ZWSOFT (688083) Deep Dive: Overdrive Kernel and ZW3D Wukong 2027 Underpin an Integrated CAD/CAE/CAM Stack, with a CAx Agent Platform as the Practical Route to Engineering AI
ZWSOFT (688083, STAR Market since 2021) builds an integrated CAD/CAE/CAM stack on its self-developed Overdrive geometric kernel and the ZW3D Wukong 2027 platform; 2025 revenue was RMB 893 million with 3D CAD at RMB 259 million, and at ZWORLD 2026 on 16 September 2026 it formally launched the CAx agent platform, while revenue for the first nine months of 2026 was RMB 580 million, up 7.80 percent, with attributable net profit of negative RMB 6.44 million.
Executive Summary
ZWSOFT (688083) is a leading Chinese vendor of design and engineering software, founded in Guangzhou in 1998 and listed on the STAR Market in March 2021. Its portfolio spans 2D CAD through ZWCAD, 3D CAD and CAM through ZW3D, CAE covering structural, fluid and electromagnetic analysis, and vertical applications including AMEP for building services and Plant for process industries. Its core asset is the proprietary Overdrive three-dimensional geometric kernel, which public sources indicate came from the 2010 acquisition of VX Corporation in the United States and has been developed in house since.
The first half of 2026 needs to be read in layers. Revenue was 360.0 million renminbi, up 7.92 percent. Net profit attributable to shareholders was negative 17.25 million renminbi against negative 52.41 million a year earlier, narrowing the loss by 35.15 million. Net profit excluding non-recurring items was negative 102 million against negative 115 million. Operating cash flow was negative 3.52 million, an improvement of 95.9 percent. Second quarter revenue was 213 million, up 2.3 percent, with attributable net profit of 21.05 million against a 4.44 million loss a year earlier, a quarterly turnaround, while adjusted profit remained negative at 32.58 million.
Geographically, China drove growth: domestic revenue was 243 million, up 16.86 percent, of which the commercial market reached 226 million, up 21.03 percent, while the education market fell 20.40 percent to 16.62 million. Overseas revenue was 117 million, down 6.86 percent, which the company attributes to disruptions at some core channel partners, Middle East geopolitics and yen depreciation. 3D CAD revenue was 104 million, up 13.4 percent, helped by new product launches.
The most important product move was ZW3D Wukong 2027, released in the first half after three years of development. It is a new-generation domestic 3D CAD, CAE and CAM integrated platform whose central upgrade is a new parametric design engine, delivering three to five times faster model regeneration and one to two times faster multi-body operations in complex equipment structural design, while addressing the associative update anomalies that earlier versions could produce when editing complex models. The company describes it as the core product foundation for entering high-end industrial design. ZW3D 365, ZWCAD 365 V2.0 with integrated AI, AMEP and Plant were released in the same period.
On AI the company proposes a three-step path: CAD plus AI features for efficiency, AI plus CAD for intelligent design software, and an AICAD native platform. The third step matters because it does not add a plugin to existing CAD but makes AI the underlying architecture, which aligns with the long-term shape of physical AI, though it remains a plan without product or revenue.
One distinction has to be kept explicit: the CAx agent platform has been launched while the AICAD native platform remains in planning. At ZWORLD 2026 on 16 September 2026 the company formally launched the CAx agent platform, an integrated CAD, CAE and CAM agent platform that connects capability units through MCP, plans more than five thousand capability units and started a co-creation programme with one hundred customers, and the third quarter report also refers to the formal launch of the ZWSOFT agent platform. The third step of the AI path, the AICAD native platform in which the software architecture is designed for AI from the ground up, still has no product and no revenue. Conflating the two would overstate the degree of AI landing, so this report labels them separately in the text and in the discrepancies.
The first nine months of 2026, disclosed October 26, 2026, show continued growth and a further narrowing of the loss. Revenue was 580 million renminbi, up 7.80 percent. Attributable net profit was negative 6.44 million against negative 40.85 million a year earlier. Adjusted net profit was negative 109 million, operating cash flow turned positive at 18.84 million, gross margin was 96.53 percent, and research investment was 299 million, down 8.76 percent and 51.51 percent of revenue. Total assets were 3.086 billion and attributable net assets 2.566 billion at the end of September 2026. The company states that Wukong drove three-dimensional design revenue close to 30 percent growth and that overseas revenue resumed growth. Against the interim report, the research expense ratio fell from 54.6 percent to 51.51 percent and operating cash flow turned positive, both signs that scale effects are beginning to appear, but adjusted profit is still a loss of above 100 million and operating profitability has not turned positive.
One framing point matters. This is a company with a real technical foundation and commercial pressure. Its geometric kernel autonomy is the scarcest asset among Chinese CAD vendors, and the Wukong 2027 parametric improvements are substantive. Yet adjusted losses remain above 100 million renminbi, research intensity is 51.51 percent of revenue, the education market is contracting, and first half administrative expense was inflated by 13.59 million of litigation-related professional fees. Against that, the first nine months delivered 7.80 percent revenue growth, a much narrower attributable loss and positive operating cash flow. Assessment must track three lines: whether 3D CAD breaks into the high end, whether CAE becomes a second growth curve, and whether the AI path moves from plan to product.
Company Overview
The Path From 2D to 3D
ZWSOFT was founded in Guangzhou in 1998 and entered with ZWCAD, a 2D product competing on AutoCAD compatibility and lower price. That path established a position in the domestic 2D market and created a lasting problem: 2D CAD has a relatively low technical barrier, price competition is intense, and the growth ceiling is visible.
