Haomo Zhixing Technology Co., Ltd.

1.1 Identity, origin, and what Haomo was built to sell

Haomo should be understood as a Great Wall Motor spinout that tried to serve two adjacent but operationally different markets with one autonomy stack: passenger-car assisted driving and low-speed logistics autonomy. Public company-issued and independent profiles converge on the founding baseline. Haomo became an independent company in November 2019 after Great Wall separated its intelligent-driving unit, and public profiles place the company in Beijing. Its product ambition was broader than one ADAS module. The company repeatedly described a full-stack offering built around HPilot and NOH/NOA-style assisted driving for passenger cars, HPark and memory-driving features, low-speed autonomous delivery vehicles such as Little Camel or HDeliver, and later DriveGPT plus the MANA OASIS compute center as the technical layer intended to bind those products together. That breadth matters for diligence because Haomo was never only a feature vendor. It was trying to become an autonomous-driving platform supplier with exportable modules, open code-and-cloud collaboration, and a commercial path spanning both OEM programs and terminal logistics. The same breadth also created execution risk, because the company had to prove product-market fit in two very different buying motions at once.[CO001, CO002, CO008, CO009, CO026, CO030]

1.2 Leadership, control, and the capital stack behind the company

The capital and governance picture was always central to Haomo’s story because Great Wall was simultaneously founder, strategic customer, and control anchor. Public reporting shows that Haomo reached unicorn status in late 2021 after raising nearly RMB 1 billion in a Series A round from GL Ventures, Meituan, Qualcomm Ventures, Shoucheng Holdings, and JZ Capital. TechCrunch later reported a RMB 100 million first tranche of Series B in February 2024 led by Chengdu Wufa, while CB Insights indexed the business as Series B-II and estimated total disclosed funding at about $245 million. Those numbers are directionally positive but not perfectly reconciled, which is itself an important diligence signal. Later reporting from Yicai said Haomo had completed seven financing rounds and that Great Wall Motor chairman Wei Jianjun remained the actual controller with a 37% stake. Great Wall’s own 2025 annual report does not disclose the startup’s full cap table, but it does confirm Haomo Zhixing as an associate and related party and records substantial cumulative loss participation tied to the investment. Leadership stability deteriorated just as funding momentum slowed. Mid-2025 coverage reported the resignation of chairman Zhang Kai and departures by senior product and technology executives, suggesting the company entered its most difficult commercial phase without a settled leadership bench.[CO003, CO004, CO005, CO006, CO007, CO014]

1.3 Commercial proof, customer footprint, and the real scale achieved before the crisis

Haomo did achieve non-trivial public commercialization markers before the 2025 crisis. The strongest historical proof came from Great Wall vehicle deployment and Beijing logistics operations rather than from a broad third-party OEM roster. Company-linked and independent sources said HPilot or NOH products had reached more than 20 Great Wall-linked vehicle models, while the official 2023 DriveGPT release said Haomo had become a supplier for three automakers after previously selling mainly to Great Wall. On the logistics side, TechCrunch said Haomo’s delivery vans had ferried close to 300,000 grocery parcels for supermarkets in Beijing, and Logistics Tech Outlook described more than 200,000 cumulative orders by July 2023 across collaborations with Meituan, Wumart Dmall, and Dada Nexus. Friendly Auto’s 2025 summary added that Little Camel versions 2.0 and 3.0 had entered commercial operation across nine use cases, and that Haomo-powered passenger vehicles had logged more than 141 million assisted-driving kilometers by May 2024. These metrics show Haomo was more than a lab project. But the pattern is also narrow. Great Wall remained the central OEM distribution channel, the logistics proof clustered around Beijing pilot zones, and independent evidence of durable multi-customer revenue scale remained thin even before the shutdown reports surfaced.[CO010, CO011, CO012, CO013, CO019, CO020]

1.4 Milestones, distress signals, and the current state of the company

The milestone record splits cleanly into an ascent phase and a breakdown phase. The ascent runs from the 2019 spinout through early financing rounds, 2021 unicorn status, the April 2023 DriveGPT launch, and the October 2023 release of lower-cost HP170, HP370, and HP570 assisted-driving products. The breakdown phase becomes visible in 2025. KrASIA’s mid-2025 translated 36Kr reporting said Haomo had only two Hyundai passenger-vehicle projects left, had missed urban-NOA delivery targets, and had reduced its Little Camel sales target to around 50 units while focusing on clearing inventory rather than launching new models. The crisis sharpened in late 2025 when KrASIA, CarNewsChina, and Yicai all reported that employees were told not to report to work, that salaries were already in arrears, and that frozen accounts and possible litigation were being discussed internally. Great Wall’s own annual report strengthens the concern indirectly by showing that it had moved its public intelligent-driving narrative toward in-house Coffee Pilot 3 and 4 plus VLA and world-model systems, while recognizing further Haomo-related investment losses. The most important present-tense overview fact is therefore not simply that Haomo once reached unicorn status; it is that by August 2026 there is still no public official statement clearly confirming restart, restructuring, or liquidation.[CO015, CO016, CO017, CO019, CO020, CO021]

