Moqi Intelligent Technology

1.1 Identity, product, and operating model

Shenzhen Moqi Intelligent Technology Co., Ltd. was registered in the Qianhai Shenzhen-Hong Kong Cooperation Zone in Shenzhen on August 27, 2025. The available company profile describes its business scope as intelligent-robot research and development, sales, and technical services, with a focus on general-purpose humanoid robots. Its stated technical approach combines a native visual-action model with a world model to support embodied perception, decision-making, and cross-scenario generalization. The same profile says the company had launched an initial embodied-intelligence prototype by April 2026 and intended to start in commercial settings such as hotel room service and last-mile logistics before pursuing household use. These are company-profile assertions rather than independently audited operating results; no source in the assigned pool establishes paid deployments, revenue, unit economics, or customer retention.[CO001, CO002, CO003, CO004]

1.2 Founders, governance, and financing

The founding narrative centers on Gao Wenli, identified as founder, chairman, and manager, and Huang Qingqiu, described as co-founder and a former Huawei autonomous-driving AI leader. The public profile also names Fu Yue and Cao Wei as directors and describes an employee ownership platform alongside institutional investors. It reports a RMB900 million angel round in December 2025, a February 2026 pre-A round, and cumulative angel-series financing above RMB1 billion with a post-money valuation above RMB7 billion by July 2026. A separate article repeats the above-RMB1 billion and above-RMB7 billion figures but uses promotional language and does not provide a term sheet, cap table, or audited financial statement. Accordingly, the financing scale is best treated as reported private-market information, while ownership economics, board rights, liquidation preferences, debt, cash balance, and runway remain open diligence items.[CO005, CO006, CO007, CO008, CO009]

1.3 Milestones, ecosystem constraints, and evidence gaps

The company has moved quickly from incorporation to financing, prototype development, subsidiary formation, and a stated commercial-entry plan. The timeline is notable for capital formation arriving before public evidence of scaled production or customer outcomes. Industry context raises both opportunity and execution requirements: China's 2026 national standard system spans computing, complete machines, applications, safety, and ethics, while public industry coverage highlights the difficulty of moving from controlled demonstrations to reliable real-world operation. Export controls and geopolitical restrictions can also affect advanced-computing access for Chinese embodied-AI companies, although the assigned evidence does not show a Moqi-specific restriction or incident. The central diligence question is therefore whether the financing and technical ambition convert into repeatable deployments, measurable safety, and disclosed commercial traction.[CO010, CO011, CO012, CO013, CO014]

1.4 Exhibits

2.1 Market boundary and adoption context

This analysis defines the market as embodied-intelligence robots that combine perception, planning, manipulation, and physical execution in human or industrial environments. Included spend covers robot hardware, onboard or edge AI, sensors and actuators, deployment, maintenance, data operations, and recurring robot-as-a-service or software layers. Excluded spend includes conventional fixed automation that does not need general-purpose physical adaptability, pure software assistants, and broad semiconductor revenue unless it is directly purchased to operate the robot. The near-term substitute set is not only another humanoid: specialized industrial robots, logistics automation, human labor, and internal process redesign can all solve parts of the same job. The market boundary therefore supports a deployment-led analysis rather than a single broad “AI” estimate.[CM001, CM002]

2.2 Evidence-constrained sizing lenses

Public evidence provides operating and policy proxies, not a company-specific revenue market size. China's Ministry of Industry and Information Technology committee described a national standard system spanning the humanoid lifecycle, while Xinhua reported more than 140 domestic manufacturers and more than 330 models released in 2025. Independent industry coverage describes AGIBOT crossing 10,000 units and Unitree reporting more than 5,500 shipments in 2025, indicating that production capacity and fleet deployment are becoming measurable market variables. These figures should not be added together or treated as market revenue: they measure different companies, periods, and operating concepts. Figure AI's reported $39 billion post-money valuation demonstrates investor willingness to fund the category, but valuation is not market size. Moqi's addressable opportunity remains an evidence gap until product price, utilization, customer segment, deployment geography, and repeatable unit economics are disclosed.[CM005, CM006, CM007, CM008, CM009, CM010]

2.3 Buyers, users, and adoption path

The most supportable initial buyer path is commercial or industrial deployment where layouts, tasks, and safety procedures can be bounded more tightly than in homes. The economic buyer may be an operations, manufacturing, logistics, facilities, or innovation leader; the user is the worker or supervisor who works alongside the robot; and the payer is the operating unit or central automation budget. A robot-as-a-service model can reduce upfront capital requirements, but the assigned evidence describes that model for AGIBOT and Figure AI rather than for Moqi, so it is a market mechanism, not a Moqi business-model fact. Adoption triggers are labor scarcity, repetitive or hazardous work, and the ability to reuse existing spaces. The adoption path still requires proof of task success, uptime, safety, maintenance cost, and payback against specialized automation or labor.[CM011, CM012, CM013, CM014]

