Legal name: Beijing Shichuang Technology Co., Ltd.
Shichuang Technology Deep Dive: SupreCAST Cloud-Native Material-Processing Simulation and Industrial Agents Advance the CAE+AI Strategy
Shichuang (Tsinghua-origin) runs SupreCAST cloud-native material-processing simulation (15,000-plus customers) and launches SupreXI/SupreHub industrial agents; a near-RMB 100 million B++ round closed in July 2026.
Executive Summary
Supreium, formally Beijing Supreium Technology Co., Ltd., was founded in 2016 in Beijing's Haidian district by Guo Zhipeng. It builds simulation software for materials processing, centered on the cloud-native die casting platform SupreCAST, which covers high-pressure die casting, low-pressure casting, gravity casting and semi-solid forming, alongside modules for mold cooling optimization and process control.
On July 24, 2026 the company announced a B plus plus round of nearly 100 million renminbi led by Guoxin Oriza with existing investors Bridge Capital and Sungiant participating. Proceeds fund product iteration on the physical simulation platform and the industrial agent platform, plus expansion in automotive, aerospace and energy. This followed earlier angel, pre-A, A and A plus rounds of nearly 100 million renminbi and a B round in the tens of millions.
The development path has three clear phases. From 2016 to 2020 the company built its own solvers from scratch rather than wrapping open-source code for quick monetization. The first product reached market in 2020. In 2024 several things happened at once: the SupreCAST cloud platform and Supreium-DPB intelligent process brain for integrated die casting launched, the A and B rounds closed, the company won Chery's dual-shot technology development project and Changan's integrated die casting digital twin upgrade project, cumulative customers passed ten thousand with nearly 80,000 mold flow simulations, and Supreium Tengyang was established to enter the southern China market.
The most recent product milestone was the SupreCAST 3.0 online launch on December 3, 2025, followed by the 2026 move into industrial agents. The company now describes two product lines: a physical simulation platform as the capability base and an industrial agent platform as the strategic layer. The latter includes SupreXI for research and development, covering product development, simulation modeling, scheme evaluation, defect diagnosis and parameter optimization, and SupreHub for production, covering parameter optimization, quality prediction, anomaly diagnosis and process control. A Zhuwen mini-program provides industry knowledge content that the company says has formed a user ecosystem in the millions.
In physical AI terms this is a textbook case of process physical AI: rather than pursuing a general physics foundation model, it encodes the mechanisms of materials processing, including filling, solidification, gas entrapment, thermal stress and die temperature evolution, into computable models and layers AI on top for surrogates, diagnosis and optimization. Founder Guo Zhipeng offers a memorable framing: large models are strong at understanding what people say and at orchestrating functions, but they need a team, meaning the industrial software capabilities built over the past decade, and without that a large model is a commander without troops.
One framing point matters. This is a Chinese industrial software company with a clear technical path and commercial validation but no disclosed financials. Its customer coverage of the top 100 die casters and mold makers and 12 leading automakers, and its wins at Chery and Changan, are verifiable. Revenue, gross margin and profitability are not, and the cumulative amount raised cannot be summed precisely. Assessment must rest on product, customers and technology rather than financials.
Company Overview
From a Tsinghua Lab to the Factory Floor
Founder Guo Zhipeng holds a doctorate in fluid mechanics and taught and researched applied mathematics at Tsinghua University, with another source also placing him at the Institute of Mechanics of the Chinese Academy of Sciences, before leaving his academic post in 2020 to work full time on the company. He has described the motivation as finding a direction that used his applied mathematics background while addressing national needs such as turbine blades for aero engines, and recognized that doing process work well requires simulation.
That background determined the technical starting point: beginning from numerical methods rather than interfaces and workflows. The company says it chose the harder road by not wrapping open-source solvers, on the reasoning that open-source solvers have ceilings and must be abandoned to break through. The choice cost time, with the first product only in 2020, and paid off technically, since the solver capability is proprietary and can be optimized for specific process scenarios.
On team composition the company states that more than half of research staff hold master's degrees or above, that core researchers come from the Chinese Academy of Sciences and Tsinghua, that dozens of invention patents and software copyrights have been filed, and that there is proprietary work in CAE solvers, meshing and GPU parallel computing. Research is split between Beijing headquarters and Suzhou.
Development Milestones
Founded in 2016, the company spent 2016 to 2020 on solvers, shipped its first product in 2020, broke through in 2024, released version 3.0 in 2025 and turned toward industrial agents in 2026. The rhythm matches the general law of industrial software: solvers mature over five to eight years, productization and market validation take another three to five, and scale comes after that.
