Hong Kong Exchanges and Clearing Limited and The Stock Exchange of Hong Kong Limited take no responsibility for the contents of this announcement, make no representation as to its accuracy or completeness and expressly disclaim any liability whatsoever for any loss howsoever arising from or in reliance upon the whole or any part of the contents of this announcement.

UBTECH

UBTECH ROBOTICS CORP LTD

深圳市優必選科技股份有限公司

(A joint stock company incorporated in the People's Republic of China with limited liability)

(Stock Code: 9880)

ANNUAL RESULTS ANNOUNCEMENT FOR THE YEAR ENDED DECEMBER 31, 2025

The board (the "Board") of directors (the "Directors") of UBTECH ROBOTICS CORP LTD (the "Company") announces the consolidated annual results of the Company and its subsidiaries (collectively, the "Group", "we", "our", "us" or "UBTECH") for the year ended December 31, 2025 (the "Reporting Period"), together with the comparative figures for the year ended December 31, 2024.

FINANCIAL HIGHLIGHTS

Our revenue increased by 53.3% to RMB2,001.0 million for the year ended December 31, 2025 from RMB1,305.4 million for the year ended December 31, 2024. Among these, revenue from full-size embodied intelligent humanoid robot products and services grew rapidly by approximately 2,203.7%, from RMB35.6 million for the year ended December 31, 2024, to RMB820.6 million for the year ended December 31, 2025, becoming UBTECH's largest source of revenue. Our gross profit increased to RMB753.8 million for the year ended December 31, 2025 from RMB374.0 million for the year ended December 31, 2024. Our gross profit margin increased by 9 percentage points to 37.7% for the year ended December 31, 2025 from 28.7% for the year ended December 31, 2024. Our loss decreased to RMB789.8 million for the year ended December 31, 2025 from RMB1,159.9 million for the year ended December 31, 2024.

MANAGEMENT DISCUSSION AND ANALYSIS

In 2025, the global artificial intelligence sector showed a trend of multi-point breakthroughs on core technologies. Key advancements, particularly in Large Language Models (LLMs), Generative AI, and Embodied Intelligence, are propelling the sector into a new era. Currently, AI technologies are transitioning from the digital realm to the physical world. As a pivotal technology in this shift, Embodied Intelligence deeply integrates AI into physical entities, such as embodied intelligent humanoid robots, endowing them with the ability to autonomously perceive, learn, and interact with their environment in real time. UBTECH continues to ramp up investment in embodied intelligent humanoid robot R&D, with a steadfast focus on industrial applications. We have successively launched the industrial versions of embodied intelligent humanoid robots, Walker S, Walker S1 and Walker S2 (as an embodied intelligence carrier, the Walker S series is designed and iterated to address real-world industrial application needs; in 2025, the Walker S2 series started mass production and delivery officially, and entered a new stage of large-scale scenario application). UBTECH remains committed to advancing key technologies, including the “humanoid brain”, “humanoid cerebellum”, and “high-performance limbs”. These technologies encompass core artificial intelligence technologies such as high-performance servo drive technology, large model technology, semantic VSLAM technology, learning-based motion control technology, visual perception technology, and multimodal interaction technology. The integration of our BrainNet 2.0 with Co-Agent technology establishes an “AI dual-cycle” for industrial embodied intelligent humanoid robots, enabling a spiral evolution of individual robot autonomy and group collaboration. Our full-stack technological advancements enable embodied intelligent humanoid robots to achieve superior task planning, dexterous manipulation, navigation and mobility, and human-computer interaction capabilities. Additionally, UBTECH has pioneered an embodied intelligent humanoid robot application paradigm for multi-tasking environments, and conducted the world’s first field trial for the multi-robot, multi-scenario, and multi-task group intelligence. In the meantime, UBTECH has spearheaded the development and establishment of multiple national standards in the field of humanoid robots and embodied intelligence. Serving as the deputy head of the Humanoid Robot Standards Working Group of the National Robotics Standardization Committee, co-leader of the Embodied Intelligence Working Group of the National AI Standardization Subcommittee, and member of the Standardization Committee for Humanoid Robots and Embodied Intelligence/co-leader of the Working Group on Complete Robots and Systems under the Ministry of Industry and Information Technology, UBTECH continued to guide the industry, including spearheading the development of key standards such as the Technical Requirements for Humanoid Robots – Part 5: Operational Tasks and Technical Requirements for Humanoid Robots – Part 6: Positioning and Navigation, and jointly leading the development of Technical Requirements for Humanoid Robots – Part 7: Human-Computer Interaction within the first batch of national standards for humanoid robots.

In 2025, UBTECH continued to focus on the large-scale deployment of our robots in industrial manufacturing scenarios. The field trials for the Walker S series entered the new stage of group intelligence, where multiple robots collaboratively executed production-line-level tasks, linking together a series of workstation tasks such as workbin handling, SPS sorting, parts assembly, process material handling, and quality inspection. As at the end of 2025, UBTECH achieved an annualized production capacity of over 6,000 full-size embodied intelligent humanoid robots, and continued to lead the global embodied intelligent humanoid robot industry in terms of order volume, covering key fields such as automotive manufacturing, smart manufacturing, smart logistics, and embodied intelligent datacenters, marking the official transition of embodied intelligent humanoid robots from field trial to commercial application.

A. Robotic Technologies

At the complete robot level, we released the third-generation industrial embodied intelligent humanoid robot, Walker S2, with the innovatively developed hot-swappable, autonomous battery-changing system. This system redefines the energy management paradigm for embodied intelligent humanoid robots with its 3-minute autonomous battery swap and 24-hour continuous operation capabilities. The bionic torso, featuring 52 degrees of freedom (DoF), is capable of performing high-difficulty maneuvers across a full spatial range of 0 to 1.8 meters, including stable load handling of up to 15 kg, flexible floor-reaching actions, extreme squatting, long-distance grasping, and flexible lumbar rotation with a range of ±162°. We released the wheeled humanoid robot Cruzr S2, featuring a waist-fold mechanism design that enables handling of 15kg payloads across a full spatial range of 0-1.8 meter, with zero-radius rotation in narrow aisles as compact as 0.8 meter and a rotation angle of ±170°. On the dexterous hand level, we released the fourth- and fifth-generation dexterous hands. In particular, the fourth-generation dexterous hand, with the aid of an 11-DoF structure and a worm gear transmission mechanism, can output a fingertip payload capacity of 12.5N, achieving sub-millimeter-level fine operations. The additional active thumb DoFs enable the dexterous hand to achieve flattening and pinching gestures, improving operational flexibility significantly. The fifth-generation dexterous hand has 19 fully-driven active rotational DoFs, and tactile and visual multimodal sensors, which can adapt to more complex industrial scenarios. At the bottom domain control level of humanoid robots, we have explored and developed a domain control solution that combines the brain and cerebellum of humanoid robots, further improving the integration of their electronic systems. Based on the application practice of the Walker S series in diverse industrial scenarios, we have developed atomic and combinatorial capabilities for humanoid robots, introduced control center nodes innovatively, and achieved efficient reconstruction at the application system level. We are also developing and refining a series of software infrastructure platforms, including an embodied intelligent humanoid robot management cloud platform, a data management platform, a robot training platform, and a simulation management platform, to support evolution toward multi-robot collaboration and group intelligence.

In 2025, we improved the preventive design system for humanoid robots, optimized the manufacturing process, strengthened module and complete robot testing, and tackled quality and delivery challenges in the mass production of full-size humanoid robots. The 1,000-unit-level small-scale mass production and delivery of the Walker S2 marks the maturity and stability of our manufacturing process.

In addition, we have integrated well-developed technologies accumulated through the Walker S series to start the R&D of new-generation full-size humanoid robots for the general commerce and education markets, focusing on functional aesthetic design, highly dynamic motion control, multimodal interactions, and open-source ecology building, etc.