1.5 Exhibits

2.1 Defining Haomo’s real market boundary before sizing it

Haomo did not sell into one simple “autonomous driving market.” It sat at the intersection of two related but economically distinct demand pools. The first is China passenger-vehicle intelligent driving, especially L2+ to L2.9 assisted-driving systems for domestic OEMs that need a mix of sensors, compute, software, and safety validation at mass-market price points. The second is low-speed autonomous last-mile delivery, where robotics, campus and community routes, grocery delivery, and retail logistics create a different set of buyers, operating metrics, and regulatory constraints. Treating those as a single TAM would inflate the opportunity and obscure the sales reality. Passenger-vehicle ADAS depends on OEM platform wins, SOP timing, chip choices, and feature penetration in vehicle price bands. Low-speed logistics depends on route density, labor substitution, utilization, city permissions, and whether customers treat the vehicle as capex, service, or hybrid infrastructure. Haomo’s own product mix reflected that split: HPilot and NOH targeted OEM deployment, while HDeliver or Little Camel targeted logistics operators and retailers. The market chapter therefore uses two parallel lenses rather than one blended headline number.[CM001, CM002, CM003, CM004, CM005, CM006]

2.2 China ADAS is still growing, but the growth is moving up-stack and toward full-stack vendors

The broad China intelligent-driving market remains favorable in direction. Research-and-markets syndications and specialist industry outlets describe 2025-2026 as a structural upgrade phase in which non-intelligent and low-level configurations continue to fade while high-level assisted-driving functions become mainstream. The most useful numbers are not generic TAM figures but penetration and architecture changes. Independent OEMs moved from a market dominated by non-intelligent configurations in 2022 to one in which high-level intelligent driving made up 62.7% by 2025, with urban L2.9 NOA penetration rising from 2.1% to 17.2%. Another supplier-side market lens showed L2.5 and L2.9-equipped new models reaching 30.2% and 34.82% respectively by the first four months of 2025. Gasgoo and DVN then show 2026 scale moving into components and industrialization: LiDAR installations exceeded 1.3 million units in the first four months of 2026 and the supplier stack concentrated around a few large players. This is positive for category demand, but it is not neutral for startup positioning. Robotics and Automation News summarized IDTechEx research showing that software-only ADAS vendors lost ground as automakers prioritized full-stack and cost-effective partnerships. That shift is strategically bad for Haomo because its commercialization problems arrived exactly as the market rewarded broader ecosystems and tighter chip-software integration.[CM008, CM009, CM010, CM011, CM012, CM013]

2.3 Autonomous last-mile delivery is real, but it still behaves like an ROI and operating-model market

Haomo’s logistics business targeted a market with real structural demand, but one that remains unforgiving operationally. Grand View Research estimated the global autonomous last-mile-delivery market at about US$1.6 billion in 2024 and projected about US$5.9 billion by 2030, while DataM put the China market alone at roughly US$6.05 billion in 2026 and nearly US$26 billion by 2035. Those market numbers should not be treated as near-term revenue pools for Haomo; they are adoption ceilings shaped by hardware cost, urban density, food and grocery demand, and smart-city investment. The more useful readthrough is the operating logic. Ground vehicles dominate because they fit controlled, repetitive routes better than aerial systems, and China’s densest use cases remain campuses, business districts, residential compounds, supermarkets, and community logistics. DataM explicitly frames the market as entering a critical investment-timing phase in which utilization, regulatory approvals, operator ratio, battery economics, and insurance frameworks determine whether pilot deployments become scalable businesses. Logistics Tech Outlook’s Haomo profile fits that picture well: it showed real parcel movement and named customers, but still in geography- and use-case-bounded environments. That means the logistics opportunity stayed open, yet the bar for durable advantage was rising toward fleet economics and ecosystem integration rather than prototype novelty.[CM020, CM021, CM022, CM023, CM024, CM025]

