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
Frontop (301313, ChiNext since 2022) is pivoting from digital exhibition halls to a physical-AI data foundation built on its self-developed FTE engine, the FunSim simulation platform and a four-tier embodied-data stack; AI 3D revenue reached RMB 175 million in 2025 (+255%), but revenue for the first nine months of 2026 was RMB 338 million, down 19.45 percent, with an attributable net loss of RMB 123 million and cash flow pressure still widening.
Executive Summary
Frontop, formally Guangzhou Frontop Digital Creative Technology, listed as 301313 on the ChiNext board, started in digital exhibition halls and is transforming into what it calls a physical AI data foundation and solutions provider. It operates three lines: integrated digital services for exhibitions and showrooms, AI 3D digital twin products and solutions, and 3D visualization products and services, with a fourth line in cultural IP operations added in 2026.
The first half of 2026 is a textbook transformation balance sheet. Revenue was 231 million renminbi, down 27.80 percent. Net profit attributable to shareholders was negative 77.25 million against negative 46.86 million a year earlier, widening losses by 64.86 percent. Adjusted net profit was negative 79.74 million against negative 47.69 million. Operating cash flow was negative 57.97 million against positive 3.44 million, a swing from inflow to outflow. Basic earnings per share were negative 0.75 and the weighted average return on net assets was negative 12.25 percent. Second quarter revenue was 89.07 million, down 52.18 percent year over year and 37.13 percent sequentially, with attributable net profit of negative 52.12 million.
The third quarter report of October 27, 2026 steepened that decline. Revenue for the first nine months was 338 million renminbi, down 19.45 percent; attributable net profit was negative 123 million against negative 96.41 million a year earlier, so losses widened further; adjusted net profit was negative 125 million; operating cash flow was negative 95.12 million. Third quarter revenue alone was 107 million, up 7.20 percent, the first positive single-quarter revenue growth since the transformation began, though the adjusted result was still a loss. Gross margin for the first nine months was 27.78 percent, up 3.75 points, and research expense grew 77.12 percent. On the liability side short-term borrowings reached 152 million, up 108.48 percent, alongside long-term borrowings of 74.64 million for an acquisition loan, and the company added a 10 percent stake in Guangdong embodied intelligence technology for 28.62 million renminbi.
Structurally there is a countervailing signal. AI 3D digital twin products and solutions reached 56.65 million renminbi, up 23.73 percent, at a 36.32 percent gross margin, making it the leading growth driver, while integrated digital services fell 45.54 percent to 129 million and drove the overall decline. Gross margin rose to 31.49 percent, up 4.87 points, while net margin fell to negative 36.66 percent, down 22 points.
The interim report states the three causes of the wider loss. First, new exhibition projects at government and enterprise customers contracted, reducing orders and slowing acceptance. Second, research expense reached 31.90 million, up 87.19 percent, directed at data business, software toolchain, an embodied intelligence data simulation platform, AI 3D digital twins and integrated solutions. Third, credit and asset impairment totaled 39.94 million, up 16.14 million year over year, reflecting high receivables and longer settlement cycles, with 50 percent provisions for receivables aged two to three years and 100 percent beyond three years.
In physical AI terms the story is using digital twin visualization capability to enter the embodied intelligence data business. Embodied models need large volumes of annotated three-dimensional scene and motion data, and Frontop spent more than a decade building scene modeling, real-time rendering and interaction design capability. Redirecting that capability toward embodied data collection and simulation is logically coherent, with the FTE engine and the FunSim simulation platform as the technical carriers.
One framing point matters. This is a company with a sound transformation direction and real financial pressure. AI 3D growing 23.73 percent at a 36.32 percent margin shows new demand and pricing power, yet revenue for the first nine months down 19.45 percent, losses up to 123 million and operating cash flow negative 95.12 million show the legacy business is contracting faster than the new business replaces it. The key questions are not about the concept but three specifics: whether AI 3D and data revenue can offset the legacy decline in 2027, whether receivable impairment has bottomed, and whether the embodied data line can move from selling projects to selling data and services.