Several 2024 events deserve separate mention. On product, SupreCAST and Supreium-DPB launched together, the latter being an intelligent process brain for integrated die casting that combines process knowledge with algorithms for die temperature control. On capital, the A and B rounds closed in quick succession. On market, the company won Chery's dual-shot technology development project and Changan's integrated die casting digital twin flexible upgrade project, both being automaker investments in frontier processes, which indicates technical acceptance. On organization, Supreium Tengyang was established to enter southern China, where die casting is concentrated.
Customer Coverage
Three sets of coverage figures circulate with different definitions. The financing release says service to more than 15,000 enterprises globally. The founder told China Daily in August 2026 that paying customers exceed 1,000 while users approach 20,000. The end-2024 framing was more than ten thousand enterprises served with nearly 80,000 mold flow simulations. The figures are consistent but mean very different things: nearly 20,000 users or trials against just over 1,000 paying customers implies a paid conversion rate near 5 percent. For a SaaS-modeled industrial software company that is not abnormal, but it means the 15,000 figure describes reach rather than payment.
On customer structure the company claims coverage of the top 100 die casters and mold makers nationally and 12 leading automakers, with named customers including Jianghuai, Seres, XPeng, Chery, Dongfeng Nissan, CITIC Dicastal, Hongtu Group and Dayang Motor, energy customers in oil and wind power, and universities including Tsinghua and Shanghai Jiao Tong. The value of that list lies in full coverage of the top 100 die casters, since die casting is a concentrated industry where winning the leaders creates demonstration effects down the long tail.
Technology
The Physics Difficulty of Die Casting Simulation
Understanding Supreium's technical value requires understanding why die casting simulation is hard. High-pressure die casting forces molten metal into a mold cavity at tens of metres per second, filling it in milliseconds, then solidifies it under high pressure. Four coupled physics are involved.
The first is free surface flow. Filling is a gas-liquid two-phase flow where gas entrapment is a major defect source. Numerically this requires interface tracking or capturing, with volume of fluid and level set methods common, and interface breakup at high fill speed demands very fine meshes.
The second is heat transfer and solidification. The interfacial heat transfer coefficient between melt and die varies with contact state and drops sharply once a gap forms, latent heat release affects the temperature field, and shrinkage porosity depends on feeding capability in the last regions to solidify. The hard part is calibrating interfacial coefficients, which relies on experimental data and empirical models.
The third is stress and distortion. Solidification shrinkage and die thermal expansion cause distortion and hot tearing, and dies suffer thermal fatigue under repeated cycles. This needs thermo-mechanical coupling with complex high-temperature constitutive behavior.
The fourth is multiscale geometry. Wall thickness ranges from 0.8 millimetres to several millimetres while dies reach metres, spanning three orders of magnitude in mesh size. Integrated die casting, which merges dozens of parts into one large casting, amplifies the problem, with single parts weighing tens to over a hundred kilograms and meshes reaching the hundreds of millions of cells.
Coupling these four is the core difficulty. Foreign tools have accumulated material databases, interfacial coefficient models and defect criteria over more than thirty years, and catching up on those empirical parameters is harder than catching up on algorithms.
Proprietary Solvers and GPU Acceleration
Supreium's technical argument has three layers. The first is proprietary solvers covering fluid, heat transfer, structures, multiphysics coupling and high performance computing. The claim of full in-house development carries real meaning in the Chinese CAE context, since it means no dependence on frameworks such as OpenFOAM and the ability to optimize discretization and solution strategies specifically for die casting.
The second is GPU acceleration and cloud-native architecture. Die casting simulation involves large meshes and small time steps, and GPU parallelism can cut a run from hours to tens of minutes. The company cites mesh sizes reaching the hundreds of millions of cells with runtimes several times shorter than traditional desktop software. Cloud-native means more than hosting: elastic compute, job scheduling, collaboration and knowledge retention live in one architecture, and users open a browser without workstations or license servers.
The third is chain simulation. The company extends single-point simulation into a cross-process chain of casting, plastic forming and heat treatment, where each step's output feeds the next, with residual stress and microstructure entering the heat treatment model and post-treatment properties entering service analysis. The value is continuity across the process chain, since real manufacturing has multiple steps and single-point simulation cannot answer whether a casting will distort after heat treatment.