2.4 Market implication: Haomo’s timing window narrowed as both markets became scale games

The market implication is not that Haomo picked the wrong categories. It picked categories that ultimately became harder for a mid-scale, strategically dependent startup to win. In passenger ADAS, the center of gravity shifted toward platforms that combine chips, software, safety, and broad OEM coverage, with Huawei, Horizon, and similar ecosystems defining the benchmark. In low-speed logistics, demand remained real but economics increasingly depended on repeatable route density, lower vehicle cost, and operational integration with giant platforms such as Meituan, Alibaba, JD, or specialized robot operators. Haomo had pieces of each story—low-cost HPilot kits for mainstream vehicles, named logistics pilots, and a research narrative around DriveGPT—but by 2025 it lacked evidence of broad OEM diversification or logistics scale strong enough to overcome its dependence on Great Wall. The underlying markets could therefore continue to grow even while Haomo’s own feasible SAM and SOM contracted. That divergence matters for valuation: investors cannot justify a bullish view by citing category growth alone when the company’s route to share capture has already been impaired by commercialization delays and a reported work stoppage.[CM030, CM031, CM032, CM033, CM034, CM035]

2.5 Exhibits

3.1 Competitive landscape: Haomo fought startups, incumbents, and substitutes at the same time

Haomo’s competitive set was broader than a simple startup leaderboard. In passenger vehicles, it had to win OEM sourcing cycles against Horizon, Momenta, Huawei-linked stacks, and other third-party intelligent-driving suppliers. In low-speed autonomy, it also had to prove delivery economics against robotaxi or robotruck operators whose technology, data, and capital bases could spill into adjacent commercial use cases. Just as important, Haomo faced substitute competition from Great Wall’s internal smart-driving roadmap, from incumbent global suppliers such as Mobileye, and from the status quo of human delivery plus conventional ADAS. That matters because Haomo tried to bridge two buying motions at once: car-program sourcing and logistics-operations automation. Public evidence suggests this multi-front competition became harder as the industry consolidated around a smaller number of scaled ecosystems with clearer chip, data, or distribution advantages. The most relevant comparison is therefore not “which startup had the best demo,” but which companies became hardest for OEMs and platform operators to avoid in real procurement decisions.[CP001, CP002, CP003, CP004, CP005, CP006]

3.2 China ADAS rivals moved toward integrated ecosystems with stronger OEM embed

Horizon, Momenta, and Huawei illuminate why Haomo’s passenger-car position weakened. Horizon’s public materials show a tightly integrated story: Journey 6 silicon, HSD assisted-driving software, reference tools, and segmented packages for RMB100k+, RMB150k+, and RMB200k+ vehicle bands. Momenta’s public partnerships and third-party coverage show another strong model: an end-to-end data flywheel sold through many OEM programs rather than through one anchor customer. Huawei’s Qiankun ADS shows a third path, combining sensor redundancy, local data density, and ecosystem power large enough to influence supplier share at the market level. These rivals differ in technical route, but they converge on the same commercial advantage: they became difficult for automakers to ignore because they offered either chip-plus-stack integration, wide model coverage, or both. Haomo’s low-cost HPilot messaging looked strategically sensible, yet by 2026 it was competing against larger rivals that could also segment products by vehicle price tier while bringing more production wins, more capital, and richer data loops.[CP007, CP008, CP009, CP010, CP011, CP012]

3.3 L4 operators and global benchmarks widened the gap in breadth and credibility

Haomo’s logistics and autonomy narrative must also be compared with companies whose scale lives outside Great Wall’s ecosystem. Pony.ai’s official materials show a company spanning robotaxi, robotruck, and personally owned vehicle smart-driving solutions, while WeRide’s platform description emphasizes L2-to-L4 reuse across mobility, logistics, and sanitation. DeepRoute’s 2026 IO 2.0 launch adds another pressure point: even newer peers are now speaking the language of VLA, multi-sensor flexibility, and secured OEM partnerships. Mobileye remains the global benchmark for embedded ADAS scale rather than a direct China-delivery rival, but that benchmark still matters because it sets expectations for mapping, data volume, and long-cycle OEM trust. The cumulative effect is that Haomo was not only outrun by one domestic rival. It was being benchmarked against several different models of competitive strength—consumer-vehicle scale, robotaxi fleet operations, vertically integrated mapping and ADAS ladders, and multi-product reuse—without clear evidence that it led any one of those dimensions by 2026.[CP018, CP019, CP020, CP021, CP022, CP023]

3.4 Moat durability depended on production lock-in, data loops, and capital staying power

The strongest competitive lesson is that autonomous-driving moats are now operational rather than rhetorical. A supplier that wins a production program gains validation history, software-hardware tuning, and recurring road data that make later replacement expensive. The suppliers that appear strongest in 2026—Horizon, Momenta, Huawei, Mobileye, Pony.ai, and WeRide—each have at least one durable moat layer in public view: OEM embed, chip control, data scale, global fleet operation, or financial resilience. Haomo’s public evidence base shows something narrower: historical commercialization tied heavily to Great Wall and Beijing pilots, a cost-conscious product portfolio, and a technical story that may have been promising but no longer sat inside the leading ecosystem narrative. Once trust deteriorated through management churn, salary-arrears reports, and the operations halt, even a technically credible stack became far harder to underwrite for new vehicle programs. The resulting moat assessment is weak: Haomo lacked a clearly durable competitive layer that looked stronger than customer concentration and execution risk.[CP033, CP034, CP035, CP036, CP037, CP038]