Company Overview
From Exhibition Halls to 3D Visualization
Frontop was founded in the 2000s and is headquartered in Guangzhou, starting with creative design and delivery of digital exhibition halls and showrooms. The core of that business is content plus space plus multimedia technology, with government and large enterprise customers. It listed on ChiNext under 301313 and raised funds to expand digital creative and 3D visualization capability.
The line is project-based, with single projects from several million to tens of millions of renminbi, delivery over months and revenue recognized on acceptance. Project business brings large ticket sizes and acceptable margins at the cost of volatile revenue, long receivables cycles and high sensitivity to customer budgets, and the first half of 2026 concentrated all three weaknesses.
Business Mix and Transformation Path
First half revenue of 231 million breaks down by product into integrated digital services at 129 million or 56.06 percent, down 45.54 percent; AI 3D digital twin products and solutions at 56.65 million or 24.55 percent, up 23.73 percent; 3D visualization products and services at 28.43 million or 12.32 percent; and other at 16.30 million or 7.06 percent including about 16 million of cultural IP operations.
By region, South China contributed 125 million or 54.28 percent, North China including the northeast 47.37 million or 20.53 percent, East China 33.31 million or 14.44 percent and other regions 24.81 million or 10.75 percent. South China above half indicates a customer base concentrated in Pearl River Delta government and exhibition projects with limited national reach.
The transformation path is explicit in the interim report: move from a traditional digital creative company to a physical AI data foundation and solutions provider, cultivating four growth poles in physical AI 3D digital twin products and solutions, embodied application solutions, data business, and consumer cultural IP operations.
Direction of Research Spending
Research expense of 31.90 million, up 87.19 percent against a 27.80 percent revenue decline, shows the transformation is funded rather than narrated. Spending goes to data business covering embodied data collection, annotation and synthetic generation; software toolchain around the FTE engine; the FunSim embodied intelligence data simulation platform; and AI 3D digital twins.
Research intensity rose from about 9.7 percent to 13.8 percent of revenue. That is high for digital creative and low for AI software, reflecting an in-between state: no longer a pure project company with low research intensity, not yet a pure product company with high intensity.
Technology
The Digital Twin Foundation
The technical assets fall into three layers. The base is three-dimensional content production covering scene modeling, materials, lighting and rendering. The middle is a real-time engine, where FTE renders and enables interaction with 3D scenes in browsers or on large displays, which is central to exhibition halls because visitors interact through touch screens, video walls and augmented reality. The top is the digital twin, mapping physical equipment, production lines and campuses into interactive 3D models connected to live data from sensors, video streams and business systems.
These three layers were refined across exhibition projects, producing engineering delivery processes covering modeling standards, material libraries, interaction templates and performance optimization. That is genuine accumulation and the starting point for embodied data.
Why Digital Twin Capability Can Transfer to Embodied Data
Embodied model training draws on three sources: teleoperated real-robot collection, simulation generation and video or motion capture. All three need three-dimensional scenes, object poses, motion trajectories and semantic annotation.
The overlap with Frontop's capability is the scene layer, since it can build high-fidelity models of factories, warehouses and showrooms including geometry, materials and lighting. In synthetic data generation that is the most basic and time-consuming step, because scanning and reconstruction or manual modeling are both expensive.
Beyond the overlap there are large gaps. Embodied data needs interactive physical simulation covering mass, friction, collision response and joint constraints, plus the outcome of robot-object interaction. That requires a physics engine such as MuJoCo, Isaac Sim or PyBullet, whereas the FTE engine is a rendering engine. FunSim is described as an embodied intelligence data simulation platform, indicating an attempt to fill that layer, but public information does not say whether the physics engine is proprietary or open source, what physical fidelity it achieves, or at what data scale.
The Four Categories of Embodied Data
The company refers to four categories of embodied data. Following common industry classification, these would be scene data, object data including models and poses, motion data covering robot trajectories and manipulation, and task data covering instructions and outcomes. The taxonomy is reasonable, and the substance lies in collection method and scale.
Three industry approaches exist. Real-robot collection through teleoperation or motion capture delivers the highest quality at the highest cost. Simulation generation is cheap but faces the sim-to-real gap. Video reconstruction is cheapest with limited precision. Frontop is more likely to lean on simulation given its strength in 3D scenes, which means confronting sim-to-real directly, since policies trained in simulation often fail on real robots.