How CAE and AI Combine
Three specific AI paths exist, with different natures.
The first is surrogate modeling. Supreium-DPB is described as an integrated die casting intelligent die temperature control system based on a coupled CAE and AI surrogate model. The logic is to generate training data with physical simulation, train a surrogate mapping process parameters such as die temperature, shot velocity and pouring temperature to quality metrics such as defect rate and dimensional deviation, then use the surrogate for real-time optimization and control. Surrogate inference costs far less than simulation, so it can sit inside a closed production loop, which is the classic pattern of physics simulation training offline and AI inferring online.
The second is defect diagnosis and parameter optimization. SupreXI targets scheme evaluation, defect diagnosis and parameter optimization, essentially encoding expert interpretation, such as what a given temperature field distribution implies about defects, into rules or models and then having an agent orchestrate simulation iterations. The technical substance lies in the accuracy of the criteria, because an agent with inaccurate criteria simply produces wrong conclusions faster.
The third is production-side agents. SupreHub targets parameter optimization, quality prediction, anomaly diagnosis and process control, fusing simulation models with shop floor data so production moves from after-the-fact analysis toward process awareness and autonomous optimization. The difficulty here is not algorithms but data, since real-time machine data such as shot curves, die temperature and vacuum level must be aligned with simulation models and cleaned of sensor noise and operating drift.
The Commander Without Troops Argument
Guo Zhipeng's view of the relationship between AI and industrial software deserves recording. He argues that large models are strong at understanding language and at orchestrating functions, but need a team of industrial software capabilities covering geometry handling, meshing, solvers, post-processing and material databases, and without that a large model is a commander without troops.
The judgment is technically correct and matches the prevailing industrial AI architecture, where the language model handles intent and orchestration, professional tools handle computation and the data layer supplies context. It also explains a common misreading, namely treating the integration of a large model as achieving industrial AI. The real threshold is tool capability rather than model interface.
The company frames this as a physical base plus industrial agents, with the base being solvers and the chain simulation platform, the agents being SupreXI and SupreHub, and physical data in between, combining simulation-generated virtual data with field-collected real data. If realized, the three-layer structure is a digital twin of the process.
Technical Shortfalls
First, depth of material databases and process experience. Die casting accuracy depends heavily on thermophysical properties, interfacial heat transfer coefficients and defect criteria, all requiring extensive experimental calibration. MAGMA and ProCAST have decades of accumulation, and a company founded in 2016 cannot match that coverage and calibration precision. This is the hardest part to catch up on.
Second, breadth and depth of solver physics. The company advertises five solver families covering mechanics, thermal, optical, acoustic and electromagnetic, while what it actually demonstrates centers on fluid, heat transfer, structures and multiphysics directly relevant to materials processing. Whether the optical, acoustic and electromagnetic solvers are commercially mature is not shown publicly, and marketing breadth exceeding demonstrated capability is common among Chinese CAE vendors.
Third, missing quantitative evidence for AI. There are no public figures on SupreXI and SupreHub effectiveness, whether diagnostic accuracy, yield improvement from optimization or response time. The DPB award is product recognition rather than customer-side quantified benefit.
Fourth, coupling depth in multiphysics. Die casting centers on flow-thermal-mechanical coupling where Supreium has accumulated experience, but whether more complex coupling such as thermal-metallurgical-mechanical with microstructure evolution is covered remains unclear.
International Academic Showcase (World Foundry Congress 2026)
From October 19 to 21, 2026, product director Shi Hao served as co-chair of the Intelligent Digital Casting and Materials Genome Engineering sub-forum at the 76th World Foundry Congress (WFC 2026, Istanbul) and on October 21 presented the talk SupreCAST: A Cloud-Native Platform for Casting Process Development and Quality Control. The talk disclosed a stack of adaptive mesh refinement, LBM, a high-performance solver, LES and VOF, claiming more than 10x speedup over comparable desktop software (company-original wording, pending independent reproduction), up to one billion cells per case and more than 90 percent parallel efficiency on 103 cores. This shows the solver capability entering the international foundry research arena, but the performance figures come from a company demonstration and are not independently verified.
Product and Commercial Model
Product Matrix
The core is SupreCAST, a cloud-native die casting and casting simulation platform covering filling, solidification, gas entrapment, shrinkage porosity, thermal stress analysis and process optimization. By process it spans high-pressure die casting, low-pressure casting, gravity casting, tilting casting, investment casting, centrifugal casting and semi-solid forming.