3.5 Exhibits

4.1 Revenue model: multiple monetization paths were visible, but real revenue quality was not

Haomo’s public materials show a mixed revenue model rather than a single pure-software business. On the passenger-car side, the company sold ADAS kits and broader smart-driving solutions into OEM programs, with HP170, HP370, and HP570 positioned as price-tiered products for low- to high-end vehicles. On the logistics side, Haomo marketed HDeliver and related low-speed autonomous vehicles, suggesting hardware revenue, deployment revenue, and possibly continuing operations or support fees. Public media and company-linked coverage also describe code, modules, hardware, software, and cloud services, implying that Haomo likely monetized a bundle of product and integration work rather than one recurring SaaS fee. Traction signals existed—20-plus vehicle models, roughly 140 million assisted-driving kilometers by spring 2024, and close to 300,000 Beijing grocery deliveries—but these are activity proxies, not audited revenue. The financial interpretation is therefore mixed: Haomo had real commercial surfaces, yet public evidence still does not show how much revenue came from product sales versus support services, nor whether any of those streams carried attractive margins or durable retention.[CI001, CI002, CI003, CI004, CI005, CI006]

4.2 Cost structure and unit economics: Haomo looked capital-heavy even before the halt

The cost picture can only be reconstructed indirectly, but the direction is clear. Haomo’s disclosed uses of funds centered on large models, compute, data, and mass production—not on near-term profitability. Its product mix also implies inherently different margin structures: mainstream ADAS kits may support decent gross margin if sold at scale, while autonomous delivery vehicles, domain controllers, sensor stacks, and fleet support generally drag the business toward hardware-like economics and working-capital needs. Comparable listed AV companies underline the problem. WeRide’s 2025 revenue of about US$97.9 million still sat beneath roughly US$196 million of R&D and a US$264 million operating loss, while Pony’s 2025 and Q1 2026 disclosures show rapid revenue growth coexisting with heavy operating investment, vehicle BOM optimization, and new capex. That does not prove Haomo’s exact burn rate, but it does show the surrounding category economics: even stronger and better-capitalized autonomy companies were not close to easy self-funding. Haomo’s combination of Great Wall concentration, logistics hardware, and frontier-model ambition likely meant a negative cash-conversion profile unless an anchor customer or new round kept absorbing the cost of scale-up.[CI011, CI012, CI013, CI014, CI015, CI016]

4.3 Capital adequacy: the public record implies financing dependency, then liquidity stress

The strongest supportable financial conclusion is that Haomo still depended on external funding or strategic support through 2024 and probably into 2025. Yicai reported that Haomo raised over CNY100 million in B1 and CNY300 million in B2 in early 2024, with proceeds earmarked for mass production, robotics expansion, and continued spending on large models, compute, and data. That use-of-funds language is growth-investment language, not mature free-cash-flow language. Late-2025 reporting then turned sharply adverse: layoffs, salary arrears, halted work, theories of frozen accounts, and no public post-crisis clarification on cash, liabilities, or restart. Great Wall’s 2025 annual report confirms Haomo remained an associate and related party, but it does not give public investors the clean runway disclosure that would let them underwrite ongoing solvency. Relative to public peers, that opacity is severe. Pony and WeRide can disclose cash balances, revenue composition, capex, and losses; Haomo cannot. The practical implication is that Haomo’s capital adequacy should be treated as unresolved-to-weak, with the burden of proof on management to show that the business still has operating liquidity, not just historical funding headlines.[CI001, CI012, CI013, CI014, CI015, CI016]

4.4 Financial verdict: underwriting is blocked by missing basics, not by a lack of category demand

Haomo’s financial verdict is negative because the missing data are the exact inputs needed for underwriting. Investors do not have public revenue, gross margin, receivables aging, cash balance, debt, covenant status, backlog, renewal rates, customer concentration by revenue, or monthly burn. Nor is there evidence that the company converted its impressive commercialization narrative into stable, diversified, high-quality earnings. The public record instead points to a familiar autonomy-company pattern: high R&D intensity, product-plus-service complexity, possible working-capital drag, and constant need for either new funding or powerful strategic sponsorship. That pattern is survivable when a company remains on a scaling path and can show balance-sheet strength, but Haomo’s late-2025 distress reporting means the same pattern becomes dangerous. The consequence for diligence is straightforward. Until management provides current cash, liabilities, active contracts, unit economics, and customer-payment data, the only responsible financial stance is that Haomo was a capital-intensive business with low reporting visibility and unresolved solvency risk by the 2026 diligence date.[CI031, CI032, CI033, CI034, CI035, CI036]