2.4 Drivers, constraints, and valuation relevance

Policy standardization is a demand-enabling driver because it can reduce uncertainty around safety, data, and lifecycle practices; China's 2026 standard system explicitly covers application, safety, ethics, computing, components, and complete systems. Production acceleration and large funding rounds are additional ecosystem signals, but neither proves customer willingness to pay. Constraints are equally material: the market must close the gap between polished demonstrations and reliable long-duration operation, while hardware, supply chain, data acquisition, generalization, and geopolitical access all remain execution risks. Harvard's discussion of semiconductor export controls shows that Chinese AI infrastructure operates within an active technology-access constraint. The valuation implication is asymmetric: a large category can support strategic financing, but Moqi's valuation cannot be underwritten from category growth without deployment evidence and disclosed economics.[CM015, CM016, CM017, CM018]

2.5 Exhibits

3.1 Competitive landscape and scale benchmarks

The relevant competitive set spans Chinese mass-production leaders, US full-stack humanoid platforms, and the status quo of specialized automation or internal robotics programs. Unitree and AGIBOT are the strongest public China scale comparators in this pool: Humanoid Index reports more than 5,500 Unitree units shipped in 2025, while Humanoids Daily reports AGIBOT's 10,000th unit and a fourfold acceleration from 5,000 to 10,000 units in three months. TrendForce independently frames both as the leading pair and projects them to represent nearly 80% of shipments. These are shipment or output indicators, not proof of profitable deployment, recurring revenue, or Morphi displacement.[CP001, CP002, CP003, CP004]

3.2 Capability, distribution, and buying trade-offs

Buyers appear to be choosing among scale, workflow integration, and the ability to support a robot after deployment. Unitree's reported volume and low-end G1 reference price create a hardware-cost benchmark; AGIBOT adds order-driven manufacturing, a rental model, and industrial partnerships; Figure provides a vertically integrated US alternative with Helix software, BotQ production, and BMW operating data. Morphi's reported thesis is different in emphasis rather than yet demonstrated outcome: coverage describes it as porting automotive production discipline, multi-sensor fusion, and a data loop into embodied AI. Public evidence in this pool does not establish realized Morphi pricing, customer retention, uptime, or deployment economics, so unknowns are retained rather than scored as advantages.[CP005, CP006, CP007, CP008, CP009]

3.3 Differentiation durability and displacement risk

A durable moat would require more than a capable demo: it would need reliable production, proprietary data from repeated real-world work, distribution or partner access, and safety and regulatory credibility. China's national standards system now covers commonality, computing, components, complete machines, applications, and safety and ethics, raising the importance of compliance execution. At the same time, the competitive evidence points to rapid commoditization pressure: AGIBOT is scaling, Unitree is pursuing a STAR Market listing, and US platforms such as Figure, Boston Dynamics, 1X, and Tesla are advancing commercialization. Geopolitical restrictions could also narrow the addressable customer set for Chinese vendors. Morphi's moat is therefore plausible but unproven until it discloses comparable production and field-performance evidence.[CP010, CP011, CP012, CP013, CP014, CP015]

3.4 Exhibits

4.1 Revenue model remains unproven

The assigned evidence does not provide a Moqi price sheet, customer contract, revenue disclosure, ARR, GMV, utilization rate, or revenue-recognition policy. That makes a conventional software-style ARR or revenue multiple inappropriate. The most decision-useful public comparator is Figure AI: Sacra describes a recurring deployment model in which customers avoid a large upfront purchase and expansion depends on fleet growth and geographic rollout, while a separate market profile still characterizes revenue as undisclosed or only estimated from early pilots. That distinction matters for Moqi: a prototype, shipment, pilot, or financing announcement is not revenue, and a list price is not realized revenue or gross margin. The current financial conclusion is therefore conditional: Moqi may ultimately monetize hardware sales, rental, service contracts, software, or a hybrid, but the public record supplied here cannot determine the mix or quality of those streams.[CI001, CI002, CI003, CI004]