Three supporting directions exist. The mold cooling module handles cooling channel design and evaluation, the process control module handles real-time monitoring and traceability of die casting parameters, and Supreium-DPB provides intelligent die temperature control for integrated die casting.
The chain simulation platform extends single-point capability to casting, plastic forming and heat treatment, and the company has planned and developed more than ten process simulation products covering investment casting, forging and heat treatment.
The industrial agent platform is the 2026 addition, with SupreXI for research and SupreHub for production. On the ecosystem side the Zhuwen mini-program targets industry knowledge sharing and the company says it has built a user ecosystem in the millions.
Business Model
The model combines SaaS subscription with project delivery. On SaaS, users subscribe and work in a browser without workstations or license servers. On projects, the company takes on automaker development work such as Chery dual-shot technology development and Changan integrated die casting digital twin upgrades.
That combination is pragmatic for Chinese industrial software. Pure SaaS is hard to sell to large manufacturers because of data security, system integration and customization needs, while pure project work does not scale. SaaS serves the long tail and standardized scenarios, projects serve leading customers and frontier processes.
On pricing, subscription lowers the barrier for smaller die casters, since traditional mold flow software carries high per-seat license costs that small plants cannot afford. Cloud pay-as-you-go reduces the threshold, which explains how the company reached nearly 20,000 users.
Market Position and Competition
In Chinese die casting mold flow simulation the main competitors are MAGMASOFT, ProCAST under ESI now part of Keysight, FLOW-3D from Flow Science, AnyCasting from Korea, and domestic players including Huazhu CAE with Huazhong University of Science and Technology roots.
Supreium's relative advantages are threefold. Cloud-native architecture is the first, since foreign tools remain largely desktop-deployed and few domestic peers are cloud-native. Price and accessibility come second, with subscription lowering total cost of ownership for smaller die casters. Localized service is third, since die casting is deeply process-specific and customers need on-site support and process consulting that local teams deliver quickly.
The disadvantages are equally clear. Material databases and process experience trail foreign tools. Brand recognition remains a barrier, since MAGMASOFT and ProCAST are still default choices in the overseas projects of leading automakers. Validation at the high end is unproven, and integrated die casting is a new scenario where every tool lacks experience, which is an opportunity for Supreium but also means no established path to follow.
The company says its market share has surpassed that of international competitors in China, a claim without third-party data that remains company framing. What can be confirmed is high coverage of leading die casters and mold makers and a price advantage in mid-market segments.
Financing and Partnerships
The funding history runs through angel, pre-A, A and A plus at nearly 100 million renminbi earmarked for international expansion, a B round in the tens of millions, and a B plus plus round of nearly 100 million in 2026 led by Guoxin Oriza with Bridge Capital and Sungiant. Investors include Zhongguancun Development Group, TusStar as an early Tsinghua-affiliated investor, Guoxin Oriza under Oriza Holdings, Bridge Capital and Sungiant.
Cumulative funding cannot be summed precisely because some rounds are undisclosed, but a rough estimate places it between 200 and 300 million renminbi. For a company that spent a decade on foundational research that is a medium amount, enough for team and research but not for a price war.
On recognition, the company won first prize at the 14th China Innovation and Entrepreneurship Competition, first prize in the Nonferrous Metals Industry Science and Technology Award, an AI innovation potential award at the fifth Industrial Software Innovation Application Competition for Supreium-DPB, and inclusion in the 2026 Beijing manufacturing digital transformation service catalog. Such credentials materially help with government projects and state-owned enterprise procurement.
Competitive Landscape
Technical Barriers of Foreign Tools
MAGMA is the benchmark for die casting simulation, with advantages in material databases, interfacial heat transfer models and defect criteria accumulated over decades, plus validation cases at global automakers. ProCAST under ESI, now Keysight, is strong in comprehensive casting process coverage and multiphysics coupling. FLOW-3D excels at free surface flow with distinctive capability in filling and gas entrapment. AnyCasting holds meaningful share in Asia, particularly Korea and China, at more accessible prices.
The shared advantage of these tools is a library of empirical parameters covering thermophysical properties, interfacial coefficients and defect criteria calibrated over decades of experiments and production. A domestic tool with comparable algorithms but without those parameters produces insufficient accuracy. That is the hardest threshold to cross and where Supreium must keep investing.
Differences From Domestic Competitors
China has many institutions and companies working on casting and die casting simulation. Huazhu CAE, with roots at Huazhong University of Science and Technology, has long accumulation in casting simulation and a deep academic foundation, alongside other university spinouts and regional software firms.