B. AI Technologies

For embodied intelligent humanoid robot group collaboration scenarios, we have iterated and upgraded BrainNet 2.0. It comes with our proprietary Co-Agent used specifically for industrial embodied intelligent humanoid robots, which establishes an "AI dual-cycle" for industrial embodied intelligent humanoid robots, and promotes the evolution of industrial scenario applications for embodied intelligent humanoid robots toward refined collaboration. We have independently developed a foundation embodied intelligence large model, Thinker, with hundreds of billions of parameters, and established a "refining and purification" pipeline for multimodal data. By adopting a strategy of "large model-assisted annotation + multi-model cross-validation", we have achieved end-to-end automated annotation for tasks such as visual scene segmentation, action sequence classification, and instruction intention parsing. Moreover, we have created a data backflow mechanism innovatively, and fed back continuous model evaluation through real-scenario data. We have developed a visual language action model (Thinker-VLA) with closed-loop feedback, achieving millimeter-level precision for loading, generalized grasping, and dual-arm collaborative work bin handling. We have built an AI model generation pipeline, forming an infinite data loop starting from the base large model, and passing through business model optimization, distillation, and quantitative deployment sequentially for lightweight implementation, and then through data governance and difficult case mining, thereby feeding high-quality data back to the base large model, and achieving the continuous iteration and efficient implementation of AI models.

We have made our Thinker visual language base large model for embodied intelligence open-sourced officially with focus on "small parameters, high performance and full open source" to address the core pain points in the transition of humanoid robots from the laboratory to industrial scenarios – spatial measurement inaccuracy and time response delay, providing industrial humanoid robots with fast response and accurate spatial

perception capabilities, and their BrainNet and Co-Agent with cognitive and decision-making support. Currently, the model weight, training toolchain and application cases have been fully open-sourced to boost the industry's technology iteration and popularization.

We have independently developed Thinker-WM, a world model for industrial embodied intelligent humanoid robots. As one of the core foundations of the dual-cycle embodied AI system, it provides comprehensive spatiotemporal modeling of the physical world, long-sequence dynamic prediction, and closed-loop decision-making support for both single-robot operation and multi-robot collaboration scenarios of industrial humanoid robots. Built upon Thinker, a vision-language foundation large model, Thinker-WM establishes core capabilities in multimodal 4D spatiotemporal modeling. By integrating multi-source industrial scene data including vision and proprioceptive sensing, it enables millimeter-precise spatial state perception, causal prediction of long-horizon action sequences, and real-time forecasting of dynamic environmental disturbances. This delivers robust environmental perception and proactive decision-making capabilities for high-precision loading and unloading, generalized grasping, dual-arm collaboration, and multi-robot coordinated operations.

We are building and operating humanoid robot data collection and testing centers in multiple locations, and establishing a one-stop embodied intelligence data collection platform. Through data flywheels, we create a virtuous cycle of "more data - better models - wider applications - even more data", strengthening our core competitiveness and accelerating the deployment of robots in industrial and commercial settings.

C. Integrated Robotic and AI Technologies

1. Integration of Positioning, Navigation, and AI

We keep iterating multi-camera semantic VSLAM technology and have built an infinite-loop link for feature extraction and matching based on the real-machine data backflow self-supervision model, achieving spatial understanding and spatial memory functions such as multi-camera semantic processing VSLAM, autonomous map updating, generalized Landmark mapping, and DSG multi-layer semantic mapping, improving the stability of localization and mapping in real industrial scenarios comprehensively, and elevating computational efficiency while reducing the occupancy of computing resources significantly. Multi-sensor fusion semantic laser USLAM solves common scenario degradation problems such as large scenarios and high dynamics, achieving significant improvement in scenario range, accuracy and stability, and supporting the practical application of wheeled humanoid robots in industrial scenarios. We have developed an omnidirectional navigation model for embodied intelligent humanoid robots that supports new capabilities such as efficient backward movement, narrow channel lateral

movement, and close range obstacle stopping. In terms of 3D spatial perception, we have developed a multimodal and multi-task OD/OCC perception model for humanoid robots independently. By producing high-quality data through a 4D true value system, we have achieved 360-degree look-around spatial perception. We have built an end-to-end navigation simulation training system based on reinforcement learning using ISAAC simulation and self-developed handheld data acquisition devices.

L4 autonomous logistics vehicles have been deployed in multiple industrial parks, and achieved normalized operation on the premise of meeting customers' operational efficiency demand. By leveraging massive data and improving perception models, and optimizing online localization and mapping algorithms with semantic information continuously, we have iterated our multi-lane navigation framework, including integrated driving-parking task decision-making capabilities, scenario-based decision-making capabilities, behavioral intent decision-making capabilities, dynamic obstacle prediction capabilities, dynamic obstacle response capabilities, and lane keeping cruise capabilities. This has strengthened the safety, stability, and operational efficiency of our L4 autonomous logistics vehicles in complex scenarios, laying a foundation for the expansion of autonomous logistics from industrial parks to airports and open roads.

2. Integration of Machine Vision and AI Technologies

We have developed leading purely binocular vision technology independently, whose accuracy exceeds that of RGBD cameras, enabling the binocular vision of embodied intelligent humanoid robots to achieve "human eye like" perception, and has been applied to delivery in multiple scenarios. In terms of technology implementation, we have developed an industry-leading passive binocular vision system that uses a deep learning-based binocular depth estimation algorithm to generate high-precision dense depth maps in real time from dual RGB images, providing embodied intelligent humanoid robots with stereo vision perception comparable to human eyes. In terms of technological and functional progress, machine learning technology is used to perform geometric imaging tasks, fully leveraging the scaling law to expand the upper limit of perception, and allowing robots to "see more accurately"; structured light dotting is replaced with RGB pixels to improve the imaging resolution, allowing robots to "see more finely"; the large model is inputted directly into both eyes to create a native spatial intelligence end-to-end large model, allowing robots to "see and think together". To meet the demand of navigation and manipulation scenarios, we have integrated lightweight model design, a high-precision knowledge distillation framework, and Gaussian-probability-based supervised training innovatively. Combined with our proprietary high-precision data acquisition equipment and multi-scenario datasets, we have

achieved industry-leading performance in computational power, precision (submillimeter level), and generalization, providing core support for the robot's precise perception and interaction.

We have built a generative pipeline for objects and scenario assets in the simulation platform, which has two generation capabilities – automated generation and 3DGS reconstruction. To date, it has generated tens of thousands of proprietary digital object assets, solving the problem of insufficient asset quantity and generalization. In addition, we have built a data generation pipeline to solve the problem of insufficient generalization in data generation, designed simulation schemes such as normal light, excessive darkness, overexposure, motion blurring, randomly placed objects, and random motions of objects, and generated ten-million-class high-value training data in total. We use data generated by the simulation platform to solve the problem of high annotation costs for real data, design technical solutions for mixed training using simulation data + real data, and build a few shot deployable visual perception capability for industrial scenarios alongside difficult case mining, achieving a positioning accuracy of millimeter at an accuracy of 99.5% in various lighting changing/stack scenarios.

3. Integration of Motion Control and AI Technologies

To address the demand for dual-arm dexterous manipulation and high-performance bipedal mobility, we have developed learning-based motion control technologies utilizing imitation learning, reinforcement learning, and generative models. In dexterous manipulation, we have optimized our model-based grasping gesture generation and screening mechanism comprehensively, achieving the continuous and stable grasping of items in work bins. By combining data from model-based grasp generation with manually collected data, we have built the VLA dexterous hand operation model oriented to object generalization, and further enhanced the model's generalization and adaptation to unknown objects by expanding the object dataset using 3D object generation models. In addition, we have applied the capture generation and VLA models to the sorting of complex out-of-order stack materials in industrial SPS and logistics sorting scenarios. In terms of full-body motion control, we have optimized and improved the data-driven hierarchical reinforcement learning framework, achieving multi-source full-body motion data acquisition methods such as motion capture and video extraction, as well as the instant redirection algorithm, and trained trajectory tracking models, achieving high-speed humanoid walking, and full-height dynamic performances such as running, boxing and dancing. In the meantime, a high-level control strategy that integrates complex terrain and interactive information perception is designed to achieve stable robot movements in going upstairs and downstairs, and playing tennis, improving robots' motion performance and adaptability in external

environments significantly. Additionally, we keep optimizing work bin handling strategies for bipedal humanoid robots using reinforcement learning technology, achieving work bin handling operations with more stable dynamic loads and more accurate landing points.

As at December 31, 2025, the number of patents granted to us reached 2,985, including 508 overseas licensed patents, reflecting an increase of 11.4% compared to the end of 2024.