The real turning point was 3D. In 2010 the company acquired VX Corporation in the United States and obtained the Overdrive geometric kernel, then built ZW3D on top of it. A geometric kernel is the heart of CAD, determining surface modeling capability, boolean robustness and large model performance, and also whether parametric history-based modeling is supportable. Very few companies worldwide develop kernels independently, with Parasolid under Siemens, CGM and ACIS under Dassault and Autodesk's in-house kernel among them, and ZWSOFT is one of the few Chinese vendors with its own.
That asset gains value under supply-cutoff risk. If a kernel depends on third-party licensing, any restriction would halt the entire 3D line. A proprietary kernel means no dependency at that link, which is the hardest evidence in its domestic substitution case.
Operating Structure in the First Half
Revenue was 360 million with core business at 359.96 million or 99.88 percent. By region, domestic was 243 million or 67.48 percent and overseas 117 million or 32.52 percent. By sales model, direct sales were 186 million with the remainder through distribution. By product, 3D CAD reached 104 million, up 13.4 percent, with 2D and CAE making up the rest.
Domestic commercial growth of 21.03 percent was the key highlight, with the company citing cooperation with benchmark customers including TCL and China Railway Shanghai Design Institute, indicating progress in both electronics manufacturing and engineering design. Education fell 20.40 percent, reflecting procurement budgets and the pace of education licensing, a small line at 16.62 million that nonetheless matters as the entry point for future talent.
The overseas decline needs disaggregation. The company cites disruptions at core channel partners, Middle East geopolitics and yen depreciation. The first two are external shocks and the third is currency. Channel problems would indicate a fragile overseas distribution system, while geopolitics and currency are cyclical. The company says it will increase support for channel partners and ZW3D promotion in the second half.
Cost Structure and Staffing
Sales, administrative and research expense growth was negative 1.2 percent, positive 24.5 percent and negative 12.0 percent respectively, totaling 459 million against 472 million a year earlier. Research expense fell 12 percent mainly because research headcount dropped by 51 to 889 and because performance assessment changes lowered compensation. Total research spending was 197 million, or 54.6 percent of revenue, still high.
The 24.5 percent administrative increase deserves separate note: litigation-related legal consulting and expert fees rose by 13.59 million. That is one-off but signals substantive legal exposure, and diligence should establish the nature of the dispute and potential liability.
A research intensity above half of revenue needs balanced reading. For a company with 360 million of revenue attacking 3D kernels and CAE, heavy spending is necessary, but the ratio also means scale effects have not appeared, and only revenue growth outpacing research spending will improve margins. The 51-person reduction is ambiguous, possibly structural optimization retaining core roles or possibly a contraction to be reversed in the second half.
Technology
Why the Geometric Kernel Is a Moat
The technical core of CAD is the kernel, responsible for geometry representation through boundary representation and NURBS curves and surfaces, geometry operations including boolean, fillet, shell and surface stitching, and topology management through parent-child relationships and update propagation among features.
The difficulty is numerical stability. Boolean operations compute intersections, unions and differences, and edge cases such as coplanar faces, tangency and tiny features accumulate floating point error, producing failed operations, missing faces or broken bodies. Industrial users tolerate such failures poorly, since a complex assembly with a thousand parts breaks the whole design process at any one failure.
Overdrive's value is more than a decade of industrial validation on routine geometry operations. Its gap against Parasolid, used in SolidWorks, NX and Solid Edge, and ACIS, used in AutoCAD among others, lies in robustness at extremes and performance on large models. That gap explains why domestic CAD struggles in high-end markets such as aerospace and automotive large assemblies: not insufficient features but insufficient stability.
The Wukong 2027 Parametric Engine
The core of ZW3D Wukong 2027 is a new parametric engine, with quantified claims of three to five times faster model regeneration and one to two times faster multi-body operations in complex equipment structural design.
Model regeneration is the central mechanism of parametric CAD. When a designer changes a dimension or feature, the software recomputes all downstream features in tree order to produce new geometry, and that performance determines interaction quality: a 30 second wait breaks the designer's train of thought while three seconds supports continuous iteration. A three to five times gain means reducing a 15 second regeneration to three to five seconds, a transition from unusable to usable.
Multi-body operations mean managing several bodies within one part file, as with weldments or multiple mold inserts, common in complex equipment design. A one to two times gain corresponds to faster selection, movement and boolean operations.
The company specifically cites resolving associative update anomalies that earlier versions could produce when editing complex models. That is a classic parametric CAD problem: circular dependencies or cross-part references can invalidate downstream features or cause geometric jumps, and solving it requires restructuring dependency graph management, which is a substantive improvement.
One qualification is needed. These are company figures without third-party benchmarks against SolidWorks or NX on the same model. Three to five times is relative to its own previous version rather than to international benchmarks, and the true gap requires head-to-head testing on the same complex assembly.
2D AI Features and the 365 Subscription
ZWCAD 365 V2.0 integrates AI and strengthens cloud collaboration. AI in 2D CAD mainly covers drawing recognition and vectorization of scanned drawings, intelligent dimensioning and layer management, and similar-design recommendations from historical drawings. The commercial value is reduced repetitive work, while the technical barrier is modest and easy for peers to follow.
The 365 subscription is a business model shift. Traditional CAD sells perpetual licenses plus maintenance while 365 sells annual subscriptions. Subscriptions make revenue predictable and raise stickiness, since stopping payment ends access, but near-term revenue recognition is lower than perpetual licenses and cash management is more demanding. Running both models indicates a gradual rather than abrupt transition.
CAE as a Second Curve
The CAE line covers structural, fluid and electromagnetic analysis positioned as design-side simulation rather than a replacement for high-fidelity solvers from Ansys. The engineering logic is clear: designers need immediate feedback on whether a wall thickness suffices or a channel will clog, without waiting days in a CAE queue.