Supreium differs through cloud-native architecture and degree of commercialization. Huazhu is strong in academia and industry standards with comparatively weaker product experience, while Supreium built as a SaaS product from the start with interfaces, collaboration and subscription closer to modern software. In the mid-market that difference is decisive, since smaller plants want something that works immediately more than the fullest feature list.
Relationship With General CAE Vendors
Ansys, Siemens, Dassault and COMSOL do not build die casting specific tools, though they have casting modules, and their strength lies in general structural, fluid and electromagnetic solving. Die casting is a highly specialized scenario where general tools require substantial customization, which creates room for specialized products.
The ceiling is in the same place: the die casting simulation market is finite because the number of die casters worldwide is finite, and die casting alone cannot support a large company. Supreium's expansion into chain simulation across forging and heat treatment and into industrial agents is fundamentally an attempt to break that ceiling.
Financing and Capital Structure
Rounds and Investors
The funding rhythm tracks the product rhythm. Angel and pre-A rounds from 2016 to 2019 covered solver development, with Zhongguancun Development Group and TusStar participating. The A and A plus rounds of nearly 100 million renminbi covered productization and international expansion, the B round in the tens of millions covered market expansion, and the 2026 B plus plus round of nearly 100 million covers the industrial agent platform.
The B plus plus structure is notable. Guoxin Oriza led, an entity under Oriza Holdings that manages national funds including an industrial machine tool fund, with existing investors Bridge Capital and Sungiant following on. Investor commentary reveals the logic: industrial software carries high technical barriers, materials processing CAE demand grows with advanced manufacturing upgrading, Supreium has systematically laid out the full metal materials processing space across low-pressure casting, gravity casting, investment casting, forging, heat treatment, continuous casting, centrifugal casting, injection molding and semi-solid forming, and it covers the equal-material and additive directions targeted by the industrial machine tool fund.
In other words part of the logic is strategic allocation by a national industry fund rather than pure financial return. That benefits the company through long-duration capital and policy resources, and it also means a degree of industrial mission around domestic substitution and key technology development.
Capital Efficiency
Ten years, cumulative funding of 200 to 300 million renminbi, nearly 20,000 users and more than 1,000 paying customers is moderate-to-good capital efficiency among Chinese industrial software companies. Domestic CAE firms generally need five to eight years of research before a usable product, and Supreium's eight-year accumulation period fits the pattern.
Proceeds go mainly to research on solvers, chain simulation and agents, and to market expansion in southern China and industry coverage. Revenue and gross margin are undisclosed, so unit economics cannot be judged, and under SaaS the question of whether customer acquisition and service cost fall below subscription revenue determines whether the model works.
Future Funding Needs
The industrial agent platform requires sustained investment in model training, data accumulation and engineering, and the more than ten process simulation products in chain simulation each need refinement. At the current pace the B plus plus round covers roughly 18 to 24 months of operations and research, and a listing or further round in 2027 or 2028 would require a clear step up in revenue.
DISCREPANCIES
1. Three customer figures. The financing release cites more than 15,000 enterprises served globally, the founder in an August 2026 interview cites more than 1,000 paying customers and nearly 20,000 users, and the end-2024 framing cites more than ten thousand enterprises with nearly 80,000 mold flow simulations. The three correspond to reach, usage and payment, with paid conversion near 5 percent. This report lists all three and does not equate the 15,000 figure with paying customers. Source tier: OFFICIAL (company) with inconsistent definitions.
2. The market share claim lacks third-party support. The founder says market share has surpassed international competitors in China. The claim has no third-party market research behind it and the definition of market share, whether by customer count, installed base or revenue, is unspecified. This report retains the statement labeled as company framing. Source tier: company statement, UNVERIFIED.
3. Three descriptions of the founder's background. One source describes him as a Tsinghua professor of applied mathematics who left in 2020 to work full time, another places him at the Institute of Mechanics of the Chinese Academy of Sciences, and others describe him as holding a doctorate in fluid mechanics. These can coexist as a sequence from Tsinghua doctorate to the academy to Tsinghua teaching to founding the company, but no complete biography is public. This report uses the description of a fluid mechanics doctorate with teaching and research in applied mathematics at Tsinghua before leaving in 2020, and notes the academy affiliation reported elsewhere. Source tier: multiple MEDIA sources, with no public OFFICIAL biography.