Division of Labor Between FTE and FunSim
FTE is the rendering and interaction base while FunSim is the embodied data simulation platform, and the division is technically sensible: rendering ensures it looks right and simulation ensures it behaves right.
The two stacks differ greatly. Rendering rests on computer graphics including rasterization, ray tracing and material models, while physics simulation rests on numerical computation including rigid body dynamics, collision detection, contact solving and soft body or fluid behavior. A company starting from rendering must build a numerical computation team, which cannot be achieved by converting a rendering team, and whether such a team exists cannot be judged from public information.
Technical Shortfalls
First, autonomy of the physics engine is unknown. If FunSim builds on open-source engines, core capability sits outside the company; if self-developed, it requires sustained investment.
Second, the sim-to-real gap. Data generated in simulation and used to train real robots amplifies error, and the common industry practice of pretraining in simulation plus fine-tuning on real data has not been disclosed.
Third, no disclosure of data scale. Embodied data value scales strongly with volume, whether millions of trajectories or hundreds of thousands, and the company has not disclosed accumulated volume.
Fourth, ambiguity in the AI content of AI 3D. Whether AI is used for automated reconstruction from images or point clouds, for semantic labeling through scene understanding, or for generative content creation changes the technical substance greatly.
Product and Commercial Model
Commercial Roles of the Three Lines
Integrated digital services are the cash engine at 129 million in the first half, down 45.54 percent, and project businesses bear the brunt when customer budgets tighten.
AI 3D digital twin products and solutions are the strategic direction at 56.65 million, up 23.73 percent, at a 36.32 percent gross margin. Customers likely overlap with exhibition clients in government, parks and factories, while the product form is closer to software plus data plus services, selling a twin system in continuous use rather than a one-off exhibit.
3D visualization products and services contributed 28.43 million or 12.32 percent as a continuation of legacy capability. Cultural IP operations added about 16 million, with cases including the Yongle Palace digital art exhibition, the Xiaoxitian relic exhibition and an ancient Egypt exhibition, marking a shift from making content for venues to operating IP directly, potentially with better margins but requiring content judgment and operational capability.
Customer Structure and Collection Risk
Customers are mainly government and large enterprises, whose payment processes are long, acceptance involves many steps and settlement follows fiscal budget cycles. Receivables remained high in the first half with provisions of 50 percent for two to three years and 100 percent beyond three, totaling 39.94 million.
That impairment equals 17.3 percent of period revenue, indicating that a meaningful share of historical projects is hard to collect. For a company with 200 to 300 million of annual revenue, impairment at that level persistently suppresses profit.
Business Models for Embodied Data
Three models are possible: selling datasets through one-time licensing, selling data services through customized collection and annotation, and selling a simulation platform through licenses plus compute. Datasets and platforms replicate while data services remain project-based.
Revenue and customers for the data business are undisclosed. Industry practice suggests buyers are robot makers, embodied model companies and research institutions, a market still early with limited buyers, uncertain ticket sizes and extremely high quality requirements on annotation precision, scene diversity and physical realism.
Distance From Project Business to Product Business
The core challenge is moving from project to product. Project revenue scales with headcount and project count while product revenue scales with licenses and subscriptions, with gross margins of 30 to 40 percent for the former and above 70 percent for the latter.
AI 3D at 36.32 percent gross margin is close to project levels, indicating customized delivery remains dominant. Standardization through reusing one twin platform across customers could lift margins, but that requires product investment and industry standardization rather than short-term effort.
Competitive Landscape
Peers in Digital Exhibition
Peers in digital exhibition and creative services include Fengyuzhu, Silkroad Visual Technology and Crystal CG, with similar project-based models, government customers and multimedia integration, competing on creative proposals, pricing and local relationships.
The industry faces the same situation as Frontop: new government venue projects contracting, maintenance of existing sites dominating and price competition intensifying. Frontop's relative advantages are 3D visualization depth and listed-company funding, with weaknesses in scale and regional concentration.