4.2 Cost and unit-economics drivers

Humanoid economics are dominated by the conversion of engineering capability into reliable, repeatable field work. Figure's comparator evidence gives useful boundaries but not a Moqi forecast: one profile cites a target price of $20,000-$30,000 and a $1,000 monthly service concept, while also warning that utilization, reliability, and maintenance determine whether the arithmetic works. A separate market profile reports Unitree at more than 5,500 units shipped in 2025, while Humanoids Daily reports an AGIBOT rental offer of EUR 899 per day. These figures illustrate scaled supply and a service-price signal, but they are competitor observations rather than Moqi costs or realized prices. Moqi's gross margin, bill of materials, warranty reserve, service labor, manufacturing yield, depreciation, working capital, and customer acquisition cost remain unverified. A credible model therefore needs per-unit contribution after service and replacement costs, not just a robot sticker price or production target.[CI005, CI006, CI007, CI008, CI015]

4.3 Capital adequacy and underwriting verdict

No assigned source reports Moqi cash on hand, monthly burn, runway, debt, project-finance obligations, planned use of proceeds, or the trigger for its next financing. The refresh context identifies those as unresolved diligence items, but it is background rather than current evidence. External industry evidence indicates why the missing data is material: Figure's profile says its large financing provides runway, while China has introduced a national humanoid-robot standard system spanning computing, components, complete machines, applications, safety, and ethics. Export-control research separately describes restrictions that can constrain access to advanced computing and create scarcity or higher prices; CSET separately frames semiconductors as a key input to economic and scientific progress. For Moqi, capital adequacy must therefore be tested against hardware scale-up, inventory and tooling needs, compute access, certification and safety work, and geopolitical contingencies. The financial verdict is research-more rather than a quantified valuation call: there is insufficient evidence to estimate revenue quality, margin path, or runway.[CI009, CI010, CI011, CI012, CI013, CI014, CI016]

4.4 Exhibits

5.1 Product definition and maturity

The only company-specific product description retained in this pool is a personal profile for co-founder and CTO Qingqiu Huang. It describes Morphi Robot as building humanoids intended to understand, adapt, and care for people, and says the team has in-house capabilities across hardware, data, and algorithms. The same profile identifies Huang's prior Huawei work across perception, prediction, end-to-end driving systems, VLM/VLA, and AI infrastructure. This is useful founder and direction evidence, but it is not a product specification: it does not name a model, SKU, customer workflow, deployment date, production volume, benchmark, or certification. The product should therefore be treated as an early-stage platform thesis rather than a verified commercial system.[CE001, CE002, CE003, CE004]

5.2 Architecture direction and technical differentiation

A plausible architecture for Morphi's stated mission is a perception-to-action stack that combines visual and language representations with robot data and control outputs, but the company has not publicly disclosed enough detail to attribute such an implementation to Morphi. The RT-2 research provides a relevant technical benchmark: it co-fine-tunes a pretrained vision-language model on robotics and web data, represents actions as tokens, takes robot camera images, and directly predicts robot actions. In more than 6,000 trials, the reported system improved performance on previously unseen scenarios from 32% for RT-1 to 62%, and reached 90% in simulation on the Language Table suite. Those results demonstrate why VLA is a credible research direction, not that Morphi has reproduced them. Huawei's disclosed WEWA design offers a second comparator, pairing cloud generation of difficult scenarios with an on-vehicle multimodal action model. Huang's profile is consistent with relevant expertise, but no Morphi paper, benchmark, repository, or system diagram in the assigned pool closes the attribution gap.[CE005, CE006, CE007, CE008, CE009, CE015]

5.3 Deployment, controls, and diligence risks

The operating environment is becoming more structured but also more demanding. China's March 2026 national standard system covers commonality, intelligent computing, limbs and components, complete machines and systems, applications, and safety and ethics; it also addresses data lifecycle and model training and deployment. Industry research describes 2026 as a commercialization inflection point, with component capacity and cost decline becoming central to scale. These are ecosystem tailwinds, not evidence that Morphi has passed safety or deployment gates. The assigned evidence contains no Morphi test results for collision avoidance, uptime, battery endurance, remote support, privacy, cybersecurity, regulatory approvals, or production quality. U.S. semiconductor export controls also make compute availability and architecture choices a relevant dependency for a China-based AI developer, although the assigned evidence does not establish Morphi's own chip stack. A diligence process should request a named platform bill of materials, safety case, incident log, security threat model, deployment references, compute plan, and reproducible task-level benchmarks before underwriting technical maturity.[CE010, CE011, CE012, CE013, CE014, CE016]

