Excelland Robot Files for a 45 Million-Share Hong Kong IPO as Losses Test Its Scale Story

Excelland Robot has opened a global offering of 45 million H shares, despite remaining unprofitable after years of commercial robot deployments. The offering places its scale claims under a public-market spotlight. Trading is expected to begin on the Hong Kong Stock Exchange on September 9, 2026.

The Wuxi-based company is offering 2.25 million shares through Hong Kong’s public offering and 42.75 million through its international placement. Those allocations represent 5 percent and 95 percent of the initial offering, respectively. Both portions remain subject to reallocation, according to the published offering details.

The immediate story concerns an IPO. The more consequential question concerns Excelland Robot’s business model. It has sold robots at considerable scale and expanded into software, leasing, and robotics as a service, or RaaS. Yet its margins remain narrow, its operating cash flow remains negative, and its latest quarterly loss widened.

That tension separates this listing from a simple robotics funding event. Excelland Robot is asking investors to value a broad commercial deployment platform before that platform has established sustainable profitability.

Its challenge also differs from the one facing humanoid specialists such as UBTECH Robotics or Unitree Robotics. Excelland Robot’s machines already perform routine work in hotels, offices, clinics, shopping centers, and controlled campuses. The listing must show that practical deployment can produce stronger economics, not merely more units in service.

The 45 Million-Share Offering Puts Deployment Scale on Trial

Excelland Robot’s offering converts its operating record from a private growth story into a measurable public-market commitment.

The subscription period runs from August 31 through September 4. The company expects to determine the final offer terms on September 7 and announce the allocation results before trading begins. Its H shares are scheduled to trade under stock code 03231.

The international placement dominates the offering. Only 2.25 million shares are initially reserved for Hong Kong’s public market, while institutional and other international investors receive 42.75 million. That structure makes professional demand an important early signal of confidence in the company’s financial path.

Excelland Robot also secured cornerstone participation associated with SenseTime and 58 Group. Cornerstone investors commit to buying an agreed allocation under specified conditions and normally accept a lockup period. Their participation can provide initial demand, but it does not eliminate pricing or execution risk after listing.

The company plans to direct approximately 48.5 percent of net proceeds toward research and development. Intended projects include cloud-based vision-language models, cloud data feedback systems, vision-language-action technology, and dexterous robotic hands.

Another 30 percent is intended for business expansion and strategic acquisitions. Sales and marketing would receive approximately 7 percent, while 4.5 percent would repay part of the company’s bank debt. The remaining 10 percent is designated for working capital and general corporate purposes, according to the reported proceeds allocation.

This allocation reveals the IPO’s central bet. Excelland Robot is not presenting its existing product portfolio as a finished business. It wants public capital to finance another round of technical development while also expanding distribution and considering acquisitions.

The absence of an over-allotment option adds another detail worth watching. An over-allotment option, often called a greenshoe, lets underwriters issue additional shares and can support post-listing price stabilization. Reporting on the offering indicates that Excelland Robot has not included that mechanism.

None of these details determines how the shares will perform. They establish what investors are being asked to underwrite: continued technical spending, broader commercialization, and a path toward better unit economics that remains incomplete.

Excelland Robot Has Scale, but Scale Has Not Delivered Profit

The company has already moved beyond pilot deployments, yet the financial benefits of that reach remain limited.

By March 31, 2026, Excelland Robot had sold more than 114,600 robots to over 5,100 customers worldwide. Its annual customer count increased from 1,793 in 2023 to 3,358 in 2025, based on prospectus data summarized by market reporting.

Those machines serve settings where repetitive movement has a clear operational purpose. Delivery robots can transport meals, parcels, supplies, or linens inside hotels and office buildings. Cleaning units can cover defined indoor and outdoor routes. Vending robots can bring retail inventory closer to customers without requiring a fixed counter.

Excelland Robot’s portfolio includes machines designed for indoor delivery, high-capacity transport, enclosed-area delivery, cleaning, guest guidance, and unmanned retail. It also sells robot modules and provides industry-focused computer vision through its Youxiaoyun platform.

This breadth matters because many commercial environments need several robot types. A hotel operator might want room delivery, floor cleaning, and guest navigation without managing unrelated systems from several vendors. A shared software and service layer can reduce integration work if the products perform reliably.

According to Frost & Sullivan data cited in the company’s listing materials, Excelland Robot ranked third among Chinese commercial delivery robot suppliers by 2025 revenue. Its reported market share was 9.8 percent. It ranked fourth in commercial cleaning robots, with an 8.8 percent share.

The company also ranked third in China’s broader commercial service robot products and solutions market. That ranking came with an 8.9 percent revenue share. These are company-disclosed third-party estimates, rather than independently audited measures of installed market share.

Customer retention offers another encouraging signal. Reported customer retention increased from 25.1 percent in 2023 to 43.4 percent in 2025. Net revenue retention rose from 100.1 percent to 120.8 percent over the same period.