Competitors in Embodied Data
Embodied data is new, with three groups of participants. Robot companies build in-house data teams such as Agibot and Unitree. Specialized data providers such as Guanglun Intelligent and Maniformer offer collection hardware, platforms and datasets. Simulation platform vendors including the NVIDIA Isaac Sim ecosystem and domestic simulation companies offer environments and synthetic data.
Frontop sits closer to the third group but with weaker physics simulation than specialists and stronger scene capability than most peers. Whether that intersection works depends on whether customers want a supplier that understands both scenes and data generation.
Difference From Pure AI Companies
Pure AI teams typically lack scenes and data while companies like Frontop lack models and algorithms, making cooperation more likely than competition, with Frontop supplying scenes and data and AI companies supplying models and training. If Frontop builds its own models, it becomes a competitor to its customers, the classic dilemma of data service providers.
Financing and Capital Structure
Listing and Financial Position
Listed on ChiNext under 301313, the company reported a weighted average return on net assets of negative 12.25 percent and return on invested capital of negative 10.71 percent for the first half. At the August 25, 2026 close the trailing twelve month price to earnings ratio was about negative 20.67 times owing to losses, price to book about 5.73 times and price to sales about 6.31 times.
Historical results are highly volatile: 2022 revenue of 609 million with attributable net profit of 18.27 million, 2023 revenue of 576 million with 10.50 million, 2024 revenue of 365 million with a loss of 187 million, and 2025 revenue of 631 million with a loss of 135 million. Two profitable years at tens of millions and two loss years at over 100 million in four years show extreme sensitivity to project cycles.
Cash Flow and Financing
First half operating cash flow was negative 57.97 million against positive 3.44 million a year earlier, financing cash flow was positive 72.89 million, up 56.70 million, and investing cash flow was negative 63.63 million. Financing inflow covered operating and investing outflows, indicating reliance on external funding to sustain transformation spending.
The structure warrants caution: continued negative operating cash flow and growing impairment would deepen dependence on financing. A price to book of 5.73 times and price to sales of 6.31 times is not cheap for a loss-making company with declining revenue, and it rests on the physical AI data foundation narrative.
Financial Position in the First Nine Months of 2026
The third quarter report of October 27, 2026 is the key node for judging the pace of the transformation. Revenue for the first nine months was 338 million renminbi, down 19.45 percent; attributable net profit was negative 123 million against negative 96.41 million a year earlier; adjusted net profit was negative 125 million; operating cash flow was negative 95.12 million; basic earnings per share were negative 1.20; the weighted average return on net assets was negative 20.26 percent; gross margin was 27.78 percent, up 3.75 points; and research expense grew 77.12 percent. Third quarter revenue alone was 107 million, up 7.20 percent, with an adjusted loss.
Two signals point in opposite directions. The first is that third quarter revenue turned positive year over year for the first time since the transformation began, but the magnitude is small and no segment breakdown is given, so it cannot be attributed to AI 3D rather than to a low base. The second is the change in the funding structure: short-term borrowings of 152 million, up 108.48 percent, plus long-term borrowings of 74.64 million for an acquisition loan, against a new 10 percent stake in Guangdong embodied intelligence technology for 28.62 million renminbi. Transformation spending is therefore no longer covered by internal research expense alone but now draws on interest-bearing external funding, which reduces balance sheet flexibility.
The third quarter report also confirms that receivable pressure continues: operating cash outflow of 95.12 million for the first nine months widened by about 37 million from the 57.97 million outflow in the first half. Losses, cash flow and borrowing deteriorating together are the core basis for keeping a cautious stance on this company.
The valuation basis also needs updating: at the October 27, 2026 close the trailing twelve month price to earnings ratio was about negative 40.69 times owing to losses, price to book about 11.92 times and price to sales about 12 times, clearly above the price to book of 5.73 times and price to sales of 6.31 times at the interim report date, which shows the market pricing the transformation narrative more highly even as losses widened.
Impairment and Asset Quality
Credit and asset impairment totaled 39.94 million, up 16.14 million year over year. The stated policy of 50 percent for two to three years and 100 percent beyond three is conservative, but rising impairment indicates aging receivables as customer settlement cycles lengthen.
Two indicators deserve monitoring: receivables turnover at 0.29 times in the first half, which is low, and the cumulative impairment balance. Further increases would keep pressuring profit.