4.5 Exhibits

5.1 Product surface: Haomo sold driver assistance and autonomous delivery into very different workflows

Haomo’s product map split across two different customer jobs. In passenger vehicles it offered HPilot and related NOH or NOA, memory-driving, parking, and safety-assistance features sold as tiered ADAS kits or broader full-stack solutions for OEMs. In logistics it offered HDeliver and the Little Camel family for short-route autonomous delivery, aimed at supermarkets, courier-type flows, campuses, airports, and other controlled environments. Public company releases stressed that Haomo was not only selling one boxed feature. It described a “6P” cooperation model spanning code, module, software, hardware, cloud services, and full-stack schemes, which implies a willingness to meet OEMs at multiple integration layers. That breadth is strategically relevant because it gave Haomo more ways to participate in customer workflows, but it also forced the company to mature two very different delivery models at once: automotive SOP-grade ADAS and operations-heavy logistics autonomy. The resulting product story is broader than many startup narratives, but the public record still leaves open how much of this surface was fully deployed versus still partly aspirational. That ambiguity is especially important for diligence because module breadth alone does not reveal whether Haomo could still service, patch, and evolve each workflow in 2026.[CE001, CE005, CE006, CE010, CE011, CE012]

5.2 Architecture and technical differentiation: DriveGPT, sensor tiers, and research tooling formed the core technical narrative

Haomo’s technical narrative centered on making assisted driving cheaper, more capable, and more explainable at the same time. The 2023 AI Day releases described DriveGPT as a GPT-style autoregressive planning model built on a “Drive Language,” with RLHF-like human-feedback training intended to make decisions safer, smoother, and more interpretable. The HP170, HP370, and HP570 product tiers show that this narrative was grounded in concrete sensor and compute bundles rather than in abstract AI branding alone: the three SKUs used materially different compute budgets, cameras, radars, parking capability, and optional LiDAR. Third-party explanation of DriveGPT emphasizes that a planning model still depends on high-quality perception inputs, which fits Haomo’s own 6P framing and the likely role of infrastructure such as MANA OASIS and upstream compute partners. Meanwhile, Haomo’s GitHub presence suggests real practitioner output around LiDAR-camera fusion, occupancy forecasting, and place recognition. The important technical conclusion is that Haomo looked like a company with meaningful research capacity and a plausible architecture thesis—but one still dependent on sensors, chips, data pipelines, compute partners, and vehicle-program execution rather than on model novelty alone.[CE002, CE003, CE006, CE007, CE008, CE009]

5.3 Maturity, roadmap, and trust: shipped enough to matter, but not enough to remove support risk

Haomo’s public roadmap contains both evidence of shipping and evidence of fragility. On the positive side, official releases show that HPilot had reached close to 20 models by April 2023 and more than 20 models by 2024, that DriveGPT was attached to a mass-production vehicle narrative, and that Little Camel 2.0 and 3.0 were said to be in commercial operation across multiple use cases. The company also claimed export progress into the EU, Israel, and later planned regions. But maturity is not the same as continuity. Later reporting said city-NOA targets slipped, Hyundai delivery programs were delayed, and the company’s active commercial surface may have narrowed sharply by 2025. Trust and compliance evidence is similarly uneven. Haomo publicized specific safety or hardware claims—such as five-star AEB for HP170 and HD-map independence—yet retained sources do not show a current public reliability dashboard, support SLA, software-release log, cybersecurity program, or post-crisis quality statement. For buyers, that gap matters: automotive and logistics customers need long-lived support and compliance proof, not just technically credible features. The maturity verdict is therefore mixed-to-adverse: historically nontrivial, presently uncertain. Another practical implication is documentation depth: there is enough public material to map the 2023 stack, but not enough current material to verify whether buyers still receive active releases, issue remediation, or long-term maintenance. In a safety-sensitive domain, that missing operational evidence narrows the value of otherwise impressive feature disclosures.[CE004, CE017, CE019, CE020, CE023, CE024]