In 2025, in terms of honors and awards, UBTECH received the “Advanced Collective Award of Guangdong Province” from the Guangdong Provincial People’s Government; became a member unit of the Guangdong Artificial Intelligence and Robotics Industry Alliance and was appointed as a Vice Chairman Unit of the Artificial Intelligence Professional Committee of the China Electronic Chamber of Commerce; UBTECH was selected for the Top 100 Patents List in Shenzhen; and received awarded the title of “Jinge Award – Outstanding Company of the Year” from www.gelonghui.com; awarded the title of “Top 20 Leading Enterprise in Embodied Intelligence” from www.jiqizhixin.com; received the AI Tianma Leading Enterprise Award from the Shenzhen Artificial Intelligence Industry Association and the Guangdong Future Industry Research Institute, the title of “SSE Eagle – Emerging Technological Innovation Enterprise” from the Shanghai Securities News, and the title of “2025 Leading Enterprise in Artificial Intelligence of the Year” from QbitAI.

The application and commercialization process of UBTECH humanoid robot products has been recognized widely by the industry: The case of “demonstration application of the industrial version humanoid robot Walker S Lite in the automotive manufacturing field” was included in the list of typical application cases of AI-empowered new-type industrialization by the Ministry of Industry and Information Technology; UBTECH received the title of “Leading Enterprise in Technology Commercialization in the Guangdong-Hong Kong-Macao Greater Bay Area” from www.ifeng.com, the “2025 Embodied Intelligence Scenario Application Award” from www.lanjinger.com, entered the “2025 China Embodied Intelligence Business Potential List” released by Jazzyear, and awarded the title of “Yanshi – Scenario Application Wisdom Award” from the Humanoid Robot Scenario Application Alliance, and the “Artificial Intelligence Commercial Landing Pioneer Award” from LeadeRobot; the smart tourist guide solution that integrates “humanoid robots + AI + national image communication” created by UBTECH was selected as a typical case for “AI + cultural tourism” application scenarios in Guangdong Province in 2025.

UBTECH has received multiple awards in terms of capital and social responsibility, including the “2025 Golden Hong Kong Stocks Award” from www.zhitongcaijing.com, and the “Zhiyuan Award – Social Responsibility Pioneer Enterprise Award” from www.cls.cn.

On April 19, 2025, at the world’s first humanoid robot half-marathon, the full-size humanoid robot Tiangong Ultra, developed by the Beijing Humanoid Robot Innovation Center with UBTECH being the initiator and general manager, was the first to cross the finish line on the 21.0975-kilometer course, claiming the championship with a total time of 2 hours, 40 minutes and 42 seconds.

At the 2025 World Humanoid Robot Games in August 2025, “Tiangong”, developed by the Beijing Humanoid Robot Innovation Center with UBTECH being the initiator and major shareholder, received the first-ever 100-meter “Flying Man” title in the history of humanoid robots. Tiangong won the 100-meter championship with 21.5 seconds, and was the only athlete to run autonomously without manual remote control in the race.

2025 ANNUAL BUSINESS REVIEW

I. Full-size embodied intelligent humanoid robot products and services

The year 2025 marked a pivotal turning point for the humanoid robot industry, transitioning from “technical showboating” to “value realization”, and is widely regarded by the industry as the inaugural year for humanoid robots’ entry into industrial manufacturing scenarios. Leveraging its full-stack, self-developed core technologies in embodied intelligent humanoid robots, UBTECH has consistently focused on industrial manufacturing scenarios and successively launched the third-generation industrial embodied intelligent humanoid robot Walker S2 as well as the fourth- and fifth-generation dexterous hands. The Company has also innovatively defined a hot-swappable autonomous battery swapping system for the robot body, providing $7 \times 24$ continuous operation capabilities for industrial manufacturing scenarios, and enhanced the dexterous hands with 19 fully actuated active rotational degrees of freedom as well as tactile and visual multimodal sensors; meanwhile, based on hundreds of millions of high-quality industrial manufacturing scenario data points accumulated in 2025, we further iterated and upgraded BrainNet 2.0, which is equipped with the self-developed Co-Agent and the dexterous-operation large model Thinker-VLM that integrates reinforcement learning and VLA, thereby providing robots with rapid response and precise spatial perception capabilities and, furthermore, stable environmental cognition and forward-looking decision-making capabilities to enable robots to achieve autonomous millimeter-level precision loading and unloading, generalized grasping, dual-arm collaborative box handling and multi-robot coordinated operations, so as to adapt to more complex

industrial manufacturing scenarios. In addition, the Company has actively invested in the research and development of world models and has built the world model Thinker-WM. In 2025, the Company underwent a profound transformation from laboratory parameter competitions to commercial deployment, and full-size embodied intelligent humanoid robots broke through the bottleneck of generalization capabilities at an unprecedented pace.

In 2025, full-size embodied intelligent humanoid robot products and services for all scenarios achieved revenue of approximately RMB820 million, representing a year-on-year increase of approximately 2,203.7%; and achieved sales volume of 1,079 units, representing a year-on-year increase of 35,866.7%.

This year, our industrial embodied intelligent humanoid robots underwent “on-the-job” training day after day in real industrial manufacturing scenarios. We focused on three core workstations – handling, sorting and quality inspection – starting from customers’ most pressing needs to address key issues at such workstations, including high staff turnover, difficulties in management and highly repetitive work. At present, empowered by the Thinker embodied intelligence base large model (mimicking the human brain), our embodied intelligent humanoid robot Walker S2 has gradually become capable of taking over specific job tasks previously performed by humans. For handling roles in industrial manufacturing scenarios, our embodied intelligent humanoid robot Walker S2 features autonomous decision-making capabilities for depalletizing and palletizing tasks, full-body coordinated control enabling handling coverage for high-, mid- and low-level totes, as well as autonomous navigation and obstacle avoidance capabilities in confined areas, thereby enabling the robot to tackle the challenge of densely arranged tote handling across multiple layers and columns; for sorting roles in industrial manufacturing scenarios, by accumulating dexterous-manipulation data from both real-robot and simulation environments, our embodied intelligent humanoid robot Walker S2 has developed the capability to grasp stacked materials in constrained environments and to manipulate flexible materials, enabling the completion of complex tasks such as flow-rack sorting in automotive plants and stripping soft protective sleeves from engineering bolts. For quality inspection roles in industrial manufacturing scenarios, our embodied intelligent humanoid robot Walker S2 offers two representative categories of inspection capabilities – vision-based and operation-based – enabling a wide range of inspection types such as SPS kit completeness inspection, charging-gun plug-in/plug-out inspection and air-conditioning leak detection inspection, and is integrated with production and manufacturing systems to achieve timely closed-loop quality data. More importantly, by leveraging over 100 million pieces of industrial manufacturing scenario data accumulated by our embodied intelligent humanoid robots on real production lines to drive iterative technology upgrades, we have enabled robots to evolve from single-robot intelligence to collective intelligence, allowing robot fleets to collaborate like teams of workers and truly become a “never-stopping” new quality productive force.

10 -

This year, in addition to industrial manufacturing applications, we enabled our embodied intelligent humanoid robots to reshape a new ecosystem for education and scientific research. We entered universities and became teaching and research platforms for higher education institutions, and, by establishing embodied intelligence industry colleges as well as data collection and training grounds, we cultivated a large pool of embodied intelligence talent for the industry, enabling developers to build on our capabilities and explore the infinite possibilities of embodied intelligence; we also enabled embodied intelligent humanoid robots to address the challenge of flexible handling. In response to the stringent requirements for efficiency and accuracy in logistics and warehousing, our bipedal humanoid robots and wheeled dual-arm robots, through an integrated “Hardware + Software + Services + Operations” model, completed basic tasks such as sorting and handling in real-world scenarios. Through the embodied intelligent humanoid robot training grounds we deployed in Wuxi and other locations, by collecting and replaying massive volumes of high-quality real-robot data, our embodied intelligent humanoid robots completed “pre-job practical training” prior to taking up their posts, significantly shortening the cycle from deployment to stable value creation; we further enabled embodied intelligent humanoid robots to open a new chapter in commercial customization. From smart commercial service models featuring deep integration of “Technology + Humanities”, to guided-tour services under “Embodied Intelligence + Cultural Tourism”, our robots, as the “Guide Ambassador” of the China Pavilion, successfully completed reception duties at the Osaka World Expo, serving over 1.9 million visitors worldwide. Our embodied intelligent humanoid robot Walker S2, at the opening ceremony of the National Games, served as an opening guest and performed on a millennium-old bronze jüding excavated from a Vietnamese royal tomb. We continue to expand into promising vertical-industry applications and to form replicable standardized solutions. Commercial embodied intelligent humanoid robots, leveraging outstanding environmental adaptability, have gradually become a key option for many leading enterprises to address practical scenario applications, truly delivering on the commercial value of “robots replacing humans”.