DISCREPANCIES
1. Revenue decline coexisting with AI growth. First half revenue fell 27.80 percent while AI 3D grew 23.73 percent. Some coverage headlines robust AI growth and downplays the overall decline, while other coverage emphasizes both revenue and profit falling. Both facts hold: the new business grows but its scale cannot offset the legacy decline, with AI 3D at 24.55 percent against legacy at 56.06 percent. This report presents both. Source tier: OFFICIAL (interim report).
2. Divergence between gross margin rising and net margin falling. Gross margin was 31.49 percent, up 4.87 points, while net margin was negative 36.66 percent, down 22 points. Margin improvement came from mix as higher-margin AI 3D grew, while net margin deteriorated from expenses with research up 87.19 percent and impairment of 39.94 million. Some interpretations treat rising gross margin as a sign of improving profitability, ignoring expense and impairment erosion. This report presents both and explains the divergence. Source tier: OFFICIAL.
3. Attribution of the wider loss. The company attributes it to three factors: contracting exhibition projects, higher research spending and higher impairment. Their natures differ as external demand, deliberate investment and asset quality, and no quantification by factor is given. This report lists all three without single attribution. Source tier: OFFICIAL (company statement).
4. Severity of negative operating cash flow. Operating cash flow was negative 57.97 million against positive 3.44 million a year earlier, a swing of 61.41 million that exceeds the deterioration in net profit, indicating collection problems are worse than the income statement suggests. Some coverage notes only that cash flow was negative without the year-on-year change. This report states the magnitude. Source tier: OFFICIAL.
5. The physical AI data foundation positioning versus actual revenue mix. The company positions itself as a physical AI data foundation and solutions provider while AI 3D at 56.65 million is only 24.55 percent of first half revenue and legacy integrated services remain 56.06 percent. This report uses the in-transition framing and does not equate positioning with current state. Source tier: OFFICIAL (positioning) against financial structure.
6. Base effects in the 2025 AI 3D figure. The 2025 annual report disclosed AI 3D digital twin products and solutions revenue of 175 million renminbi, up 255 percent, against 56.65 million up 23.73 percent in the first half of 2026. The deceleration from 255 to 23.73 percent is sometimes read as stalling growth, but the 2025 surge may reflect a low base and concentrated recognition of new orders. This report presents both growth rates and flags base effects. Source tier: OFFICIAL (annual and interim reports).
7. Two definitions of research intensity. Research expense was 31.90 million, up 87.19 percent and 13.8 percent of revenue, while a broader research investment figure including capitalized amounts could differ. The capitalization ratio is undisclosed. This report uses the expensed basis. Source tier: OFFICIAL.
8. Missing disclosure on the embodied data business. The company lists data business among four growth poles without disclosing revenue, customers, data volume or collection method, while some interpretations treat it as an existing business. This report labels it a planned new business rather than realized revenue. Source tier: OFFICIAL (strategy statement) with operating data absent.
9. The magnitude of the revenue decline differs by reporting period. The interim report records first half revenue of 231 million renminbi, down 27.80 percent, while the third quarter report records first nine months revenue of 338 million, down 19.45 percent, with the third quarter alone at 107 million, up 7.20 percent. All three figures come from periodic reports and all are true, but they cover different periods, so citing only the interim 27.80 percent decline overstates the current overall drop, while citing only the third quarter 7.20 percent growth conceals that the first nine months are still down. The deterioration in losses and cash flow is consistent across both bases: attributable net profit for the first nine months was negative 123 million against negative 96.41 million a year earlier, and operating cash flow was negative 95.12 million. This report uses the third quarter report as the current basis and retains interim figures as half-year facts. Source tier: OFFICIAL (interim report) and OFFICIAL (third quarter report) in parallel, with the third quarter report as the current basis.
Risks and Diligence Gaps
First, the legacy business is declining faster than the new business replaces it. Legacy fell 45.54 percent while AI 3D grew 23.73 percent, and at current scale AI 3D would need to double for two consecutive years to offset the decline, which sets the revenue pressure for 2027.