5.4 Exhibits

6.1 Buyer, user, and use-case segmentation

The assigned evidence points to industrial manufacturing and logistics as the first commercially legible customer segments for Chinese humanoid robots. MarketsNXT describes automotive assembly, electronics manufacturing, logistics, healthcare, government, and consumer use cases, while DirectIndustry characterizes logistics and manufacturing as the first key deployment areas. These are different buying motions: an automotive or factory buyer is purchasing throughput, safety, integration, and serviceability; a showroom or hospitality operator is purchasing visitor interaction and labor substitution; a research customer is purchasing a platform for experimentation. No assigned source establishes which of these segments Moqi serves, who signs the contract, or whether the payer is a facility, integrator, distributor, or end customer. That missing buyer-to-use-case mapping is material because a general-purpose humanoid demonstration does not establish a repeatable procurement path.[CU001, CU003]

6.2 Adoption trajectory and named proof

Peer evidence shows that the category is moving beyond demonstrations in some cases, but the quality of proof varies materially. UBTECH's company-issued release reports several hundred Walker S2 units entering mass production and delivery, orders exceeding 800 million yuan since early 2025, and collaborations with named automotive, logistics, and electronics companies. DirectIndustry gives a more cautious read on the sector: it reports customer deployments for AgiBot while also quoting an academic expert who says many public examples should not be interpreted as robots ready for large-scale industrial deployment. Humanoids Daily reports AgiBot's 10,000-unit milestone and lists logistics, hospitality, and industrial environments, but the article is not evidence that Moqi has matched those volumes. The assigned sources contain no named Moqi customer, production pilot, unit shipped, utilization statistic, or independently verified outcome.[CU004, CU005, CU006, CU007, CU010, CU014, CU016]

6.3 Retention, expansion, and concentration

There is no disclosed evidence in the assigned pool for Moqi's net revenue retention, gross retention, churn, renewal rate, contract length, cohort behavior, customer satisfaction, or concentration. The peer material illustrates why these metrics matter: industrial deployments require integration into production lines and continuing operational support, while a robot used for training or a public demonstration may create data and publicity without recurring revenue. UBTECH describes a closed loop from application to delivery and iterative improvement, but that is a company description of its own operating model, not evidence about Moqi. The same distinction applies to market forecasts and shipment claims: category growth can support demand plausibility, but it cannot establish customer durability. Until Moqi discloses signed-customer counts, active deployments, repeat orders, and revenue concentration, expansion and key-account risk remain open.[CU008, CU009, CU013]

6.4 Customer diligence conclusion

The strongest conclusion is asymmetric. External evidence supports a real and increasingly commercial customer problem in Chinese factories, logistics, service locations, and training environments; it does not provide a customer proof point for Moqi. The sector also has an adoption-quality risk: DirectIndustry records expert caution that many demonstrations remain technology demonstrators, and SCIO describes a national standard system that only began covering the industry's lifecycle, applications, safety, and ethics in 2026. For an investor, the next evidence threshold is a source-verifiable customer ledger: named buyer, site, task, production-versus-pilot status, units, paid terms, uptime or utilization, outcome, renewal or expansion, and the denominator behind every aggregate count. Harvard's review of semiconductor export controls also frames compute access as a policy dependency for Chinese AI developers, not as evidence of customer traction. Until the customer proof package is available, Moqi's customer thesis should be treated as unproven despite favorable category-level demand signals.[CU010, CU011, CU012, CU013, CU015]

7.1 Ranked risk map

The highest-consequence risk is physical-world execution: a general-purpose humanoid must combine software, actuators, sensing, human interaction, and reliable maintenance in environments where failure can injure people or interrupt customer operations. China's 2026 national standard system spans the industrial chain and lifecycle and explicitly includes safety and ethics, while industry commentary describes production yield, site stability, and delivered cost as the next commercialization test. These sources establish the external bar, not Morphi's compliance status. Morphi's disclosed founder background is a capability positive, but it is not evidence of product safety, certification, field reliability, or commercial controls. The investment implication is that the risk premium should remain high until Morphi discloses test protocols, incident handling, deployment conditions, and measurable operating results.[CR001, CR002, CR003, CR004]

7.2 Regulatory, legal, and operating exposure

Regulatory exposure is moving from a future possibility toward an operating requirement. The State Council Information Office account says the national system covers commonality, computing, components, complete machines, applications, and safety and ethics, and was developed by more than 120 institutions and users. That breadth raises the likelihood that product design, model training, data handling, maintenance, and application-specific controls will be assessed together. Independent coverage of a whistleblower lawsuit against Figure AI also illustrates that humanoid-robot safety allegations can become litigation and reputational events even when the allegation concerns a competitor, not Morphi. No retained source establishes a Morphi lawsuit, recall, breach, or outage; that absence is a disclosure gap rather than evidence that none exists.[CR005, CR006, CR007, CR008]