11 -

II. Non-embodied intelligent humanoid robot products and services

In 2025, UBTECH continued to innovate in non-embodied smart humanoid robots. With strong support from independent developed “Dianling (點靈)” agent, AI Wukong (AI悟空), a humanoid robot product offering the experience of a “digital life companion”, became a new favorite in the consumer market. Meanwhile, the Yanshee educational humanoid robot, serving as a preferred platform for the implementation of embodied smart technology, talent cultivation, and algorithm validation, underwent a profound transformation from a “programming teaching tool” to an “embodied smart research foundation”. This advancement has promoted the development of an AI education ecosystem and fostered interdisciplinary embodied smart technology talents for the future. In 2025, sales of non-embodied intelligent humanoid robot products and services reached 12,759 units.

III. Other intelligent robot products and services

Education intelligent robot products and services

In 2025, UBTECH continued its core strategy for AI education, driving continuous iterations of hardware and software products, the implementation of teaching scenarios, and international expansion. Through the mature delivery of services and brand market penetration, the Company built a solid foundation for the entire AI education ecosystem.

In the K-12 education sector, we continued to deepen scenario implementation and expand product services. Youth AI and Robot Technological Innovation Education Bases have been completed and delivered in multiple cities. These are expected to serve tens of thousands of local students annually, providing a standardized vehicle for the popularization of regional AI education. We continuously iterated software and hardware, such as the AI Education Cloud, the Yuanxiang PBL platform, and the “Expert Teacher” system. Features like centralized control, FPV, and competitive modes were added to UGOT to enhance out-of-the-box playability. We introduced hardware bases, including the UDECA AI main controller and bionic case library, deeply empowering teaching under the new curriculum standards. We continued to broaden the scope of product services with the release of the C-end (consumer) version of the Yuanxiang PBL platform to meet diverse user needs.

12 -

In 2025, UBTECH strengthened its brand influence and industry voice through the dual paths of industry events and tournament ecosystem operations. We successfully participated in industry events, including the 85th China Educational Equipment Exhibition, the 63rd Higher Education Expo China, the World Robot Conference, the IROS International Conference on Intelligent Robots and Systems, and the Guangdong-Hong Kong-Macao Greater Bay Area AI and Robotics Industry Conference. We centrally showcased integrated services for industry-education fusion and scientific innovation education, receiving high praise from the international academic community and effectively enhancing industry recognition and brand penetration. We further optimized and iterated core competition categories within its tournament services, initially building a tournament ecosystem that covers both popularization and elite cultivation, connecting domestic and international stages. Our currently implemented “White List” and self-owned brand tournaments cover nearly 3,500 participating teams, with geographical reach extending to Hong Kong, Macao, and overseas countries such as Malaysia and Russia. The internationalization of these tournaments has significantly increased, and several teams in the “Youchuang Weilai” category were honored as national-level “Innovation Stars.” Simultaneously, we conducted referee training in multiple cities across the country, laying a solid talent foundation for the professional and standardized operation of competitions.

Logistics intelligent robot products and services

In 2025, UBTECH’s logistics intelligent robot products and services continued to deepen their presence in five core industries: new energy vehicles, new energy battery production lines, tire manufacturing plants, 3C electronics factories, and e-commerce/3PL. Drawing on our self-developed ACU robot core controller and UPIlot robot operating system, complemented by the information system on three cloud platforms: WMS, MES, and CTS, we deepened the implementation of the industry’s first full-stack unmanned logistics services, with the core products including the Wali Series U lifting mobile robots, Wali Series T latent towing robots, Wali Series F light- and heavy-load unmanned forklifts, Wali Series H heavy-load omnidirectional robots, Wali Series C stacker robots, and the Chitu L4-level unmanned logistics vehicle. Through integrated indoor and outdoor intelligent scheduling and multi-machine collaborative operations, we overcome the “last 10 meters” challenge in intra-facility logistics. This enables customers to build a fully unmanned logistics ecosystem, covering sorting, transportation, warehousing, and distribution, ultimately achieving a genuine closed-loop unmanned logistics system.

13 -

On the R&D front, UBTECH focused our product optimization, upgrades, and new development efforts on meeting client demands for end-to-end logistics scenarios. Our R&D is concentrated on the industrial mobile robots and L4-level unmanned logistics vehicles. We continued to refine our integrated delivery model of "Hardware + Software + Service + Operation", collaborating with clients and industry chain partners to pioneer innovative unmanned logistics applications and create benchmarks for unmanned logistics services. Through product innovation and upgrades, we helped customers achieve a qualitative leap in warehousing and logistics efficiency, transitioning unmanned logistics from merely "usable" to "highly efficient". In the AI field, based on our self-developed VLA framework and large model pre-training data, we advanced multi-modal alignment and the decoupling of "perception-action." This significantly enhanced the controllability and generalization of agents in complex tasks and formed a "Simulation-Teleoperation-Reinforcement Learning" (Sim2Real) closed-loop pipeline. This provides an iterable technical foundation for agents to enter the physical world, realizing the "awakening" of embodied intelligence from algorithmic models.

On the robot product side, UBTECH achieved a transition from scenario depth to all-domain intelligent mobility. We launched the Wali C50L + U50 bin-stacking and mobile robots, directly addressing cost-efficiency and fulfillment pain points in industries such as e-commerce/3PL to achieve high-density storage and efficient flow of bin-level materials. We introduced the Wali H3000 heavy-duty omnidirectional robot, which utilizes omnidirectional motion control technology to perform heavy-load handling tasks in narrow spaces, providing a new solution for building highly flexible and efficient logistics models. We released the next-generation Wali U600/U1500 lifting mobile robots, which cover mainstream load capacities up to $1500\mathrm{kg}$ through flexible combined applications and meet high-capacity demands across all industries with an exceptional load-to-volume ratio. We also launched the Chitu $\alpha$ , a new generation of outdoor unmanned logistics vehicles, pioneering an in-cabin fusion of LiDAR and binocular perception. Capable of operating in all weather conditions, it enables unmanned logistics from closed campuses to open urban roads and last-mile delivery.

14 -

On the strategic partnership side, UBTECH transitioned from single-point breakthroughs to ecosystem co-construction. We reached an embodied intelligence cooperation agreement with Volcano Engine (under ByteDance), focusing on multi-modal large models, VLA models, interactive ecosystems, and AI cloud-native infrastructure to accelerate the industrial implementation of large models in the field of embodied intelligence. We entered into a strategic partnership with KENGIC, a subsidiary of Digital China, to jointly build an embodied intelligence application ecosystem for unmanned warehousing, creating a full-stack unmanned logistics application model for the 3PL industry. We collaborated with M8M, a world-renowned smart logistics service provider, focusing on high-potential aviation logistics scenarios to drive the entry of unmanned logistics systems into the airport sector and create benchmark cases for airport unmanned logistics. We partnered with Aurotek Corporation, a smart robot service provider, to deeply explore the Taiwan and global markets, providing flexible unmanned logistics systems and exporting replicable Chinese services for industrial manufacturing upgrades to the world. We also collaborated with Sun Robotics, a Thai automation system integrator, to jointly build a localized product delivery and service network in Thailand, tapping into the "blue ocean" of the Southeast Asian smart manufacturing market.

On the corporate honors side, UBTECH moved from industry validation to authoritative recognition. With our outstanding technical advantages and comprehensive competitiveness in the field of unmanned logistics, we successfully passed the "National High-Tech Enterprise" certification. We received provincial recognition for its specialization, refinement, differentiation, and innovation, earning the title of "Specialized, Refined, Differentiated, and Innovative" enterprise in Jiangsu Province. We joined the ranks of high-growth innovative enterprises in the province, being selected for the "Potential Unicorn" list of Jiangsu in 2025. Our Wali F3000 heavy-duty unmanned forklift achieved cage stacking with millimeter-level positioning accuracy, winning the "Intelligent Stacking Architect" award at the world's first Unmanned Forklift Application Scenario Competition.