Second, receivable impairment risk. Impairment of 39.94 million, turnover of 0.29 times and government-heavy customers mean continued fiscal tightness could push impairment higher.
Third, negative operating cash flow and deepening reliance on financing. The first half operating cash outflow of 57.97 million widened to 95.12 million for the first nine months, while short-term borrowings doubled to 152 million and a further 74.64 million of acquisition loans were taken on. Tighter funding conditions would constrain transformation spending, and higher financing costs would widen the loss against a return on net assets already down to negative 20.26 percent.
Fourth, unknown physics engine capability. FunSim's physical fidelity is central to embodied data quality, yet engine provenance, fidelity and simulation scale are undisclosed.
Fifth, the sim-to-real gap. Policies trained on synthetic data may fail on real robots, a shared industry difficulty, and without real data collection capability the data business loses value.
Sixth, uncertainty in the embodied data market. Buyers are limited in number and many choose to build in-house teams, leaving third-party data service market size unvalidated.
Seventh, valuation diverging from fundamentals. A price to book of 5.73 times and price to sales of 6.31 times price in transformation expectations rather than current results, leaving material downside if progress lags.
Composite Assessment
Technology scores 6.5. Three-dimensional content production and real-time rendering are genuinely engineered through the FTE engine, scene modeling and interaction templates, digital twin capability transfers to the scene layer of embodied data, and the four-category taxonomy is reasonable. Deductions for unknown physics engine autonomy, a missing sim-to-real loop, undisclosed data scale and unclear AI content in AI 3D.
Business scores 5.5. AI 3D at 56.65 million with a 36.32 percent margin and 23.73 percent growth shows demand and pricing power, and cultural IP operations open a consumer path. Deductions for revenue down 27.80 percent, losses up 64.86 percent, operating cash flow turning negative and AI 3D margins still near project levels.
Finance scores 4.5. Listed status provides financing access and 72.89 million of financing inflow supports transformation. Deductions for two consecutive loss years at 187 million in 2024 and 135 million in 2025, a loss of 123 million over the first nine months of 2026, 39.94 million of impairment, an operating cash outflow of 95.12 million, and the interest cost pressure from short-term borrowings doubling to 152 million alongside a return on net assets of negative 20.26 percent.
Product scores 6.0. Three lines plus cultural IP cover government, enterprise and consumer segments, and AI 3D digital twins with FunSim align with embodied intelligence demand. Deductions for low product standardization, unverified physics simulation and a data business not yet at scale.
Market scores 6.5. Embodied intelligence and humanoid industrialization create surging data demand, physical AI draws both policy and capital attention, and the company differentiates on scenes and visualization. Deductions for the contraction of digital exhibition, an early embodied data market with limited buyers, and strong competitors building in house or specializing in data.
Overall 5.8. This is a company whose transformation logic holds while execution is halfway: it uses a decade of 3D scene capability to enter embodied data, the direction is right and AI 3D growth and margin prove demand, yet legacy decline, receivable impairment and negative operating cash flow show the cost of transformation arriving all at once. The judgment on physical AI is that scene and rendering capability are real while physics simulation and data scale remain unverified, and the shift from a project-based digital creative company to a product-based physical AI data provider depends on FunSim achieving physical fidelity and the data business producing disclosable revenue in 2027.