5.4 Exhibits

6.1 Customer segmentation: the real payer was usually an automaker or platform operator, not the end user

Haomo served at least two distinct customer motions. In passenger vehicles, the direct payer was the automaker or vehicle-program team buying an ADAS stack, while the driver was only the downstream user. In low-speed logistics, the payer was a retailer, supermarket operator, instant-delivery platform, or scenario owner deciding whether autonomous delivery could lower labor pressure and maintain service levels. That distinction matters because logo count can overstate diversification. Great Wall was not just one customer among many; it was the company’s historical parent ecosystem, largest visible production channel, and likely core source of commercialization credibility. The logistics roster added important proof that Haomo technology reached real operating routes, but those relationships look more like programmatic route deployments than like a broad, transparent recurring-revenue base. The diligence question is therefore not whether Haomo ever had customers. It did. The harder question is how much of the economic value sat inside Great Wall and a small number of Beijing-centered delivery programs rather than inside a diversified set of independent contracts.[CU001, CU002, CU003, CU004, CU005, CU006]

6.2 Adoption trajectory and named-customer proof: Great Wall and Beijing delivery programs are the core evidence

The most credible adoption evidence is concentrated in two places. First, multiple 2024 and 2025 sources say HPilot had already been equipped in more than 20 vehicle models and mostly sold through the Great Wall ecosystem, with assisted-driving mileage above 120 million kilometers by February 2024. Second, Haomo’s autonomous-delivery programs produced real order-flow evidence in Beijing. Independent and company-linked sources cite nearly 300,000 grocery parcels by early 2024, more than 200,000 cumulative orders by July 2023, and a Wumart Dmall route base spanning more than 60 communities in Shunyi. Logistics Tech Outlook and EqualOcean also tie those deployments to Meituan, Dada Nexus, Wumart Dmall, and Alibaba. That said, the quality of proof varies by logo. Great Wall is clearly production-grade. Wumart Dmall has specific route and order outcomes. Meituan has strong strategic logic and very large autonomous-delivery scale as a platform, but public sources do not quantify how much of Meituan’s live fleet specifically used Haomo vehicles. Dada and Hyundai are weaker still: both are publicly mentioned, but contract economics and current deployment continuity remain under-documented.[CU010, CU011, CU012, CU013, CU014, CU015]

6.3 Retention, expansion, and concentration: the expansion logic is visible, but durability is mostly inferred

Haomo’s public customer record supports an obvious land-and-expand theory but offers very little true retention disclosure. In ADAS, the natural expansion path is one OEM program growing into more models, more trims, and more lifetime vehicle volume. In delivery, the expansion path is one district or route growing into more communities, more use cases, and denser order throughput. Those pathways appear in the public record, but investors still do not get the underwriting metrics that matter most: NRR, GRR, churn, contract length, renewal timing, customer-level revenue, or attach rates by customer. The best proxy for durability is Great Wall continuity, because Haomo remained an associate and related-party ecosystem name in Great Wall’s 2025 annual reporting. Yet that same fact also creates concentration risk. If Great Wall drove most ADAS economics and the public logistics roster stayed narrow, then customer stickiness and customer concentration rise together. The late-2025 layoffs and work-stoppage reports make the problem sharper: even if historical customer proof was genuine, present-day continuity and support quality can no longer be assumed.[CU028, CU029, CU030, CU031, CU032, CU033]

6.4 Customer verdict: real pre-crisis proof, weak diversification, and major freshness gaps in 2026

The customer verdict is not that Haomo lacked market adoption. It is that the adoption evidence never graduated into a clean, diversified, continuously verified customer base. Great Wall is the one relationship that clearly reached production depth. Wumart Dmall and adjacent delivery partners provide persuasive evidence that Haomo solved real last-mile problems in Beijing routes. But the public record remains too thin on customer economics, too concentrated around a handful of relationships, and too stale after the 2025 operating crisis to support a strong durability conclusion. The Hyundai thread is especially revealing: a late-2024 article described Haomo integration into a Beijing Hyundai EV, but official Hyundai 2026 materials named Momenta instead. Likewise, current Dmall recognition proves the retail counterparty still scales AI operations, yet it does not prove Haomo still powers those workflows. For diligence purposes, customer traction should be scored as historically proven, current continuity as unresolved, and diversification as weaker than Haomo’s headline logo set first suggests.[CU036, CU037, CU038, CU039]

6.5 Exhibits

7.1 Regulatory and legal risk: China now expects ADAS and autonomy companies to behave like safety-critical manufacturers