Net revenue retention measures how revenue from an existing customer group changes after expansions, reductions, and departures. A result above 100 percent means spending by retained and expanding customers more than offset lost revenue within that group.

However, deployment volume should not be confused with profitable scale. A large installed base creates maintenance obligations, support costs, and upgrade expectations. It produces operating leverage only when recurring revenue and product margins grow faster than the costs of serving that base.

That distinction creates pressure for Excelland Robot. Its listing story depends less on proving that service robots can find users. The company has already supplied evidence of adoption. It must now prove that those deployments can support a durable public company.

Recurring Services Are the Answer to Weak Hardware Economics

Excelland Robot’s most important transition is from selling machines toward earning more revenue throughout each machine’s operating life.

Robot product sales generated approximately 42.5 percent of the company’s 2025 revenue. Industry-focused computer vision solutions contributed about 37.5 percent. RaaS arrangements accounted for roughly 15.7 percent, while modules and repair services supplied a smaller share.

RaaS lets customers use robots through service agreements instead of purchasing every machine outright. The model can reduce upfront adoption barriers for customers while giving the supplier recurring revenue. It also places more responsibility for maintenance, utilization, and asset performance on the provider.

That tradeoff is central to Excelland Robot’s IPO. Hardware sales can create revenue quickly, but competitive markets often constrain margins. Recurring contracts can increase lifetime customer value, although they require capital and disciplined fleet operations.

Excelland Robot’s AI vision business provides another route beyond basic hardware. The company offers application-specific visual models that help systems interpret objects, people, and operating conditions. Vision-language models connect visual information with language-based instructions, giving robots a richer way to understand tasks and environments.

The company says it has moved from navigation systems led by lidar-based simultaneous localization and mapping toward sensor fusion. Sensor fusion combines information from cameras, lidar, and other inputs to improve positioning and perception.

It also describes an architecture that connects on-device vision models with cloud-based vision-language models. On-device systems can respond without waiting for a remote service, while cloud models can support broader interpretation and fleet-level learning.

That approach can spread development costs across several product lines. For example, localization work developed for a cleaning robot can potentially support a delivery model after adaptation. Reusing perception, navigation, and fleet-management components would shorten development cycles.

The financial evidence remains mixed. Overall gross margin improved from 6.9 percent in 2023 to 14.3 percent in 2024, then slipped to 13.9 percent in 2025. It declined further to 11.2 percent during the first three months of 2026.

Some individual businesses moved in a more favorable direction. The reported margin for robot product sales increased from 5.4 percent in 2023 to 9.1 percent in 2024. The company’s AI vision solutions reached an 18.9 percent margin in 2025, while RaaS moved from a gross loss to a gross profit.

These changes support the strategy, but they do not prove its success. Software, services, and leasing must become large enough to lift consolidated margins. Otherwise, hardware competition and service costs will continue to control the company’s economics.

Commercial Service Robots Face a Different Race Than Humanoids

Excelland Robot is competing for operational reliability while much of the robotics market competes for technical spectacle.

UBTECH Robotics has focused substantial attention on humanoid systems for industrial and service applications. Unitree Robotics is associated with quadruped and humanoid platforms. Dobot and other specialists target collaborative industrial automation, while Pudu Robotics and Keenon Robotics compete in commercial delivery and service environments.

Excelland Robot sits closer to Pudu and Keenon than to humanoid-first companies. Its machines address structured jobs with measurable outputs, including deliveries completed, floors cleaned, hours operated, and service incidents avoided.

That positioning gives Excelland Robot a clearer near-term commercialization path. Hotels, hospitals, and property operators do not need general-purpose intelligence for every task. They often need a machine that can navigate predictable routes, operate safely around people, and connect with elevators or access systems.

A practical robot can create value without resembling a person. It can also reach commercial deployment sooner because the scope of its job is narrower. The disadvantage is that specialized machines face price comparison, feature convergence, and pressure from established suppliers.

Humanoid developers face higher technical risk, but they can present a larger potential market. A sufficiently capable general-purpose robot could theoretically perform several jobs with one hardware platform. Excelland Robot instead relies on a family of purpose-built machines tied together by reusable software.

The IPO therefore tests two competing ideas about robotics investment. One emphasizes future generality and rapid capability gains. The other emphasizes existing deployments, narrower tasks, and incremental improvements to operating efficiency.

Excelland Robot’s reported position in delivery and cleaning robots gives it credibility within the second model. Its customer relationships with hotel groups such as H World Group and BTG Homeinns also provide environments where repeat deployments are possible.

Yet relationships do not guarantee supplier lock-in. Large hotel and property groups can test competing fleets, negotiate lower prices, or split procurement among several manufacturers. Software integration and service quality become more important as the physical products grow more similar.