The technical difficulty is the balance between accuracy and speed. Coarse meshes and simplified physics buy speed at the cost of certification-grade results. ZWSOFT's approach shares geometry and parameters with CAD so simulation updates with the design, which partly compensates, since users care more about how results change with a parameter than about absolute values.
What the AI Three-Step Path Means and How Far It Is
The three steps are CAD plus AI features, AI plus CAD intelligent design software, and an AICAD native platform.
The first adds AI to existing CAD through intelligent dimensioning, drawing recognition and command suggestions, and is already shipping in ZWCAD 365 V2.0. The second has AI drive design, generating geometry from natural language or intent and completing design tasks automatically, which requires AI to understand design intent and engineering constraints, a much harder step. The third rebuilds the architecture for AI rather than layering AI on traditional CAD.
The third aligns with physical AI: if future design means a person states requirements, AI generates options, simulation verifies them and the person confirms, then CAD changes from a drafting tool into a workbench for design agents. Three prerequisites must hold, namely geometric kernels callable efficiently by AI, real-time simulation feedback, and structured representation of design knowledge. None is mature.
Product and Commercial Model
Product Matrix
Four lines exist. 2D covers ZWCAD and ZWCAD 365 for general drafting and subscriptions. 3D covers ZW3D, ZW3D Wukong 2027 and ZW3D 365 across part design, assembly, molds and CAM. CAE covers structural, fluid and electromagnetic simulation for design-side checking. Vertical applications cover AMEP for building services and Plant for process industries.
The logic is All-in-One CAx, letting customers complete 2D drafting, 3D modeling, simulation checking and machining programming within one vendor stack and reducing tool switching and data conversion. For small and mid-sized manufacturers without resources for multiple software and data flows, that integration has real appeal.
Revenue Mix and Customers
Full year 2025 revenue was 893 million renminbi, of which 3D CAD was 259 million. In the first half of 2026 3D CAD reached 104 million, up 13.4 percent and about 29 percent of revenue. 2D remains the base while growing more slowly, consistent with industry patterns where 2D is mature and competitive and 3D is the increment.
The customer base spans industry leaders such as TCL and China Railway Shanghai Design Institute alongside small and mid-sized firms. Leaders provide benchmark effects and high-end scenario validation while smaller customers provide volume and cash flow, and the difficulty of this dual-track strategy lies in the cost of two sales and service systems.
Pricing and Channels
Pricing is positioned below international vendors and above domestic peers, using cost performance to take share from foreign tools and technical leadership to hold off domestic rivals. That positioning is proven in 2D against AutoCAD and still being tested in 3D.
On channels the company emphasizes integrated direct and distribution sales, with direct teams serving leaders and distribution covering smaller firms and regional markets. Domestic commercial growth of 21 percent indicates the system is working while the 6.86 percent overseas decline indicates the international distribution system is not yet robust.
Competitive Landscape
The International Trio
CAD is dominated by Autodesk with AutoCAD, Inventor and Revit, Siemens with NX, Solid Edge and Parasolid, and Dassault with CATIA, SolidWorks and CGM or ACIS. Their shared advantages are proprietary kernels, complete industry solutions and mature ecosystems of plugins, training and certification.
The gap has three layers. Functionally, routine design is largely at parity with differences at extremes such as very large assemblies, complex surfaces and high precision. On performance, Wukong 2027 narrowed the regeneration gap while large model loading and interaction still trail. The ecosystem gap is largest, since plugin markets, textbooks and certification represent decades of accumulation that domestic tools cannot replicate quickly.
Domestic Peers
Domestic CAD peers include Gstarsoft, mainly 2D with some 3D, and CAXA under Digiwin covering 2D, 3D and CAM. ZWSOFT's relative advantages are a proprietary geometric kernel, since Gstarsoft's 3D depends on third-party kernels, a complete CAD, CAE and CAM matrix, and the capital and brand of a STAR Market listing.
Gstarsoft is strong in 2D share and channels and CAXA in CAM and education, with competition concentrated in replacement projects at small and mid-sized manufacturers where price competition is intense.
Comparison in CAE
In CAE ZWSOFT faces Ansys, Siemens, Dassault and COMSOL alongside domestic players such as PERA, INTESIM and Tenfong. Its design-side positioning avoids head-on competition with high-fidelity solvers while capping CAE revenue to that segment.
Whether the positioning holds depends on whether customers pay for fast but approximate simulation. Engineering practice suggests yes, since design decisions are numerous and fast feedback outweighs precision, provided accuracy is not so poor as to mislead design.
Financing and Capital Structure
STAR Market Listing and Balance Sheet
The company listed on the STAR Market in March 2021, raising funds for 3D CAD, CAE and related research and industrialization. At the end of June 2026 total assets were 3.093 billion renminbi, down 1.9 percent from the prior year end, and attributable net assets were 2.607 billion, down 2.8 percent; at the end of September 2026 total assets were 3.086 billion and attributable net assets 2.566 billion. The balance sheet is solid with ample cash.
The Nature of the Loss
First half attributable loss was 17.25 million while adjusted loss was 102 million, with the difference coming mainly from government subsidies and other non-recurring gains. Operating profitability has therefore not turned positive, and the narrowing of the reported loss partly depends on non-recurring items.
Over the first nine months the attributable loss narrowed to 6.44 million against negative 40.85 million a year earlier, adjusted loss was 109 million, and operating cash flow turned positive at 18.84 million. The gap between the attributable and adjusted figures remains on the order of 100 million, so the near break-even reported result still depends substantially on non-recurring items, but the cash flow turn is a harder signal than the profit line.