4. Founding year versus full-time commitment. The company was founded in 2016 but the founder left Tsinghua for full-time work in 2020, meaning 2016 to 2020 was a part-time phase of team building that affected productization speed. This report separates the two dates. Source tier: OFFICIAL (interview disclosure).
5. Round history and cumulative amount cannot be summed precisely. Disclosed rounds include angel, pre-A, A and A plus at nearly 100 million, B in the tens of millions, and B plus plus at nearly 100 million, but some amounts are undisclosed, so cumulative funding can only be estimated at 200 to 300 million renminbi. This report gives a range labeled as an estimate. Source tier: company announcements and media, partially incomplete.
6. The five solver families versus demonstrated capability. The website claims breakthroughs in five solver families covering mechanics, thermal, optical, acoustic and electromagnetic, while the scenarios actually shown center on fluid filling, thermal solidification, structural thermal stress and multiphysics coupling, with no evidence for commercial maturity of optical, acoustic and electromagnetic solvers. This report distinguishes marketing framing from demonstrated capability in the technology section. Source tier: OFFICIAL (website) contrasted with product material.
7. Inconsistent product naming. The same product appears as SupreCAST, as the Chinese brand name, and as SupreCloud in third-party material. The official brand is SupreCAST, and SupreCloud appears in third-party sources. This report uses SupreCAST. Source tier: OFFICIAL over third-party.
8. Exclusivity of the first cloud-native claim. The company describes SupreCAST as the first cloud-native simulation platform for materials processing in China. The first claim depends on how cloud-native and materials processing are defined and cannot be independently verified. This report does not use the exclusivity claim. Source tier: company marketing, UNVERIFIED.
9. Three headcount figures coexist, and they are not interchangeable. Registry information gives an insured-person figure of 64 for the legal entity, with registered capital of 7.8039 million renminbi (raised from 7.14863 million on August 24, 2026, and earlier from 6.754613 million on January 9, 2025); the registry scale classification for 2025 is small enterprise. A general encyclopedia entry citing the registry lists a personnel scale of 50 to 99. A commercial company database, updated October 26, 2026, instead lists 21 to 50. These are three different bases: statutory social-insurance participation at the entity, a registry-derived personnel range, and a commercial-database estimate. They are recorded side by side and not converted into one another. The company narrative that emphasizes nearly 20,000 users, more than 1,000 paying customers and a user ecosystem in the millions describes platform reach rather than headcount, so it is kept as a separate basis. Source tier: business registry (OFFICIAL) versus commercial database (MEDIA) versus company marketing (UNVERIFIED).
10. Cumulative financing has two incompatible figures. The main text estimates roughly 200 to 300 million renminbi of cumulative capital from the individual rounds, while a commercial company database records four financing events totalling about 80 million renminbi. The two differ by more than a factor of two and cannot both be right: the database total appears to count only the disclosed amounts of the rounds it tracks, whereas the main-text estimate treats undisclosed rounds and round labels as contributing capital. Neither figure has been confirmed by the company, and the valuation is undisclosed. Adjudication: record both, take neither as settled, and treat the capital-efficiency discussion as resting on a wide range. Source tier: commercial database (MEDIA) versus estimation in this report (UNVERIFIED).
Risks and Diligence Gaps
First, financials are entirely undisclosed. Revenue, gross margin, profitability, average contract value and retention are not public. Under SaaS, cloud infrastructure and implementation service costs determine gross margin, and if subscription prices are close to service cost, more customers mean larger losses. This is the core gap in judging the model.
Second, depth of material database accumulation. Die casting accuracy depends on calibration of thermophysical properties, interfacial heat transfer coefficients and defect criteria over long experimental campaigns. Whether the gap to MAGMA and ProCAST is wide cannot be judged from public information, but it determines high-end competitiveness.
Third, the industrial agent validation gap. SupreXI and SupreHub are 2026 initiatives without published customer quantification on diagnostic accuracy, yield improvement or response time. The DPB award is product recognition rather than effect validation.
Fourth, technical uncertainty in integrated die casting. Integrated die casting is a new process where every simulation tool lacks experience. Winning Chery dual-shot and Changan projects is an opportunity, but simulation accuracy in new processes is itself unstable, and disappointing project outcomes would affect follow-on orders.
Fifth, customer concentration and project dependence. Automaker projects are large in value but limited in number, so revenue may concentrate in a few customers, and project-based recognition and collection cycles pressure cash flow.