15 -

In 2025, the global shipment volume of UBTECH’s logistics intelligent robot products and services reached nearly 3,000 units, cumulatively spanning more than 60 cities and regions across 6 countries. Full-stack unmanned logistics have been implemented in various industries, including automotive manufacturing, new energy, and airports. The Company will continue to increase deployment scale and delivery density, covering more industries and key scenarios. This will enable unmanned logistics to move from indoors to outdoors, from low-speed to high-speed, and from manufacturing to distribution, achieving full-scenario coverage and accelerating the realization of an end-to-end unmanned logistics system. By continuously creating benchmark cases for unmanned factories, unmanned warehouses, and unmanned delivery services, we push forward the intelligent upgrade of industrial manufacturing with resilience and determination, injecting powerful momentum into the new era of global unmanned logistics.

3. Other sector-tailored intelligent robot products and services

In 2025, in terms of R&D, leveraging UBTECH’s longstanding accumulation in the R&D of industry-specific smart robotic products, it continued to deepen its technological capabilities in embodied intelligence and model applications, iterated on a series of industry-specific smart robotic products and their supporting intelligent service solutions, launched a new large-model version of industry-customized robots, and introduced new products for general-purpose wheeled embodied intelligent robots. Through multimodal interactive design and computing power upgrades, the Company achieved a major breakthrough in industry-level group intelligent operations and adaptive capabilities in complex environments.

In 2025, in terms of market expansion, UBTECH accelerated its global layout and penetration across multiple scenarios: its commercial robot series products were incorporated into the overseas store construction standards of a leading domestic new energy vehicle company, deepening its collaboration in both the domestic market and overseas expansion. The Company signed strategic agreements with ecosystem partners, including listed companies, key laboratories, and industry clients, to jointly explore the application of embodied intelligence through the pathways of “Robot Urbanization” and “Industry Services.” Furthermore, by collaborating with strategic corporate clients across more than 50 countries overseas, we continued to expand its global market presence, achieving innovative explorations in the multi-field implementation of sector-tailored intelligent robot products and services.

16 -

IV. Other intelligent hardware devices

In 2025, UBTECH successfully launched the smart lawnmowers M10 series for yard scenarios, the PC01, PC10, PC20, and PC200 series of pool robots, and the PS100 pool surface cleaning robot. Following these, we developed the smart lawnmowers M20 series, which is set to hit the market soon. We also initiated the R&D of the next-generation high-end flagship, the PC300. This product focuses on “all-domain 3D cleaning” for pools, integrating a smart environment sensing system with AI cleaning planning algorithms to further push the boundaries of high-end products. Currently, a comprehensive pool robot product line has been established, covering various price points and usage scenarios. In the field of smart home, we launched the T25 and T30 smart robot vacuums, along with the cost-effective E10 and the high-end all-in-one station product X1, continuing to broaden the robot vacuum product matrix. Regarding smart pet hardware, we developed and successfully launched the C30 rake-style series and C40 extra-large series of smart cat litter boxes, as well as the F20 series of smart refrigerated dual-temperature pet feeders. Additionally, we introduced more cost-effective products such as the PG01/PG01 Plus series pet grooming vacuum and the PD10 smart drying box. We continued to refine its product layout around fundamental pet care needs. Its product strength has been highly recognized by users and clients, earning the Excellent Partner Award from major overseas channel clients, while the sub-brand Airrobo was honored as a 2025 Forbes China Pet Industry Value Brand.

In 2025, on the market side, UBTECH continued to cultivate mainstream domestic and overseas markets, deeply expanding its reach across major online platforms, O2O channels, KA chains, and distribution clients. The Company pushed forward the refined operation of online e-commerce platforms; notably on mainstream platforms like Amazon, several of UBTECH’s smart pet hardware products consistently ranked among the top of the sale lists. We also remained active in mapping out offline channel client resources in Europe, America, Australia, and Southeast Asia, establishing stable partnerships with multiple large-scale, well-known distributors. The smart cat litter box series continued to maintain a leading position in global market share.

17 -

18 -

Employees and remuneration policy

As of December 31, 2025, we had 2,550 full-time employees, 2,545 of which were based in Chinese Mainland, with the majority working at our headquarters in Shenzhen with the rest located at our other offices in Liuzhou, Beijing, Xiamen, Jiujiang, Shijiazhuang and Hangzhou, etc. Total remuneration cost decreased to RMB914.6 million in 2025, compared to RMB940.7 million in 2024, primarily due to a decrease in share-based payment expenses. We recruit our employees mainly through campus recruitment, internal recruitment, social recruitment, referral by headquarters and internal staff recommendation (also through the open recruitment process). We emphasize that we have always adhered to the principles of equal competition, information disclosure and two-way selection in the open recruitment process. Whether it is in written examination, interview or recruitment conditions, we shall adhere to the principle of equal competition, ensure that recruitment information is disclosed transparently and communicate equally with candidates. We offer competitive remuneration package to our employees, which are generally based on their qualifications, industry experience, position and performance. In order to motivate, retain and reward talents for their contribution to the development of our Group, we have approved and adopted several equity incentive schemes since 2015. We further adopted the 2024 H share incentive scheme (the "H Share Incentive Scheme") in September 2024 upon approval by our shareholders (the "Shareholders") in order to, among other things, improve the Company's incentive mechanism to attract, motivate and retain Directors, supervisors, senior management and employees who have made outstanding contributions to our sustainable operation, development and long-term growth. For further details, please refer to our announcement dated July 31, 2024. We regularly evaluate the performance of our employees and reward the well-performed with bonus and promotion.

We invest in continuing education and training programs, including regular and tailor-made internal and external training, for our employees to improve their professional knowledge, and management skills, upgrade their skill sets and keep abreast of the industry standards in their respective positions. Pre-employment induction training and orientation is provided to all new hiring. We also organize activities to provide our employees with a deeper understanding of our culture.

19 -

Capital expenditures

The Group’s capital expenditure in 2025 was primarily related to the construction of projects such as Shenzhen headquarters building, Hangzhou Lin’an project, Jiujiang industrial park and Wuxi UQI PARK, which was amounting to RMB613.5 million, representing an increase of 53.2% from RMB400.4 million in 2024.

Right-of-use assets and lease liabilities

The related right-of-use assets and lease liabilities are located in the PRC. As of December 31, 2025, the Group’s right-of-use assets were RMB41.2 million (RMB58.1 million as of December 31, 2024) and its lease liabilities were RMB49.4 million (RMB71.9 million as of December 31, 2024). In 2025, depreciation expenses for right-of-use assets amounted to RMB27.3 million and interest expenses on lease liabilities amounted to RMB2.6 million.

Future Outlook

UBTECH has been adopting embodied intelligent humanoid robots and AI as the core strategy, focusing strategically on the research and development of embodied intelligence technology and its commercial application, and working hard on the breakthrough towards the commercialization of embodied intelligent humanoid robot technology and AI technology, committed to expanding the reach of embodied intelligent humanoid robots from industrial and commercial applications to every household. Specifically, we will focus on the following areas:

Embodied Intelligent Humanoid Robots

We will continue to invest in the R&D and mass production of new models of the Walker S series of industrial embodied intelligent humanoid robots. We will continuously introduce new materials and processes to improve the lightweight and integration level of the complete machine, deeply iterate the generalization and stability of autonomous operation algorithms, and consolidate our leading advantages in the industrial sector. Meanwhile, we will launch a new generation of more intelligent, user-friendly and open commercial and educational versions of embodied intelligent humanoid robots, targeting commercial scenarios such as commercial guidance and reception, corporate image display, and interactive event performances, as well as educational scenarios including university scientific research, system integration development, and AI teaching assistance. We will deeply iterate on the full-stack technology of humanoid robots, make breakthroughs in key embodied intelligence technologies including integrated joints, bionic five-fingered dexterous hands, BrainNet architecture, Co-Agent,

foundation large models, world models, learning-based motion control, spatial intelligence, and the ROSA system, and promote the “platformization” of common technologies. Through strategic technical planning and key breakthroughs, we will fully support the synergistic evolution of hardware, software, and services. We aim to continuously enhance the lightweight design, integration level, system stability, swarm coordination, and individual autonomy of embodied intelligent humanoid robots. This will enable deep coordination between robots and their surroundings, providing stable, reliable, safe, and practical services across diverse industrial, commercial, and educational scenarios. We are committed to accelerating the large-scale implementation of embodied intelligence technologies and products, comprehensively elevating the levels of intelligence and flexibility in related fields. Furthermore, we will build a robot developer community and an open robot ecosystem platform, attracting more partners to join the robotic ecosystem and accelerating the “breakthrough” of embodied intelligent humanoid robots from single-scenario applications to multi-scenario integration.