References
- Frontop 2026 interim report, disclosed August 26, 2026: revenue 231 million renminbi down 27.80 percent, attributable net profit negative 77.25 million against negative 46.86 million a year earlier, adjusted net profit negative 79.74 million against negative 47.69 million, operating cash flow negative 57.97 million against positive 3.44 million, basic earnings per share negative 0.75, weighted average return on net assets negative 12.25 percent, gross margin 31.49 percent up 4.87 points, net margin negative 36.66 percent down 22 points
- Business mix from the same report: integrated digital services 129 million or 56.06 percent down 45.54 percent, AI 3D digital twin products and solutions 56.65 million or 24.55 percent up 23.73 percent at a 36.32 percent gross margin, 3D visualization products and services 28.43 million or 12.32 percent, other 16.30 million or 7.06 percent including about 16 million of cultural IP operations
- Regional mix from the same report: South China 125 million or 54.28 percent, North China including the northeast 47.37 million or 20.53 percent, East China 33.31 million or 14.44 percent, other regions 24.81 million or 10.75 percent
- Cost and impairment data from the same report: research expense 31.90 million up 87.19 percent, period expenses 124 million up 11.75 million with an expense ratio of 53.58 percent up 18.57 points, credit and asset impairment totaling 39.94 million up 16.14 million, receivables provision policy of 50 percent for two to three years and 100 percent beyond three
- Strategy statement from the same report: accelerating the shift from a traditional digital creative company to a physical AI data foundation and solutions provider; cultivating four growth poles in physical AI 3D digital twin products and solutions, embodied application solutions, data business and consumer cultural IP operations; research directed at data business, software toolchain, embodied intelligence data simulation platform, AI 3D digital twins and integrated solutions
- Second quarter 2026 data: revenue 89.07 million down 52.18 percent year over year and 37.13 percent sequentially, attributable net profit negative 52.12 million down 115.0 percent year over year and 107.36 percent sequentially, adjusted net profit negative 52.40 million down 111.02 and 91.73 percent, gross margin 33.51 percent up 6.29 points, net margin negative 62.65 percent
- Historical results from third-party financial data: 2022 revenue 609 million with attributable net profit of 18.27 million, 2023 revenue 576 million with 10.50 million, 2024 revenue 365 million with a loss of 187 million, 2025 revenue 631 million with a loss of 135 million
- Valuation data: at the August 25, 2026 close, trailing twelve month price to earnings about negative 20.67 times, price to book about 5.73 times and price to sales about 6.31 times
- Cultural IP business from the interim report: operations including the Yongle Palace digital art exhibition, the Xiaoxitian relic and sculpture art exhibition and an ancient Egypt artifacts and digital art experience
- Industry background on embodied data from public sources: training data comes from teleoperated real-robot collection, simulation generation and video or motion capture; requirements cover three-dimensional scenes, object poses, motion trajectories and semantic annotation; simulation is cheap but carries a sim-to-real gap, and the common practice is simulation pretraining followed by real-data fine-tuning
- Competitive landscape from public sources: digital exhibition peers include Fengyuzhu, Silkroad Visual Technology and Crystal CG; embodied data participants include robot companies building in-house teams such as Agibot and Unitree, specialized data providers such as Guanglun Intelligent and Maniformer, and simulation platform vendors including the NVIDIA Isaac Sim ecosystem and domestic simulation companies
- Technical background from public sources: rendering engines rest on computer graphics including rasterization, ray tracing and material models while physics simulation platforms rest on numerical computation including rigid body dynamics, collision detection, contact solving and soft body or fluid behavior; the stacks differ greatly and team capability does not transfer directly
- Frontop 2026 third quarter report, disclosed October 27, 2026: revenue for the first nine months 338 million renminbi down 19.45 percent, attributable net profit negative 123 million against negative 96.41 million a year earlier, adjusted net profit negative 125 million, operating cash flow negative 95.12 million, basic earnings per share negative 1.20, weighted average return on net assets negative 20.26 percent, gross margin 27.78 percent up 3.75 points, research expense up 77.12 percent; third quarter revenue 107 million up 7.20 percent; short-term borrowings 152 million up 108.48 percent and long-term borrowings 74.64 million for an acquisition loan; a 10 percent stake in Guangdong embodied intelligence technology for 28.62 million renminbi
- Valuation data: at the October 27, 2026 close, trailing twelve month price to earnings about negative 40.69 times, price to book about 11.92 times and price to sales about 12 times
This report is compiled from public sources for informational purposes only and is not investment advice. Financial figures follow company periodic reports; revenue, customers and data volume for the embodied data business are undisclosed, and related judgments are analytical and labeled as uncertain.
Related papers
Related tools
Related cases
- DrivAerML High-Fidelity Road-Car Aerodynamics Dataset and Surrogates
- AirfRANS High-Fidelity Airfoil Surrogate Benchmark Case
- Physics-Informed CNN for Battery Pack Temperature Fields
- Building Energy Simulation and Surrogates: The EnergyPlus Case
- OpenFOAM Industrial Multiphysics Pipeline: From Meshing to Data Supply
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
- 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
- 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
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.