Full year 2025 revenue was 893 million and first nine months 2026 revenue was 580 million, up 7.80 percent. Analyst expectations for 2026 are 976 million to 1.003 billion, which requires 396 million to 423 million in the fourth quarter, or roughly twenty percent or more year-on-year growth in that single quarter against the fourth quarter of 2025. The company's account of Wukong driving three-dimensional design revenue close to 30 percent growth and of resuming overseas growth supports that pace, while education contraction and the degree of overseas channel recovery remain variables.
Research Spending and Payback
A 54.6 percent research ratio in the first half of 2026 is high for industrial software; over the first nine months research investment was 299 million, down 8.76 percent and 51.51 percent of revenue, so the ratio fell as revenue grew. Against Autodesk at about 20 percent and Dassault at about 25 percent, ZWSOFT spends more than twice as intensively, explained by scale: the absolute amount is similar but the revenue base is smaller.
Payback is long. Geometric kernels, parametric engines and CAE solvers each require five to ten years of sustained investment, so investors need a long horizon while public markets assess quarterly results, a tension common to all hard technology companies.
DISCREPANCIES
1. The gap between attributable and adjusted losses is often omitted. First half attributable loss was 17.25 million while adjusted loss was 102 million, a gap of about 84.75 million from government subsidies and other non-recurring gains. Some coverage reports only a 35.15 million narrowing without noting the operating loss above 100 million. This report lists both and explains the gap. Source tier: OFFICIAL (interim report).
2. Sustainability of the quarterly turnaround. Second quarter attributable net profit was 21.05 million, a turnaround, while adjusted profit was still negative 32.58 million in the same quarter. The reported turnaround depends on non-recurring items and seasonal revenue recognition, with second quarter revenue of 213 million above first quarter revenue of about 147 million. This report labels it a book turnaround rather than an operating turnaround. Source tier: OFFICIAL.
3. Wukong 2027 efficiency gains lack external benchmarks. The company cites three to five times faster regeneration and one to two times faster multi-body operations relative to its own previous version without head-to-head testing against SolidWorks or NX on the same model. This report labels them as relative to itself rather than as comparisons with international benchmarks. Source tier: OFFICIAL (company figures) without third-party benchmarks.
4. Attribution of the overseas decline. The company attributes the decline to channel partner disruptions, Middle East geopolitics and yen depreciation, which differ in nature, with the first operational and the latter two external. No quantification by factor is disclosed. This report lists all three without single-factor attribution. Source tier: OFFICIAL (company statement).
5. The relationship between lower research expense and fewer staff. Research expense fell 12 percent, attributed to 51 fewer researchers at 889 and to performance assessment changes. This could be efficiency optimization or spending contraction. This report notes both readings and the need to observe whether headcount recovers in the second half. Source tier: OFFICIAL (data) and inference (interpretation).
6. One-off factors in administrative expense growth. Administrative expense rose 24.5 percent mainly because litigation-related legal and expert fees increased by 13.59 million. The expense is one-off but the nature of the litigation, potential liability and follow-on effects are undisclosed. This report labels it one-off and lists it as a diligence gap. Source tier: OFFICIAL (interim report).
7. Technical boundaries of the proprietary kernel claim. The company emphasizes Overdrive kernel autonomy, while public sources indicate it came from the 2010 VX Corporation acquisition, and the split between post-acquisition development and original technology is undisclosed. This report describes it as acquired and continuously developed in house rather than claiming development from scratch. Source tier: OFFICIAL crossed with public sources.
8. Stage assessment of the AI three-step path, and the distinction between the CAx agent platform and the AICAD native platform. The company proposes three steps, with the first, CAD plus AI features, already shipping, the second in development and the third at planning stage. Some interpretations treat the three as already achieved. In addition, the CAx agent platform and the AICAD native platform are two different objects: the former is the CAD, CAE and CAM integrated agent platform formally launched at ZWORLD 2026 on September 16, 2026, connecting capability units through MCP, planning 5,000-plus capability units and starting a hundred-customer co-creation program, which is a productised release; the latter refers to an AICAD native platform whose software architecture is designed for AI from the ground up, which still has neither product nor revenue. Ruling: CAx has shipped and AICAD remains at planning stage. This report labels the two separately and does not treat the CAx launch as evidence that AICAD has landed. Source tier: OFFICIAL (strategy statement and ZWORLD 2026 launch) with actual progress differing.
9. Interim and third-quarter reporting periods coexist. The interim figures are first-half revenue of 360 million renminbi up 7.92 percent, attributable net profit of negative 17.25 million, adjusted net profit of negative 102 million, operating cash flow of negative 3.52 million and research spending at 54.6 percent of revenue; the third-quarter figures are nine-month revenue of 580 million renminbi up 7.80 percent, attributable net profit of negative 6.44 million, adjusted net profit of negative 109 million, operating cash flow of positive 18.84 million and research spending at 51.51 percent of revenue. Both sets come from periodic reports and both are true, but they cover different periods. This report uses the third-quarter figures when describing current operating conditions and keeps the interim figures as a half-year factual record; statements about narrowing losses and the cash flow turning positive follow the third-quarter basis. Source tier: OFFICIAL (interim report) and OFFICIAL (third-quarter report) in parallel, with the third quarter as the current basis.
Risks and Diligence Gaps
First, persistence of adjusted losses. Adjusted losses have continued for two years across 2025 and the first nine months of 2026, with adjusted net profit still negative 109 million in the third quarter, indicating the operating profit model is not yet proven. Although the attributable loss has narrowed to 6.44 million and operating cash flow has turned positive at 18.84 million, neither substitutes for an adjusted turnaround. Failing an adjusted turnaround in the fourth quarter of 2026, market patience with the scale-effect story would erode.
Second, overseas uncertainty. Overseas revenue is 32.5 percent of the total and typically carries higher prices, so channel problems, geopolitics and currency together could keep overseas growth under pressure.