Sixth, policy dependence in domestic substitution. Part of growth comes from policy-driven localization requirements in defense, state-owned enterprises and government projects. Slower policy pacing or foreign price competition would affect growth.
Seventh, talent and geography. Research concentrates in Beijing and Suzhou while die casting clusters in Guangdong and the Yangtze River Delta. Supreium Tengyang was created to be closer to southern customers, but service radius and response speed remain competitive factors.
Composite Assessment
Technology scores 7.5. Proprietary solvers plus cloud-native delivery plus chain simulation constitute a real difference in domestic die casting CAE, the physical base plus agent orchestration architecture is correct, and GPU acceleration with hundred-million-cell meshes reaches commercial grade. Deductions for insufficient material database accumulation, unclear maturity of optical, acoustic and electromagnetic solvers, and missing quantitative AI metrics.
Business scores 7.0. The SaaS plus project model is pragmatic, nearly 20,000 users and more than 1,000 paying customers show the product is usable, and high coverage of leading die casters and the top 100 mold makers provides demonstration effects. Deductions because revenue is undisclosed, paid conversion is near 5 percent, and the high end remains dominated by foreign tools.
Finance scores 6.0. Cumulative funding of 200 to 300 million renminbi over ten years is moderate capital efficiency, and the Guoxin Oriza-led round brings long-duration capital and industrial resources. Deductions for entirely undisclosed financials, an unclear path to profitability and project-based cash flow pressure.
Product scores 7.5. SupreCAST covers seven casting processes, version 3.0 improved completeness, the mold cooling and process control modules fill in tooling and shop floor links, and DPB addresses a specific integrated die casting need in die temperature control. Deductions for agents that have just started, a material parameter gap against foreign tools, and inconsistent product naming that weakens brand recognition.
Market scores 7.5. China is the world's largest die casting producer, integrated die casting creates new simulation demand, domestic substitution policy provides entry opportunities, and the company has accumulated leading customers and industry awards. Deductions for a finite market size given the number of die casters, foreign tools holding the high end, and high policy dependence.
Overall 7.1. This is among the clearest in path and most commercially grounded of the domestic CAE companies covered: ten years from solvers to a cloud platform, SaaS lowering the barrier for smaller die casters, chain simulation breaking past single-point tools, and now the physical base connected to industrial agents. Its weaknesses are financial opacity, material database depth and unvalidated AI. The judgment on physical AI is that the physical base of filling, solidification, thermal stress and multiphysics coupling is real and industrially validated and that the argument about large models needing industrial software as their team hits the mark, while the value of process physical AI must ultimately be proven by lower defect rates and fewer trial molds rather than by customer counts and awards.
References
- Supreium B plus plus financing announcement, July 24, 2026: nearly 100 million renminbi led by Guoxin Oriza with existing investors Bridge Capital and Sungiant participating; proceeds for product iteration on the physical simulation platform and industrial agent platform and expansion in automotive, aerospace and energy; a decade of work from the physical simulation engine to the chain simulation platform to the cloud computing framework
- Product and customer information in the same announcement: SupreCAST as the first cloud-native simulation platform for materials processing in China covering high-pressure die casting, low-pressure casting, gravity casting and semi-solid forming, serving more than 15,000 enterprises globally across product development, mold design, process validation, defect prediction and trial mold optimization; the industrial agent platform including SupreXI for research and SupreHub for production; the Zhuwen mini-program forming a user ecosystem in the millions; a long-term goal of serving one million industrial customers worldwide
- Investor commentary in the same announcement: industrial software carries high technical barriers and materials processing CAE demand grows with advanced manufacturing; Supreium has a systematic layout across the full metal materials processing space including low-pressure casting, gravity casting, investment casting, forging, heat treatment, continuous casting, centrifugal casting, injection molding and semi-solid forming, covering the equal-material and additive directions of the industrial machine tool fund; domestic substitution achieved in the segment, with mature and reusable proprietary multiphysics solvers and more than ten process simulation products planned and under development
- China Daily interview, August 25, 2026: Guo Zhipeng as founder and president, founding the company in 2016 and leaving Tsinghua University in 2020 to work full time; a non-benchmarking path starting from customer needs and working backward to software form, producing the cloud CAE product SupreCAST; more than 1,000 paying customers and nearly 20,000 users; a claim that market share has surpassed international competitors in China; the argument that large models need a team of industrial software capabilities or remain a commander without troops