2. Industrial Vertical Domain Embodied Intelligence Dataset

We will rely on the multiple national multimodal data collection and testing centers for humanoid robots that we construct and operate to accumulate diverse and high-quality training data for industrial scenarios. We expect to collect more than ten million pieces of industry-scarce data annually. In terms of data collection scenario design, we will focus on typical industrial scenarios with large-scale application potential for humanoid robots, such as handling, sorting, and loading/unloading, while also covering typical commercial scenarios, providing sufficient data “fuel” to achieve “breakthroughs” in these scenarios. The scope of data collection includes not only basic actions such as grasping, placing, and alignment, but also extended long-range complex tasks. Furthermore, based on real-robot teleoperation data collection, we will deploy ontology-free data collection solutions and use world models to continuously generate high-quality simulation data, expanding the sources of embodied intelligence data through multiple routes.

– 20 –

21 -
Multi-modal Perception and Large Model Technology

Driven by the high-dimensional decision-making demands in complex production line tasks, we will continue to iterate a large multi-modal reasoning model for humanoid robots, which, as the core engine of the super cerebrum, will promote the continuous self-evolution of the BrainNet architecture and Co-Agent and unleash the outburst of collective intelligence. The multi-modal reasoning model of UBTECH humanoid robots will be trained and improved based on hundreds of millions of high-quality industrial data sets accumulated by the Walker S series, quickly training professional work types based on RAG (retrieval-augmented generation) technology. Consequently, the decision-making and generalization capabilities of robots across various workstations will be significantly improved, providing strong support for the large-scale application of embodied intelligent humanoid robots in industrial scenarios. We will explore and develop high-performance, production-grade perception solutions. The core of this effort is to transform laboratory perception algorithms into standardized capabilities suitable for large-scale production and stable delivery. Through multi-modal data reconstruction, performance consistency assurance, and a verifiable delivery system, we will support the stable mass production and rapid iteration of products such as robots and autonomous driving systems in complex scenarios.

Embodied Intelligence Decision-Making Technologies

For the scenario of embodied intelligent humanoid robots collaborating in groups, we will continue our investment in the development of group intelligence technology for embodied intelligent humanoid robots to iterate BrainNet architecture. This architecture integrates reasoning nodes and skill nodes, flexibly connected via edge-cloud collaboration, and establishes a collaborative system between the cloud-based super brain and the edge-side intelligent cerebellum. We aim to strengthen the global high-dimensional decision-making capabilities of the super brain based on multi-modal embodied inference large models, while strengthening the reasoning capabilities of the super cerebrum and the skill supply abilities of the intelligent cerebellum. This will improve the cloud-based AI system's efficiency in dynamic task decomposition, multi-robot intelligent scheduling, and global resource optimization for production-line-level tasks. Furthermore, through multi-modal stereoscopic vision perception, dynamic path planning, and collaborative decision-making, typical industrial operations can be implemented, such as autonomous depalletizing and autonomous shelving, increasing the success rate of task planning and accumulating more industrial scenario datasets and multi-scale models. We will continuously

iterate the industrial embodied intelligent humanoid robot Co-Agent, enhancing its “spontaneous behaviour chain” capability to empower robots with intent understanding, task planning, tool invocation, and autonomous exception handling. By improving the closed-loop system of “active perception – active reasoning – active action” for diverse tasks, we aim to advance embodied intelligent humanoid robots to a new level of “autonomous perception” from “passive execution”, supporting the implementation of large-scale cluster operations and flexible production in industrial scenarios. We will deeply integrate Thinker-WM, the Thinker-VLA vision-language-action model, and the embodied navigation system to form a more reliable intelligent closed loop: “simulation pre-training – real-machine validation – data backflow – model iteration”.

5. End-to-End Motion Control Technology

Building on our end-to-end learning-based motion control technologies, we will continue to make breakthroughs in the research and development of task-oriented full-body motion control and dexterous manipulation technologies. This aims to meet the highly generalized operational demands of complex tasks such as handling, loading/unloading, stacking, and sorting, as well as human-robot collaboration. In the direction of full-body motion generation and control, we will expand diverse full-body motion datasets and construct a universal full-body trajectory tracking model to support a wide range of mobile manipulation tasks. We will develop a full-body trajectory generator capable of receiving upper-level commands and environmental information to achieve real-time dynamic motion generation and tracking, leveraging hybrid force/position control at the end-effectors to expand implementation in industrial scenarios. In the direction of industrial handling, we will integrate reinforcement learning to improve handling speed and success rates, training highly efficient handling methods and motion trajectories to achieve autonomous depalletizing and palletizing decision-making with enhanced stability and efficiency. In the direction of palletizing/depalletizing skills, starting with the handling of bins of different sizes, heights, and weights, we will utilize omnidirectional force control to achieve broader and more universal palletizing modes. In the direction of industrial loading and unloading, we will combine traditional model-based control with reinforcement learning to improve the success rate and efficiency of peg-in-hole operations, driving the deployment of loading and unloading models. Simultaneously, we will utilize large-scale teleoperation data and simulated synthetic data to train VLA-RL models, enabling automated material grasping and sorting across the industrial and logistics sectors.

22 -

23 -
Embodied Navigation and the Autonomous Driving Technology for L4 Level Driverless Vehicles

Embodied navigation technology will target the commercial implementation of bipedal and wheeled humanoid robots and their large-scale industrial application as its core objectives, focusing on three key areas for intensive iteration: multi-scenario adaptation, high-robustness operation, and cross-platform compatibility. By deeply integrating core capabilities such as 3D perception, spatio-temporal memory models, multi-sensor fusion, and 3D cross-terrain spatial planning, we aim to bridge the upstream and downstream technical chains. This will enable an end-to-end closed loop of "perception-decision-action-feedback" for robots in complex environments. We will strengthen the precise prediction of dynamic obstacles and self-adaptive response, significantly enhancing the operational adaptability of robots in unstructured scenarios. Leveraging large model technology and integrating reinforcement learning, imitation learning, simulation training systems, and real-machine operational data feedback mechanisms, we will continuously optimize model generalization. This will result in a standardized and reusable technical solution, achieving production-grade deployment of this technology in scenarios such as commercial complexes, industrial parks, and industrial production lines.

Focusing on the multi-dimensional spatial intelligence upgrade, we will continue to advance cutting-edge autonomous driving technologies, including spatial memory, spatial perception and spatial cognition & interaction. We aim to consistently enhance capabilities in unknown obstacle detection, task decision-making, scenario decision-making, driving task behavioural intent decision-making, dynamic obstacle response, and lane cruising. By integrating game decision-making control with end-to-end technologies, we will comprehensively enhance the autonomous driving safety, stability, and operational efficiency of autonomous vehicles in complicated environments, so that the technology will be rolled out in more scenarios in the park to accelerate the expansion and implementation from parks to open roads.

FINANCIAL REVIEW

Overview

We generate revenue primarily from sales of (i) full-size embodied intelligent humanoid robot products and services, (ii) non-embodied intelligent humanoid robot products and services, (iii) other intelligent robot products and services; and (iv) other intelligent hardware devices. Our revenue in 2025 increased by 53.3% to RMB2,001.0 million compared to RMB1,305.4 million in 2024 owing to changes in revenue mix as detailed below.

Revenue by products and services

	Year ended 31 December
	2025		2024		Year-on-Year Change (%)
	Amount RMB'000	% of Revenue	Amount RMB'000	% of Revenue
Full-size embodied intelligent humanoid robot products and services	820,557	41.1	35,619	2.7	2,203.7
Non-embodied intelligent humanoid robot products and services	47,963	2.4	41,606	3.2	15.3
Other intelligent robot products and services	628,656	31.4	756,580	58.0	(16.9)
Other intelligent hardware devices	499,207	24.9	468,998	35.9	6.4
Others^{(Note)}	4,616	0.2	2,558	0.2	80.5
Total	2,000,999	100.0	1,305,361	100.0	53.3

Note: “Others” primarily included sales of raw materials and spare parts.

Our revenue attributable to full-size embodied intelligent humanoid robot products and services increased by approximately 2,203.7% from RMB35.6 million in 2024 to RMB820.6 million in 2025, primarily due to the comprehensive acceleration of large-scale scenario-based applications, which enabled us to successfully convert our technological accumulation into large-scale order deliveries.

Our revenue attributable to non-embodied intelligent humanoid robot products and services increased by 15.3% from RMB41.6 million in 2024 to RMB48.0 million in 2025, primarily due to the sales launch of new models.