The company’s ability to support indoor and outdoor operations could help differentiate its product family. It could also increase complexity because outdoor environments introduce weather, uneven surfaces, pedestrians, and changing routes.

Excelland Robot must show that its shared technology platform reduces that complexity faster than its product range creates new support costs. Public investors will ultimately judge the company through revenue quality and margins, not the number of scenarios shown in product demonstrations.

What the Growth Numbers Do Not Resolve

Revenue growth and narrower annual losses have improved the story, but the latest quarter exposes continuing financial pressure.

Excelland Robot’s revenue increased from approximately RMB244 million in 2023 to RMB267 million in 2024. It reached about RMB318 million in 2025, representing growth of roughly 19 percent during the latest full year.

Its annual net loss narrowed from approximately RMB251 million in 2023 to RMB151 million in 2024. The loss fell again to about RMB111 million in 2025. Lower operating expenses and a more favorable revenue mix contributed to that improvement.

Research spending remained material. Excelland Robot recorded approximately RMB73 million in research and development expenses during 2025. Its research team included 140 people by the latest reporting date, representing 35.5 percent of employees.

The company’s unusually high 2023 research expense included a one-time purchase of visual perception algorithms. Excluding that transaction helps explain why later research spending appeared substantially lower. It does not remove the need for continued investment in models, navigation, hardware, and cloud infrastructure.

The first quarter of 2026 was less reassuring. Revenue reached approximately RMB77 million, up about 5.7 percent from the corresponding period. Gross profit declined slightly, and gross margin fell from 12.1 percent to 11.2 percent.

The quarterly net loss widened to approximately RMB31 million from RMB26 million. Its net loss margin deteriorated from 35.9 percent to 40.8 percent, according to a review of the company’s financial disclosures.

The balance sheet creates another constraint. At March 31, Excelland Robot reportedly held approximately RMB64 million in cash and carried RMB194 million in short-term borrowings. Trade receivables stood near RMB131 million, while operating cash flow for the quarter was negative.

Receivables deserve particular attention because reported revenue does not immediately become usable cash. Slow collection can force a growing company to borrow more, especially when it must manufacture hardware before receiving customer payments.

The company attributes its losses partly to insufficient scale economies and continued research spending. That explanation is plausible, but investors still need evidence that additional scale will improve margins rather than intensify price competition.

Cost reductions also carry limits. Sales and marketing spending fell considerably between 2023 and 2025. Continuing to cut those expenses could become harder if Excelland Robot uses IPO proceeds to pursue new markets and acquire customers.

The skeptical case is therefore straightforward. Excelland Robot has reduced annual losses, but it has not established a consistent relationship between revenue growth, stronger gross margins, and positive cash generation.

Its IPO capital can extend the timeline for reaching that point. Capital alone cannot guarantee that the underlying economics will change.

Three Signals Will Define the Listing After September 9

The final allocation, post-listing financial results, and recurring-revenue mix will show whether this IPO strengthens the business or only funds its losses.

The first signal is the final allocation and institutional demand. Excelland Robot’s international placement represents 95 percent of the initial offering, making that order book especially important. Strong demand near the final terms would indicate that professional investors accept the company’s risk profile.

Demand must still be interpreted carefully. Cornerstone commitments can reduce the freely available allocation, while short-term IPO activity does not measure operational performance. Trading stability after the initial session will provide more information than the opening print alone.

The second signal is the next set of revenue, gross-margin, and cash-flow figures. Revenue growth without margin improvement would weaken the scale argument. A higher gross margin paired with lower operating cash consumption would offer stronger evidence that product mix and recurring services are working.

Investors should examine receivables alongside revenue. Faster sales growth provides limited protection if cash collection slows or customers require longer payment terms. Short-term borrowing will also show whether IPO proceeds are improving financial flexibility.

The third signal is the balance between hardware sales, AI vision solutions, and RaaS. Excelland Robot’s long-term case becomes stronger if service and software revenue expands without producing higher support costs. It becomes weaker if recurring arrangements consume capital while hardware margins remain in single digits.

Customer quality matters within that mix. Growth among repeat hotel, healthcare, office, and property customers would carry more weight than isolated deployments. Rising retention and net revenue retention would indicate that customers are expanding usage after their first purchase.

Product evidence will matter too. The company plans to invest in cloud-based models, vision-language-action systems, and dexterous manipulation. Investors should look for deployed products and customer contracts rather than treating research priorities as completed capabilities.

Excelland Robot’s September listing will not settle the commercial service robot debate. It will create a clearer reporting framework for testing one side of it.

The company has already demonstrated that specialized robots can reach thousands of customers. It has not yet shown that deployment scale reliably converts into durable margins and cash.

That is the question readers should carry beyond the offering date. Watch the first public financial updates, the recurring-revenue share, and the movement in gross margin. Together, those measures will show whether Excelland Robot is building a scalable operating platform or financing another expensive stage of expansion.