The most important contextual change for Haomo is that China’s regulatory environment for intelligent driving became materially stricter just as Haomo’s own operating condition worsened. The February 2025 MIIT and SAMR notice tightened product admission, recall, incident reporting, and OTA governance for intelligent connected vehicles with combined driver-assistance systems. Those rules matter directly to Haomo because its historical pitch depended on frequent software iteration, evolving NOA capability, and safety-sensitive automation claims. The new framework requires fuller testing and verification, clearer system boundaries, record filing, and recall-style handling when OTA changes fix safety defects. Beijing’s autonomous-vehicle regulation and the 2026 cross-border automotive-data guide further show that public-road and data-governance expectations are rising, not falling. Then the Xiaomi SU7 crash accelerated enforcement pressure on marketing language and rollout caution across the sector. For a company already in distress, this is dangerous: a restart would not happen in a permissive policy window but under a more demanding safety, disclosure, and compliance regime than the one Haomo scaled under in 2023 and 2024.[CR001, CR002, CR003, CR004, CR005, CR006]

7.2 Operational and customer-continuity risk: the late-2025 halt turns every historical traction signal into a support question

Haomo’s operational risk is best understood as a continuity problem. Before the crisis, the company had real adoption signals: more than 20 vehicle models, over 120 million assisted-driving kilometers, and large delivery-order counts in Beijing retail routes. But once public reporting says work halted, salaries fell into arrears, and hundreds of employees faced uncertainty, every historical proof point becomes a new risk question. Who is still patching software? Who is answering incident escalations? Which customers still receive support? What happens to warranties, route operations, or ADAS issue remediation if the operating team has been cut back? The customer chapter already showed that Haomo’s base looked concentrated before the halt; concentration makes a disruption more severe because too much of the installed base may sit with a small number of counterparties. Current Dmall recognition and Meituan fleet scale prove the counterparties remain strategically relevant, but they do not prove Haomo remains the active provider. In a safety-sensitive category, the absence of current support proof is itself a major risk event.[CR015, CR016, CR017, CR018, CR019, CR020]

7.3 Partner, compute, and people risk: Haomo depends on counterparties it does not fully control

Haomo’s dependency stack spans customers, regulators, chips, and people. Great Wall is both the anchor customer and the control ecosystem, which gives Haomo historical credibility but also means any change in Great Wall sourcing strategy could damage revenue, data access, and strategic relevance at the same time. The public record already shows Great Wall working with other intelligent-driving suppliers, so Haomo cannot assume exclusivity. On the technology side, Haomo’s large-model narrative around DriveGPT increases exposure to compute availability and cost. Nvidia’s 2025 China export restrictions are not abstract macro noise; they show that advanced AI compute capacity can become constrained, delayed, or materially more expensive. That matters more for a company pitching frontier-model autonomy than for one selling only mature low-cost L2 kits. The people layer is equally serious. Layoffs and halted operations increase attrition risk for senior engineers and product leads, while China’s more explicit 2025 non-compete guidance means retaining technical staff through restrictive covenants is not frictionless. A weakened team, contested IP defenses, and an uncertain compute path together create a high execution-risk bundle.[CR025, CR026, CR027, CR028, CR029, CR030]

7.4 Financial/model risk and kill criteria: missing restart proof is now as important as strategy

The financial risk here is not merely that Haomo was capital intensive. It is that public evidence no longer shows a functioning financial bridge from historical traction to current survivability. There are no public audited 2025 or 2026 Haomo financial statements, no disclosed cash balance, no debt schedule, no customer receivables aging, and no enforceable Great Wall funding backstop in the public record. At the same time, Haomo’s model remained structurally demanding: low-cost ADAS, autonomous-delivery vehicles, and frontier-model R&D each pull cash and management attention in different ways. ML Vehicle’s reporting that Little Camel 2025 sales targets were low and Hyundai-related delivery timing slipped suggests the business was already struggling to convert product breadth into scaled, reliable commercial execution before the halt. That leaves a clear IC conclusion. Until management can prove restarted operations, current cash, active customers, and compliance readiness, the burden of proof sits entirely with the company. In practical diligence terms, those missing facts are not ordinary information gaps; they are thesis-break triggers.[CR036, CR037, CR038, CR039, CR040, CR041]

7.5 Exhibits

8.1 Recommendation: avoid unless entry is explicitly distressed and backed by fresh diligence

The right present-tense call is avoid, not because Haomo never built anything real, but because the public evidence no longer supports underwriting it like an operating growth company. Haomo did achieve meaningful milestones: unicorn status in 2021, additional B1 and B2 funding in 2024, deployment into more than 20 passenger-vehicle models, and last-mile delivery volume measured in hundreds of thousands of parcels or orders. In a different context those facts could support a track recommendation. But by late 2025 the story changed from commercialization to continuity. Multiple outlets reported that work stopped, salaries were in arrears, and company accounts may have been frozen. Great Wall’s 2025 annual report still shows Haomo as an associate with accumulated losses, yet there is no public bridge from that disclosure to a restart, recapitalization, or clean residual-equity story. That gap matters more than market size or product ambition. Investors are not being asked to price only autonomy software optionality; they are being asked to price whether the underlying operating entity still has enough team, liquidity, compliance readiness, and customer support capacity to matter. Without new diligence that proves those points, the only defensible stance is to assume heavy impairment versus the old unicorn narrative.[CV001, CV002, CV004, CV005, CV018, CV019]