Third, difficulty of high-end entry. Wukong 2027 improved parametric performance while gaps against NX and CATIA in very large assemblies and complex surfaces remain. Aviation and automotive mainstream design flows demand extreme stability, and domestic tool replacement requires very long validation.
Fourth, litigation risk. Litigation-related intermediary fees of 13.59 million in the first half indicate substantive legal disputes whose nature and potential liability are undisclosed, with possible effects on future profit.
Fifth, the CAE ceiling. The design-side positioning avoids head-on competition at the cost of a limited market, and without CAE becoming a second growth curve the company remains dependent on CAD alone.
Sixth, uncertain returns on AI investment. The three-step path requires sustained spending while willingness to pay for AI features is unvalidated. The CAx agent platform was launched in September 2026 with a hundred-customer co-creation program, but the company has disclosed no revenue, subscription count or paying customer count for it, so the level of landing can currently be described only as a shipped platform with 5,000-plus planned capability units. If customers treat AI as expected but not worth a premium, investment becomes cost rather than revenue. Note also that the CAx platform and the still-planned AICAD native platform are two different objects, and the launch of the former is not evidence that the latter has landed.
Seventh, the long-term effect of education contraction. A 20.4 percent decline matters little financially but matters strategically, since students who learn a tool tend to use it professionally, so this line deserves strategic rather than purely profit-based assessment.
Composite Assessment
Technology scores 7.5. The Overdrive kernel is the scarcest asset among Chinese CAD vendors, the Wukong 2027 parametric upgrade with three to five times regeneration and one to two times multi-body gains is substantive, and integration of CAD, CAE and CAM forms a complete toolchain. Deductions for robustness at extremes and large model performance still trailing international benchmarks, limited CAE solver depth, and early-stage AI.
Business scores 6.5. Domestic commercial growth of 21 percent, 3D CAD growth of 13.4 percent and cooperation with benchmark customers such as TCL and China Railway Shanghai Design Institute show acceptance in real scenarios, and the All-in-One CAx strategy suits small and mid-sized manufacturers, with the CAx agent platform formally launched on September 16, 2026 and a hundred-customer co-creation program started; three-dimensional design revenue in the first nine months rose close to 30 percent and overseas returned to growth. Deductions for adjusted losses still above 100 million, a 20.4 percent education contraction and no disclosed revenue or customer count for the CAx platform.
Finance scores 6.5. Total assets of 3.086 billion and attributable net assets of 2.566 billion as of the end of September 2026 indicate a solid balance sheet; nine-month revenue of 580 million renminbi was up 7.80 percent, the attributable loss narrowed to 6.44 million, operating cash flow turned positive at 18.84 million, and the research spending ratio fell from 54.6 percent to 51.51 percent as scale effects begin to appear. Deductions for adjusted losses still at 109 million, an 8.76 percent year-on-year decline in absolute research spending, administrative expense burdened by litigation and growth dependent on the domestic market alone.
Product scores 7.5. ZWCAD, ZW3D, CAE, AMEP and Plant cover the main links from design to machining, the 365 subscription opens a business model transition, and Wukong 2027 targets high-end complex equipment needs. Deductions for 2D commoditization, insufficient CAE accuracy and depth, and AI features not yet differentiated.
Market scores 7.5. Domestic substitution policy is explicit, China leads the world in manufacturing scale, rising software legalization creates increment, and industrial software is a policy priority. Deductions for the deep ecosystem barriers of the international trio, price competition among domestic peers, and unrevived paying capacity among downstream small firms.
Overall 7.0. This is a company with a solid foundation and slow conversion: kernel autonomy is its least replaceable asset and Wukong 2027 shows it can keep narrowing the gap with international benchmarks, and nine-month revenue growth of 7.80 percent with narrowing losses and positive operating cash flow is a positive signal on delivery pace; yet adjusted losses still above 100 million, the education contraction and a third AI step still at planning stage show it remains far from being a world-class industrial software company. The judgment on physical AI is that its physical core in geometry, parametric design and CAE is a real engineering asset, that the CAx agent platform has shipped and connects capability units through MCP, and that the AICAD native platform direction matches the long-term shape of design agent workbenches, while the distance from adding AI to CAD to rewriting CAD for AI is one full product generation.