Our revenue attributable to other intelligent robot products and services decreased by 16.9% from RMB756.6 million in 2024 to RMB628.7 million in 2025, primarily due to the fact that certain projects under contract in 2025 for logistics intelligent robots and intelligent robotic services had not yet been delivered and accepted by the end of the year, resulting in a decrease in revenue.

Our revenue attributable to other intelligent hardware devices increased by 6.4% from RMB469.0 million in 2024 to RMB499.2 million in 2025, primarily due to the continuous launch of new products driving revenue growth.

Adjustments have been made on disclosure categories of revenue due to change in the segments. The breakdown of revenue by category for 2024 is shown in the table below:

	Year ended 31 December
		2025		2024		Year-on-Year Change (%)
		Amount RMB'000	% of Revenue	Amount RMB'000	% of Revenue
Education smart robot products and services	Full-size embodied smart humanoid robot products and services	61,039	3.0	-	-	-
	Non-embodied smart humanoid robot products and services	24,367	1.2	33,568	2.6	(27.4)
	Other smart robot products and services	327,690	16.4	329,830	25.2	(0.6)
	Subtotal	413,096	20.6	363,398	27.8	13.7

		Year ended 31 December
		2025		2024		Year-on-Year Change (%)
		Amount RMB'000	% of Revenue	Amount RMB'000	% of Revenue
Logistics smart robot products and services	Full-size embodied smart humanoid robot products	45,134	2.3	-	-	-
	Other smart robot products and services	228,591	11.4	321,664	24.7	(28.9)
	Subtotal	273,725	13.7	321,664	24.7	(14.9)
Other sector-tailored smart robot products and services	Full-size embodied smart humanoid robot products	714,384	35.7	35,619	2.7	1,905.6
	Other smart robot products and services	72,375	3.6	105,086	8.1	(31.1)
	Subtotal	786,759	39.3	140,705	10.8	459.2
Consumer-level robots and other hardware devices	Non-embodied smart humanoid robot products and services	23,596	1.2	8,039	0.6	193.5
	Other smart hardware devices	499,207	24.9	468,997	35.9	6.4
	Subtotal	522,803	26.1	477,036	36.5	9.6
Others(Note)	Others(Note)	4,616	0.2	2,558	0.2	80.5
Total	Total	2,000,999	100.0	1,305,361	100.0	53.3

Note: “Others” primarily included sales of raw materials and spare parts.

Gross profit and gross profit margin

Our gross profit increased by 101.6% from RMB374.0 million in 2024 to RMB753.8 million in 2025. Our gross profit margins were 37.7% and 28.7% in 2025 and 2024, primarily due to structural changes, as high-margin full-size embodied intelligent humanoid robot products and services became our largest source of revenue.

27 -

Other income

Our other income decreased by 1.6% from RMB37.6 million in 2024 to RMB37.0 million in 2025, primarily due to a decrease in value-added tax refunds.

Selling expenses

Our selling expenses decreased by RMB52.9 million from RMB523.9 million in 2024 to RMB471.0 million in 2025, primarily due to a decrease in share-based payment expenses. As a percentage of the Group’s revenue, selling expenses decreased from 40.1% in 2024 to 23.5% in 2025, primarily due to the significant growth in revenue. Employee benefits expenses decreased by RMB33.3 million from RMB237.5 million in 2024 to RMB204.2 million in 2025, primarily due to a decrease in share-based payment expenses. In 2025, other key expenses include advertisement and promotion expenses of RMB129.1 million, sales commissions of RMB46.1 million, travel expenses of RMB26.9 million.

Administrative expenses

Our administrative expenses decreased by RMB34.4 million, from RMB370.0 million in 2024 to RMB335.6 million in 2025, primarily due to a decrease in share-based payment expenses. As a percentage of the Group’s revenue, administrative expenses decreased from 28.3% in 2024 to 16.8% in 2025, primarily due to the significant growth in revenue. In 2025, administrative expenses mainly comprised employee remuneration and benefit expenses of RMB240.0 million, depreciation and amortization of RMB19.2 million, office expenses of RMB15.5 million, professional service fees of RMB15.1 million.

Research and development expenses

Our research and development expenses increased by RMB29.4 million from RMB478.1 million in 2024 to RMB507.5 million in 2025, mainly due to the combined effect of increased investment in the research and development of embodied artificial intelligence and robotic technologies and a decrease in share-based payment expenses. As a percentage of our revenue, it decreased from 36.6% in 2024 to 25.4% in 2025, primarily due to an increase in revenue. In 2025, our research and development expenses mainly comprised staff remuneration of RMB409.4 million, software tools and consumer goods of RMB37.7 million, and materials consumed of RMB22 million.

Credit impairment losses

Our net credit impairment losses slightly decreased from RMB156.0 million in 2024 to RMB151.4 million in 2025. Credit impairment losses for the year are primarily due to delayed payments from certain government-related customers, and corresponding impairment was provided in 2025 on a prudent basis.

28 -

Finance income

Our financial income decreased from RMB6.0 million in 2024 to RMB0.6 million in 2025, primarily due to the combined effect of an increase in interest income and an increase in exchange losses.

Income tax expense

Income tax expense increased to RMB15.1 million in 2025 as compared to RMB10.7 million in 2024, primarily due to the increase in the pre-tax income of the subsidiaries subject to taxation.

Loss for the year

As a result of the foregoing, our loss for the year decreased from RMB1,159.9 million in 2024 to RMB789.8 million in 2025, primarily due to the significant growth in revenue (especially from full-size embodied intelligent humanoid robot products and services) and gross profit, as well as the continuous decline in the expense ratio.

Non-GAAP measures

To supplement our consolidated financial statements, which are presented in accordance with PRC GAAP, we also use “adjusted net loss for the year (non-GAAP measure)” and “adjusted EBITDA (non-GAAP measure)” as additional financial measures. We present this financial measure because it is used by our management to evaluate our financial performance. We also believe this non-GAAP measures provide additional information to investors and others in their understanding and evaluating our results of operations in the same manner as they help our management. However, these non-GAAP measures do not have a standardized meaning prescribed by PRC GAAP and therefore, they may not be comparable to similar measures presented by other companies.

We define “adjusted net loss for the year (non-GAAP measure)” as net loss for the year by adding back share-based payments (being non-cash in nature). We define “adjusted EBITDA non-GAAP measure” as “adjusted net loss for the year (non-GAAP measure)” adding back (i) interest expenses, (ii) income tax expense, (iii) credit impairment loss and asset impairment loss; (iv) depreciation of fixed assets and amortization of long-term prepaid expenses, depreciation of right-of-use assets and amortization of intangible assets, which are non-cash in nature, and (v) deducting interest income from it.

The following table sets forth the reconciliation of net loss and “adjusted net loss for the year (non-GAAP measure)” and “adjusted EBITDA (non-GAAP measure)” for the years ended December 31, 2024 and 2025:

	Year ended December 31,
	2025
RMB’000	2024
RMB’000
Reconciliation of net loss and “adjusted net loss for the year (non-GAAP measure)” and “adjusted EBITDA (non-GAAP measure)”
Net loss for the year	(789,823)	(1,159,924)
Add:
Share-based payments	99,300	245,326
Adjusted net loss for the year (non-GAAP measure)	(690,523)	(914,598)
Add:
Interest expenses (Note)	2,614	26,976
Income tax expense	15,143	10,665
Depreciation of right-of-use assets	27,300	30,802
Depreciation of fixed assets and amortization of long-term prepaid expenses	50,791	46,006
Amortization of intangible assets	8,431	8,339
Credit impairment loss	151,439	156,014
Asset impairment loss	39,272	43,219
Less:
Interest income	41,971	15,946
Adjusted EBITDA (non-GAAP measure)	(437,504)	(608,523)

Note: Interest expenses included interest expenses on borrowings, net of amount capitalised and interest expenses on lease liabilities.

Liquidity and capital resources

Cash used in operation in 2025 was approximately RMB784.1 million (RMB833.6 million in 2024). As of December 31, 2025, the Group had cash and cash equivalents of approximately RMB4,887.9 million (RMB1,191.0 million as of December 31, 2024) with approximately RMB1,123.0 million of external bank borrowings (RMB1,537.6 million as of December 31, 2024).