7.3 Dependencies, financial resilience, and thesis-break triggers

Morphi is exposed to ecosystem and financing risks that are common to hardware-plus-AI businesses: component availability, compute and data access, manufacturing partners, customer integration, and enough cash to iterate through field failures. Public market intelligence describes a sector in which hardware scale is concentrating among a few leaders, while another industry account cites capital concentration, a large 2025 funding inflow, and the shutdown of some U.S. projects before mass production. Those figures are industry-level and should not be imputed to Morphi. A practical monitoring plan therefore needs cash runway, burn, unit cost, supplier concentration, deployment uptime, safety incidents, and paid pilot conversion. A thesis break would be a material safety event, an inability to meet applicable standards, repeated field failures, or financing that depends on an unverified valuation rather than operating proof.[CR009, CR010, CR011, CR012, CR013]

7.4 Diligence priorities

The public record supports a focused diligence agenda rather than a conclusion that Morphi has already failed any control. First, obtain the product safety case: hazard analysis, operating envelope, emergency stop design, human-robot interaction testing, incident log, and insurance. Second, test the commercialization model with named paid deployments, uptime, intervention rates, customer retention, gross margin by unit, warranty returns, and production yield. Third, map dependencies across chips, actuators, batteries, cloud or edge compute, training data, and contract manufacturers. Fourth, reconcile funding and valuation with cash on hand, monthly burn, runway, liquidation preferences, and milestone-linked financing. Until those records are supplied, residual exposure is high and the appropriate action is to track with explicit kill criteria rather than infer safety or traction from founder pedigree or sector momentum.[CR014, CR015, CR016]

7.5 Exhibits

8.1 Recommendation and entry discipline

The public record supports a valuation reference, not a fully underwritten price. S&P Capital IQ reports a CNY 1 billion angel round for Moqi Technology (Beijing) announced on July 6, 2026, with convertible preferred stock and more than ten participating venture institutions. Dealroom separately lists Moqi Intelligent Technology among July 2026 unicorns at $1 billion. Those are meaningful financing signals, but neither source supplies revenue, gross margin, customer count, cash balance, liquidation preferences, or a cap table. The recommendation is therefore track/research-more: the financing and investor participation are credible positive signals, while the absence of operating and preference data prevents a defensible intrinsic valuation or target return. Entry discipline should require verified commercial deployments, unit economics, runway, and security or governance terms before treating the round mark as investable value.[CV001, CV002, CV003, CV004]

8.2 Thesis and anti-thesis

The thesis is that Moqi is participating in a capital-rich humanoid-robotics market and reached unicorn status with a large syndicate, while China's standardization effort may reduce ambiguity around lifecycle, safety, and application requirements. The anti-thesis is that a financing mark can overstate realizable value when commercialization and terms are opaque. Independent comparables illustrate both sides: CNBC reports Physical Intelligence raised $400 million at a $2.4 billion post-money valuation, while TechNode reports Unitree exceeded RMB 12 billion post-money after Series C and had over RMB 1 billion of 2024 revenue. These benchmarks are not directly interchangeable with Moqi: they involve different stages, products, geographies, and disclosed operating evidence. Agility's own site makes the execution bar concrete by describing 65,000 hours of production experience and an on-site validation process, a level of deployment proof not publicly demonstrated for Moqi in the assigned evidence.[CV006, CV007, CV008, CV009, CV010, CV011]

8.3 Scenarios, downside triggers, and diligence

A scenario framework is more reliable than a point estimate because the available valuation anchor is a financing event rather than a disclosed operating multiple. The base case holds the reported $1 billion mark as a reference only and assumes no verified revenue or customer KPI uplift. The bull case requires evidence of repeatable paid deployments, production reliability, and a financing or strategic event that validates the mark without excessive dilution. The bear case assumes the mark compresses if commercialization slips, hardware and deployment costs remain high, or a down round exposes preference overhang. The most important diligence asks are the convertible preferred-stock terms, ownership and liquidation preferences, cash and burn, customer contracts, safety metrics, and production economics. A thesis-break trigger is a new financing below the July mark without offsetting operating proof; a positive trigger is independently verifiable recurring deployment revenue with retention and gross-margin evidence.[CV012, CV013, CV014, CV015]

8.4 Evidence boundaries

Public evidence is sufficient to establish a financing and a contemporaneous unicorn classification, but not sufficient to establish Moqi's current revenue, ARR, runway, customer concentration, ownership dilution, or exit readiness. Dealroom's list is a valuation-status reference rather than a statement of operating performance, and the S&P Capital IQ item is an announcement record rather than a cash-available calculation. The comparable-company sources provide context, not a mechanically transferable multiple. The recommendation should move only when the missing denominators and terms are obtained from the company, lead investors, filings, or customer references.[CV017, CV019]

8.5 Exhibits