8.2 The key valuation fact is the break between the historical unicorn mark and the current distress signal

Haomo’s historical price anchors are easy to describe but unsafe to reuse. The 2021 Series A created a unicorn narrative at more than $1 billion valuation, and the company still raised more than CNY400 million across B1 and B2 in 2024. If the company had then shown stable 2025 revenue growth, current customer continuity, and sponsor-backed liquidity, those legacy marks would still matter. Instead, the public record shows the opposite direction. By mid-2025 commercialization problems were already visible in weak Hyundai diversification, reduced low-speed vehicle targets, and organizational turbulence. By November 2025, multiple reports described a halt, salary arrears, and operational paralysis. Great Wall remained the actual controller ecosystem and its annual report still carried Haomo-related losses, but the same filing also emphasized Great Wall’s own Coffee Pilot, VLA, and world-model path rather than a simple Haomo rescue narrative. That combination makes legacy pricing stale. Historical investors were funding a scaling autonomy supplier; current investors would be funding a restructuring, rescue, or asset-preservation story. Those are fundamentally different valuation problems, and the discount between them should be very large.[CV001, CV002, CV004, CV006, CV007, CV008]

8.3 Public comparables define a ceiling, not a shortcut to value

Pony.ai, WeRide, and Mobileye are useful, but mainly as proof of what Haomo is missing. In August 2026 Pony.ai and WeRide still carried multi-billion-dollar market caps despite modest revenue because the market could at least see current operations, audited or filing-backed disclosure, large cash balances, and ongoing commercialization momentum. Mobileye traded at a much higher absolute value because it combined scale, cash generation, and a mature ADAS installed base. Haomo has no comparable disclosure stack. There is no public 2025 or 2026 revenue statement, no disclosed cash balance, no visible cap-table update, no clear support backstop, and no verified restart announcement. That means public-comp math should be used only to cap optimism. If live and capitalized AV peers with real financial disclosure are worth roughly $1.9 billion to $6.8 billion in public markets, then a distressed private company with unresolved continuity should trade at an extreme discount to them, not at a narrative premium. The fact that CompaniesMarketCap and SEC-linked filing directories can point investors directly to peer annual reports underscores the contrast: Haomo’s problem is not a lack of industry comparables but a lack of current company-level observability.[CV009, CV010, CV011, CV012, CV013, CV014]

8.4 Scenario valuation should be framed as distressed optionality, not precision pricing

Because current financials are absent, valuation has to be scenario-based. The bear case is not normal multiple compression; it is liquidation or an asset sale in which wage claims, liabilities, preference rights, and lost customer confidence leave little or no residual for common equity. The base case assumes a controlled rescue, most plausibly through Great Wall or another strategic buyer, in which some engineering assets, customer relationships, and support obligations survive. That scenario can justify a non-zero equity value, but still at a severe haircut to the historical unicorn mark because the company would be restarting from distress under tougher regulation. The bull case is not simply that autonomy sentiment improves. It requires documented restart, fresh capital, proof that core customers are still live and supported, and evidence that the compliance stack for OTA, data, and safety is restart-ready. Only then could Haomo move from distressed optionality toward a late-stage growth valuation again. Even in that better case, the public evidence today supports a value far below the old headline mark because the continuity break has already occurred.[CV022, CV023, CV027, CV028, CV029, CV030]

8.5 What would move the recommendation: proof of solvency, support continuity, and sponsor economics

This recommendation can change, but only through concrete evidence that closes the biggest underwriting gaps. The first requirement is a current balance-sheet package: cash, debt, payables, payroll arrears, contingent liabilities, and any litigation or account-freeze history. The second is customer continuity: which Great Wall, Wumart/Dmall, Meituan, Dada, or other programs are still active, and who is servicing them today. The third is governance and capital structure: whether Great Wall or another sponsor has enforceable rescue obligations, what preferences sit ahead of common equity, and whether minority investors are already economically out of the money. The fourth is regulatory readiness under the stricter 2025-2026 regime: evidence of OTA controls, incident response, product-admission status, and automotive-data compliance. If those materials show a funded restart with active customers and a clean rescue structure, the fair range can move up sharply from distressed levels. Until then, the chapter’s discipline remains simple: do not confuse historical commercialization proof with current investability.[CV021, CV030, CV034, CV035, CV036, CV037]

8.6 Exhibits