References
- ZWSOFT 2026 interim report, disclosed August 14, 2026: revenue 360.04 million renminbi up 7.92 percent, core business revenue 359.96 million or 99.88 percent, attributable net profit negative 17.25 million against negative 52.41 million a year earlier with the loss narrowing 35.15 million, adjusted net profit negative 102 million against negative 115 million, operating cash flow negative 3.52 million up 95.9 percent, basic earnings per share negative 0.1017; total assets 3.093 billion down 1.9 percent from the prior year end and attributable net assets 2.607 billion down 2.8 percent
- Regional and product data from the same report: domestic 243 million or 67.48 percent up 16.86 percent, including commercial 226 million or 62.85 percent up 21.03 percent and education 16.62 million or 4.62 percent down 20.40 percent; overseas 117 million or 32.52 percent down 6.86 percent; direct sales 186 million; 3D CAD revenue 104 million up 13.4 percent
- Cost and staffing data from the same report: sales, administrative and research expense growth of negative 1.2, positive 24.5 and negative 12.0 percent, totaling 459 million against 472 million a year earlier; research spending 197 million at 54.6 percent of revenue; 889 research staff, 51 fewer than a year earlier; administrative growth mainly from 13.59 million of litigation-related legal and expert fees
- New product information from the same report: ZW3D Wukong 2027 as a new-generation domestic 3D CAD, CAE and CAM integrated platform developed over three years, with a new parametric design engine delivering three to five times faster model regeneration and one to two times faster multi-body operations in complex equipment structural design while addressing associative update anomalies in complex model editing; ZW3D 365, ZWCAD 365 V2.0 with AI and cloud collaboration, AMEP and Plant released in the same period
- Second quarter 2026 data: revenue 213 million up 2.3 percent, attributable net profit 21.05 million against a 4.44 million loss a year earlier, adjusted net profit negative 32.58 million against negative 40.80 million, earnings per share 0.1241
- Broker commentary, August 2026: domestic commercial recovery with short-term overseas pressure; forecasts of 2026 to 2028 revenue at 976, 1,091 and 1,231 million renminbi in one estimate and 1,003, 1,139 and 1,299 million in another; attributable net profit of 108, 150 and 204 million or 67, 151 and 217 million or 130, 191 and 228 million across three estimates; risks including core technology licensing cutoff, intensifying competition, product development shortfalls and litigation
- ZWSOFT AI strategy as cited in broker commentary from the interim report: a three-step path of CAD plus AI features, AI plus CAD intelligent design software and an AICAD native platform
- ZWSOFT 2026 third-quarter report, disclosed October 26, 2026: revenue for the first nine months of 580 million renminbi up 7.80 percent, attributable net profit negative 6.44 million against negative 40.85 million a year earlier, adjusted net profit negative 109 million, operating cash flow positive 18.84 million turning to net inflow, comprehensive gross margin 96.53 percent, research spending 299 million renminbi down 8.76 percent and at 51.51 percent of revenue; total assets 3.086 billion and attributable net assets 2.566 billion as of the end of September 2026; company figures cite three-dimensional design revenue close to 30 percent growth driven by Wukong, resumed overseas growth and the formal launch of the ZWSOFT agent platform
- ZWSOFT ZWORLD 2026 conference, September 16, 2026: formal launch of the CAx agent platform integrating CAD, CAE and CAM, connecting capability units through MCP, planning 5,000-plus capability units and starting a hundred-customer co-creation program
- Company background from public sources: founded in Guangzhou in 1998 and listed on the STAR Market in March 2021; acquired VX Corporation in the United States in 2010 to obtain the Overdrive three-dimensional geometric kernel; products including ZWCAD, ZW3D, ZWSOFT CAE, AMEP and Plant
- 2025 operating data from public sources: full year revenue 893 million renminbi with 3D CAD at 259 million
- Competitive landscape from public industry sources: CAD dominated by Autodesk with AutoCAD, Inventor and Revit, Siemens with NX, Solid Edge and Parasolid, and Dassault with CATIA, SolidWorks and CGM or ACIS; domestic peers including Gstarsoft with mainly 2D and CAXA under Digiwin covering 2D, 3D and CAM
- Industry background from public sources: geometric kernels handle representation through boundary representation and NURBS, operations including boolean, fillet, shell and surface stitching, and topology management of feature parent-child relationships and update propagation; boolean numerical stability under coplanar, tangent and tiny-feature conditions is the core robustness difficulty; Parasolid and ACIS are the main commercial kernels
- Industry background from public sources: parametric model regeneration recomputes downstream geometry in feature tree order after a dimension or feature change, and its performance determines interaction quality; multi-body operations manage several bodies in one file such as weldments and mold inserts; design-side simulation trades coarse meshes and simplified physics for speed and suits trend judgment rather than certification
This report is compiled from public sources for informational purposes only and is not investment advice. Financial figures follow company periodic reports; product performance and market share claims are company-reported without third-party benchmarking, and related judgments are labeled as uncertain.
Related papers
Related tools
Related cases
- IASC-ASCE Structural Health Monitoring Benchmark and Physics-Informed Damage Identification
