates (50% → 70%) and overall manufacturing digitalization deepening. A-share industrial software P/E multiples of 30–60x in late 2025 are broadly comparable to global peers. Key risks: EDA company valuations incorporating digital EDA commercialization expectations that may disappoint; ERP company valuations requiring cloud migration and large enterprise substitution progress to materialize.
The development of China's industrial software proves a simple truth: software's core value lies in solving real engineering problems in specific scenarios. Future winners will be those who accumulate irreplicable know-how in specific industrial scenarios and translate that know-how into software intelligence—the fundamental path to realizing "software for a nation."
（全文终）