- China Worker Network and Economic Information Daily interview, September 2, 2026: proprietary solvers developed from source code since 2016; 2016 to 2020 on foundational work with the first product in 2020; 2024 launch of SupreCAST and Supreium-DPB, closing of the A and B rounds, wins on Chery dual-shot technology development and Changan integrated die casting digital twin upgrades, more than ten thousand enterprises served and nearly 80,000 mold flow simulations, and establishment of Supreium Tengyang for southern China; an eight-year valley of death
- Company website, accessed 2026: positioning around algorithms bridging industrial logic with product areas covering CAE solvers, software and platforms, analysis and evaluation, optimization and recommendations, and collaboration and connection; a claim of five solver families covering mechanics, thermal, optical, acoustic and electromagnetic; applications in automotive and transportation, aerospace and 3C electronics manufacturing
- Company website news, December 11, 2025: SupreCAST 3.0 online launch held on December 3 under the theme of new order and intelligent manufacturing
- Company website news, April 8, 2026: Supreium-DPB received an AI innovation potential award at the fifth Industrial Software Innovation Application Competition for a coupled CAE and AI surrogate model approach to integrated die casting intelligent die temperature control; the company was included in the 2026 Beijing manufacturing digital transformation service catalog with its combined offering
- Company website news, September 21, 2026: product director Shi Hao presented on reconstructing aerospace research and manufacturing efficiency with chain simulation and AI agents at the 2026 Aerospace Materials and Manufacturing Process Conference
- Company website news, August 27, 2026: the fourth Supreium Cup challenge and the 2026 user conference were held in Shenzhen, focused on CAE plus AI industrial intelligence practice
- Third-party company profile: Beijing Supreium Technology founded in 2016 and headquartered in Haidian with research in Beijing and Suzhou; founder Guo Zhipeng holding a doctorate in fluid mechanics and previously at the Institute of Mechanics of the Chinese Academy of Sciences; core researchers from the Chinese Academy of Sciences and Tsinghua; angel, pre-A and A rounds completed with investors including Zhongguancun Development Group and TusStar; national high-tech enterprise and Beijing specialized and sophisticated SME; more than half of research staff holding master's degrees or above with dozens of patents and software copyrights; GPU acceleration and cloud-native architecture supporting mesh sizes in the hundreds of millions of cells; modules for mold cooling optimization and process control
- Competitor information from public sources: MAGMASOFT as the die casting benchmark with strengths in material databases, interfacial heat transfer models and defect criteria; ProCAST under ESI, now Keysight, strong in casting process coverage and multiphysics coupling; FLOW-3D strong in free surface flow; AnyCasting with meaningful Asian share; Huazhu CAE with Huazhong University of Science and Technology roots among domestic players
- China Development Network and the company WeChat account (October 2026): product director Shi Hao served as co-chair of the Intelligent Digital Casting and Materials Genome Engineering sub-forum at the 76th World Foundry Congress (WFC 2026, Istanbul, October 19 to 21, 2026) and presented a SupreCAST technical talk; disclosed an algorithm stack of adaptive mesh refinement, LBM, LES and VOF with more than 10x speedup, up to one billion cells and more than 90 percent parallel efficiency on 103 cores (company-original figures, not independently verified); the company also visited ANKIROS 2026 around the same time.
- Business registry records (2026): employee count 64; registered capital of 7.803922 million renminbi with paid-in capital of 7.14863 million; registered capital raised from 7.14863 million on August 24, 2026 and earlier from 6.754613 million on January 9, 2025; scale classification is small enterprise, identified in 2025; unified social credit code 91110108MA005TE7XD, registered at 35 Tsinghua East Road, Haidian District, Beijing.
- Commercial company database (updated October 26, 2026): lists personnel scale as 21 to 50 people, and records four financing events totalling about CNY 80 million, with the B-plus round in July 2026 at tens of millions of renminbi (Guoqi Yuanhe Fund, Jiaqiao Capital, Xianjin Investment), the B round in December 2024 (Shenzhen Capital Group, Jiaqiao Capital), the A-plus round in June 2022 (As Angel, Sequoia China, Fangxin Capital among others) and the Pre-A round in November 2020 (TusStar, As Angel). Both the personnel scale and the cumulative financing total are database figures and were not confirmed by the company.
This report is compiled from public sources for informational purposes only and is not investment advice. The company is private and revenue and financial data are undisclosed; customer counts and market share claims are company-reported without third-party verification, and related judgments are labeled as uncertain.
Related tools
Related cases
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
- 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.