- TopoDiff: Diffusion-Model Topology Optimization and Its Evaluation Metrics
- DrivAerML High-Fidelity Road-Car Aerodynamics Dataset and Surrogates
- AirfRANS High-Fidelity Airfoil Surrogate Benchmark Case
- Physics-Informed CNN for Battery Pack Temperature Fields
Related reports66
- 51World Deep Dive: A Listed Digital Twin Company Where 51Sim Becomes the First Growth Engine While 51Earth Nearly Vanishes, and Two Irreconcilable Financing Naming Conventions Must Both Be Kept
- AMI Labs: A USD 1.03B Seed Bet on World Models, LeCun's Non-Generative Path and Its Blank Scorecard
- Anshi Pacific (PERA) Deep Dive: A Beijing Domestic-CAE Leader Founded in 2003 Whose PERA SIM General Simulation, AIROM Reduced-Order Models, AgentX Multi-Agent and Jingzhi iGPT/Whale Span from Industrial Simulation to Physics AI
- AspenTech Deep Dive: The Process-Simulation Incumbent Acquired by Emerson and Its Hybrid Industrial-AI Route
- Autodesk Deep Dive: Neural CAD Foundation Models and a $200M World Labs Bet on Physical-World AI, with NAVASTO Adding CFD Surrogate Capability
- Bentley Systems Deep Dive: iTwin Digital Twin and Generative AI Empower Full-Lifecycle Infrastructure Engineering
- BeyondMath Deep Dive: A Foundation Physics AI on the Neural-Operator Route, a $19M Consortium Budget, and Three Misread Facts
- Cadence Deep Dive: System-Level Multiphysics EDA and the Millennium M2000 Supercomputer Anchor Physical AI, with AuraStack Agentic AI and Fidelity GPU CFD as Key Levers
- COMSOL Deep Dive: Four Decades of Multiphysics Solvers, Entering Physical AI via DNN Surrogate-Model Training and All-Physics GPU Acceleration
- Dassault Systèmes Deep Dive: 3DEXPERIENCE and SIMULIA Virtual Twin Physics Behavior Anchor Industrial AI, with LEO Virtual Companion and GPU Explicit Dynamics as Physical-AI Levers
- DP Technology: From the Gordon Bell Prize to National-Level Unicorn Status and the Door of a Hong Kong Listing, the AI-for-Science Leader's Technical Delivery and Commercial Blank
- Emmi AI: From Linz Spin-out to a EUR 330M Mistral Acquisition in 17 Months
- Fainite Deep Dive: A Zurich Physics-AI Simulation Startup Whose Caltech, ETH Zurich, Cambridge and Google Team Uses Physics-Aware Deep Learning to Cut Simulation from Tens of Hours to Seconds, With Only Venture Kick Grants and No Priced Round
- Fysics AI: The PhysX Founder's Second Act, a Domestic Differentiable Physics Engine and Its Ecosystem Play
- Frontop (301313) Deep Dive: From Digital Exhibition Halls to a Physical-AI Data Foundation - FTE Engine, FunSim Simulation and the Gap Between Positioning and Revenue
- Godela Deep Dive: A YC X25 Physics-AI Simulation Startup Whose Ex-Apple, Google and Intel Team Uses Physics-Constrained Surrogate Models to Cut Simulation from Days to Seconds, at Seed Stage with Pilot-Only Access
- HollyWave (688682) Deep Dive: RDSim AI Electromagnetic Simulation and Engineering Agents Close the Simulation-Measurement-Product Loop
- IAT (Alt) Deep Dive: China's First Listed Independent Automotive Design Company Pivots to AI-native R&D with the Yuanfang, Yufeng and Taiyi 3.0 Product System and an NVIDIA Omniverse Partnership, Reporting RMB 975M Revenue and a Widened RMB 250M Net Loss in 2025
- Inductiva Deep Dive: A Closed Case in Open-Solver Orchestration, Eight Premises That Must Be Corrected, and Where Co-Founder and CTO Hugo Penedones Went Next
- Intesim (INTESIM) Deep Dive: A Home-grown Chinese CAE Leader Founded in 2009 with Self-developed Solvers for Structure, Fluid, Electromagnetics, Thermal and Acoustics and Multiphysics Coupling, a ~RMB 100M B Round from CNII and Shenzhen Capital, and 2025 Revenue and Bookings Both above RMB 100M
- iSoftStone (301236) Deep Dive: A ChiNext-Listed IT Services Giant Whose TianShu Platform Builds a Three-Layer Physical-AI Stack with Self-Developed TwinHub Digital Twin, FSim Industrial Simulation, CAE Multiphysics and Phyxis Embodied Control
- Jua Deep Dive: The Earth Physics Transformer as a Large Physics Model for Energy Trading
- Keysight (with ESI Group) Deep Dive: Virtual Prototyping and Hybrid AI Bridge Measurement and Simulation, with Natural-Language Assistants in EDA
- Lightwheel Deep Dive: A Physical AI Data and Evaluation Infrastructure Startup Founded by a Former NVIDIA Simulation Lead, Seven Rounds Totalling Roughly CNY 2.65 Billion at a Valuation Above USD 2 Billion, Seats on Both the Newton and EgoVerse International Standards Committees, With Two Incompatible Accounts of Q1 Orders
- Luminary Cloud: The Physics AI Model Factory Backed by NVentures
- MathWorks Deep Dive: MATLAB/Simulink and Simscape Multiphysics Power Physical AI, with R2026a Copilots and an Agentic Toolkit for Edge Intelligence
- Maniformer (Mifeng) Deep Dive: An Agibot-Spun Physical-AI Data Infrastructure Firm Whose Bodyless MEgo Capture Hardware Is in Mass Production and Which Closed Three Rounds in Half a Year Led by China Telecom
- Mistral AI: How Physics AI Moved From Language Models to EUV Lithography and Nuclear Power
- Monolith AI: The British Engineering-AI Firm That Cut Physical Test Counts Instead of Accelerating Simulation, Now Inside CoreWeave
- Monumo Deep Dive: A Cambridge Deeptech Company Redesigning Electric Motors with the Anser AI Engine, Tens of Millions of Multiphysics Simulations per Day, GBP 10.5M Raised in Total, and a Series A Being Raised
- NAVASTO: Moving Solver Insight Onto the Designer's Desk, and How Autodesk Bought That Entry Point
- Navier: Handing CFD Workflows to Agents, a Seed-Stage Bet on AI for Hardware Engineering
- Neural Concept: An EPFL Spin-out Betting on Neural Surrogates, with $138M Raised
- NP Co. (NP Company) Deep Dive: A 2025-founded Paris Inria-born AI-native simulation startup that raised a EUR 6M pre-seed led by Partech with Mistral AI co-founders as angels, claiming a pre-trained physics foundation model at 1,000x speedup — most claims unverified
- NVIDIA Deep Dive: Building the Full-Stack Physical-AI Infrastructure with PhysicsNeMo, Warp, Cosmos World Foundation Models and Omniverse
- Physical Intelligence Deep Dive: A $2.1B Bet on the GPT Moment for Robotics, With pi0.7 Self-Reporting 60-80 Percent Success on Unseen Tasks and Neither Pricing nor Revenue Disclosed for the Pi API
- Physical Superintelligence Deep Dive: A $58M Seed Bet on Virtual Physicists, With Emmy Targeting AI Data Center Multiphysics Optimization as Its First Commercial Proof Point
- PhysicsX: Pretrained Geometry and Physics for Industrial Engineering at a $2.4B Mark
- Prometheus: A $41B Mark, $18.2B Raised, and Not Yet a Product
- PTC Deep Dive: Creo/Onshape and the ThingWorx Digital Thread Power Physical AI, with NVIDIA Isaac Sim Bridging Robotics Design to Simulation
- Qianjue Technology Deep Dive: How a Body-Agnostic Embodied Brain Pairing the Predictive World Model Polibrain with Brain-Inspired Partitioning and Polynomial Representation Reached a Hundred Thousand Cleaning and Service Robots
- Quanscient Deep Dive: Cloud-Native and Quantum-Powered Multiphysics Simulation as the Data Engine for AI-Native Hardware Engineering
- Rescale Deep Dive: An HPC Operating System and Desktop Engineering Workbench in the Cloud, a $115M Series D, and Five Misread Facts
- Schrödinger Deep Dive: Physics-Based FEP+ Simulation and Machine Learning Power Molecular Discovery, with the 2026 Bunsen Agentic Co-Scientist
- Secondmind Deep Dive: A Cambridge Engineering-AI Firm Whose Bayesian Active Learning Cuts CAE Simulation and Test-Bench Runs by up to 90%, with Roughly USD 73M Raised and Mazda Both Customer and Investor
- Shichuang Technology Deep Dive: SupreCAST Cloud-Native Material-Processing Simulation and Industrial Agents Advance the CAE+AI Strategy
- ShonCloud (Shunyun) Deep Dive: Meshless Particle-Method CFD for EV and Nuclear Fluid Multiphysics
- Shuzhi Maijin Deep Dive: A Shenzhen Fully Self-Developed SNN-plus-LBM CFD Startup Claiming Second-Level Results on Hundreds-of-Millions of Grids on Consumer Hardware, With a Tens-of-Millions-RMB Angel Round From a Ministry-of-Finance-Affiliated Fund and an FPGA Migration Plan
- Siemens (incl. Altair) Deep Dive: The 2025 Altair Acquisition Unites Simcenter and Altair, with PhysicsAI Geometric Deep Learning and PhysicsAI Generate as the Physical-AI Core
- Silurian Deep Dive: A Generative Forecasting Transformer Earth-System Foundation Model for Weather Risk on Energy and Infrastructure Assets
- SimScale Deep Dive: A Cloud CAE Platform Built on Core-Hour Subscriptions, EUR 52M of Equity Kept Separate from EUR 25M of Debt, and Five Misread Facts
- Skild AI Deep Dive: A General Robot Foundation Model Incubated at CMU's Robotics Institute, With a USD 1.4B Series C at a USD 14B Valuation, a Conflicting Series B Size, and an ARR Figure That Is a Run-Rate Rather Than Revenue
- Songying Deep Dive: A Huawei-Cloud-Alumni Startup Building a Domestic Physical AI Operating System, With ORCA OS Positioning Against NVIDIA Omniverse, Three Incompatible Financing Narratives, and a Verifiable Third-Party Anchor in a CNY 20 Million Minority Investment
- Sudo Tech Deep Dive: Zero-Real-Data Simulation Training, the Fully In-House Sudo R1 Stack, and an Atomic-Skill Commercialisation Strategy from a Startup Valued Above CNY 20 Billion in 17 Months
- Suochen Technology: The Pain of Becoming a Physics-AI Platform from a CAE Vendor
- Synopsys (Ansys) Deep Dive: The 2025 Ansys Acquisition Turns the EDA Leader into a Silicon-to-Systems Engineering Powerhouse, with Ansys SimAI and the NVIDIA Physical-AI Partnership as the Core Physics-AI Levers
- TenFong Deep Dive: An Academician-Led Chinese Industrial Simulation Software Vendor Whose Official Timeline and Database Records Use Incompatible Round Names, With an Official Financing PDF and an Investment Database Both on File
- Tianfu (Nanjing Tianfu Software) Deep Dive: A 2011-founded Chinese Intelligent-CAE Veteran That Closed a Several-Hundred-Million-RMB Series C, Pivoted to an Industrial World Model in 2024, With Unreconcilable Financing and Valuation Records and No Disclosed Revenue
- Tongyuan Soft Control Deep Dive: MWORKS (Syslab/Sysplorer) as a Domestic MATLAB/Simulink Alternative, with MWORKS AI and the MOGICK Engineering Agent Advancing Home-Grown Physical AI
- Vinci AI: Turning Physics Simulation from a Checkpoint into a Continuous Constraint
- Westlake Robotics Deep Dive: Westlake University's First Robotics Spin-out Builds a Dual-Pretrain Brain-and-Cerebellum Architecture with WR1 Whole-Body Foundation Model and GAE Behaviour Expert
- World Labs: Fei-Fei Li's Spatial-Intelligence Unicorn, Acquired by AMD for $8.2B
- XVERSE Deep Dive: A Simulation-World Route to Physical AI Built on the XVERSE Model Family, an Edge-Cloud 3D Engine and the Physics-Based MotionGen Model
- SimWE: From In-House Solvers to a Physical-AI Engine, the Domestic CAE Leader and What Comes Next
- Yunjing Zhifang Deep Dive: A Peking-University-Incubated Cloud-Native GPU CAE Company with the Simversus Platform and the YJ-DORGP Small-Sample Intelligent Solver, Reporting 275 Percent Order Growth, 160 Percent Revenue Growth and Breakeven in 2025
- Zhongke Diwiji (Five Ages) Deep Dive: A CAS-Institute-of-Automation and Tsinghua Spin-out Whose FAM Embodied Foundation Model Uses Action Optical Flow for Cross-Embodiment Generalisation, Reaching 97% Success from 3 to 5 Demos, with EraBrain and Jiling Robots Serving Sinopec, Leapmotor and Midea
This report is compiled from public sources. It is for reference only and does not constitute investment advice. Data is current as of the publication date; please refer to the company’s official filings.