In terms of gearing, the Group’s gearing ratios (defined as total interest-bearing borrowings and lease liabilities divided by total equity) as of December 31, 2024 and 2025 were 71.6% and 16.0%, respectively. The current ratios of the Group (defined as total current assets divided by total current liabilities) as of December 31, 2024 and December 31, 2025 were 1.4 times and 3.0 times respectively. As of December 31, 2025, the Group had no material contingent liabilities. With the cash and bank balances in hand and the net proceeds received from the several rounds of placing of new H Shares under the general mandate granted to the Directors to allot, issue and deal with additional H Shares pursuant to a resolution of the Shareholders passed at the annual general meeting of the Company held on May 21, 2025 (the “General Mandate”) in 2025 and early 2025, the Group’s liquidity position remains strong to meet its working capital requirements.

As of December 31, 2025, our interest-bearing bank and other borrowings were RMB1,123.0 million. Our interest-bearing bank and other borrowings decreased from RMB1,537.6 million as of December 31, 2024 to RMB1,123.0 million as of December 31, 2025 mainly in respect of daily operations and the payments of the construction cost of the headquarter building of the Group located in Shenzhen.

Pledge of assets

As of December 31, 2024 and 2025, 100% equity interest of Shenzhen UBTECH Technology Industrial Co., Ltd.* (深圳市優必選科技實業有限公司), a subsidiary of our Group, and the land use rights held by it and some patent rights owned by the Group, have been pledged to secure bank loans of our Group mainly for our construction of property, plant and equipment.

Treasury policies and exposure to fluctuations in exchange rates

Most of the Group’s revenues are denominated in RMB as its operations are mainly located in China. As of December 31, 2025, approximately 30.2% (81.2% as of December 31, 2024) of the Group’s bank balances and cash was denominated in RMB and 69.6% (5.0% as of December 31, 2024) was denominated in United States Dollars, which mainly represented the proceeds from the placing of new H Shares under the General Mandate. The remaining 0.2% (13.8% as of December 31, 2024) was denominated in HKD, Euros, Singapore Dollars, Pounds or Japanese Yen. The Group continues to adopt a conservative approach in its foreign exchange exposure management. For the year ended December 31, 2025, the Group did not have a foreign currency hedging policy in respect of foreign currency transactions, assets and liabilities. The Group reviews its foreign exchange risks periodically and uses derivative financial instruments to hedge against such risks when necessary.

30 -

FINANCIAL INFORMATION

CONSOLIDATED INCOME STATEMENT

All amounts in RMB'000 unless otherwise stated

Item	Note	2025	2024
I. Revenue	5	2,000,999	1,305,361
Less: Cost of sales	5, 6	(1,247,186)	(931,352)
Taxes and surcharges		(19,610)	(9,891)
Selling expenses	6	(471,011)	(523,930)
General and administrative expenses	6	(335,623)	(370,006)
Research and development expenses	6	(507,483)	(478,108)
Finance expenses		594	5,983
Including: Interest expenses		(2,614)	(26,976)
Interest income		41,971	15,946
Add: Other income		36,952	37,635
Investment (loss)/income		(47,159)	20,381
Including: Share of investment loss of an associate and a joint venture		(57,532)	(5,818)
Gains or losses from fair value changes		56	8
Credit impairment losses		(151,439)	(156,014)
Asset impairment losses		(39,272)	(43,219)
Gains on disposals of assets		972	40
II. Operating loss		(779,210)	(1,143,112)
Add: Non-operating income		6,023	2,297
Less: Non-operating expenses		(1,493)	(8,444)
III. Total loss		(774,680)	(1,149,259)
Less: Income tax expenses	7	(15,143)	(10,665)

31 -

32 -

Item	Note	2025	2024
IV. Net loss		(789,823)	(1,159,924)
Classified by continuity of operations
Net loss from continuing operations		(789,823)	(1,159,924)
Classified by ownership of the equity
Net loss attributable to shareholders of the Company		(703,191)	(1,123,590)
Profit or loss attributable to minority interests		(86,632)	(36,334)
V. Other comprehensive loss, net of tax
Other comprehensive income that will not be subsequently reclassified to profit or loss
Change in fair value of investments in other equity instruments		(2,158)	(1,278)
Other comprehensive income that will be reclassified to profit or loss
Translation difference on foreign currency financial statements		(9,560)	(2,189)
VI. Total comprehensive loss		(801,541)	(1,163,391)
Total comprehensive loss attributable to shareholders of the Company		(714,909)	(1,127,057)
Total comprehensive loss attributable to minority interests		(86,632)	(36,334)
VII. Losses per share
Basic and diluted losses per share (RMB)	8	(1.55)	(2.67)

All amounts in RMB'000 unless otherwise stated

CONSOLIDATED BALANCE SHEET

Assets	Note	December 31, 2025	December 31, 2024
Current assets
Cash at bank and on hand		4,918,605	1,223,715
Financial assets held for trading		50,014	15,008
Notes receivables		39,764	36,819
Accounts receivable	9	1,302,224	913,498
Prepayments		125,226	117,801
Other receivables	10	128,346	26,435
Inventories		576,526	460,500
Other current assets		136,253	112,417
Total current assets		7,276,958	2,906,193
Non-current assets
Long-term receivables		196,963	140,351
Long-term equity investments		149,211	23,352
Investment in other equity instruments		66,987	49,795
Right-of-use assets		41,247	58,063
Fixed assets		241,743	139,035
Construction in progress		1,742,139	1,308,779
Intangible assets		425,319	447,448
Goodwill		-	10,351
Long-term prepaid expenses		24,251	28,635
Other non-current financial assets		3,042	-
Other non-current assets		72,948	21,523
Total non-current assets		2,963,850	2,227,332
Total assets		10,240,808	5,133,525

34 -

Liabilities and Shareholders’ equity	Note	December 31, 2025	December 31, 2024
Current liabilities
Short-term borrowings	11	622,858	793,365
Notes payable		11,998	–
Accounts payables	12	701,444	547,808
Contract liabilities		91,119	61,425
Employee benefits payable		139,449	124,344
Taxes payable		145,837	82,498
Other payables		537,544	391,094
Non-current liabilities to be settled within one year		84,532	70,982
Other current liabilities		57,245	21,031
Total current liabilities		2,392,026	2,092,547
Non-current liabilities
Long-term borrowings	13	439,400	708,630
Lease liabilities		25,523	36,553
Deferred income		58,255	47,961
Deferred tax liabilities		–	302
Total non-current liabilities		523,178	793,446
Total liabilities		2,915,204	2,885,993
Shareholders’ equity
Share capital	14	503,401	431,623
Capital surplus		14,284,489	8,477,261
Treasury shares		(57,716)	–
Other comprehensive income/(loss)		(6,200)	5,518
Accumulated losses		(7,505,988)	(6,802,797)
Total equity attributable to the shareholders of the Company		7,217,986	2,111,605
Minority interests		107,618	135,927
Total shareholders’ equity		7,325,604	2,247,532
Total liabilities and shareholders’ equity		10,240,808	5,133,525

35 -

1. GENERAL INFORMATION

UBTECH ROBOTICS CORP LTD (hereinafter referred to as the “Company”) was a joint stock company established in the People’s Republic of China (the “PRC”) on March 31, 2012. The registered address and headquarters of the Company is in Shenzhen City, Guangdong Province, the PRC. The Company’s shares were listed on the main board of The Stock Exchange of Hong Kong Limited on December 29, 2023. As of December 31, 2025, the total share capital of the Company was RMB503,401,373, each with par value of RMB1. The major shareholder of the Company is Mr. Zhou Jian, and as at December 31, 2025, he held 23.5% of the equity interest. The Company has had no ultimate controlling shareholder.

The Company and its subsidiaries (collectively, the “Group”) are primarily engaged in the research and development, design, production and sales of robotic products and provision of ancillary service and solutions.

2. PREPARATION BASIS OF FINANCIAL STATEMENTS AND SIGNIFICANT ACCOUNTING ESTIMATES

The financial statements are prepared in accordance with Accounting Standards for Business Enterprises – Basic Standard, and the specific accounting standards and relevant regulations issued by the Ministry of Finance on February 15, 2006 and in subsequent periods (hereafter collectively referred to as the “Accounting Standards for Business Enterprises”).

The financial statements are prepared on a going concern basis.

The Group applied the accounting policies and accounting estimates based on its business operations characteristics, including measurement of expected credit losses of receivables, valuation of inventories, depreciation of fixed assets, amortisation of intangible assets and right-of-use assets, revenue recognition and measurement, etc.

3. SEGMENT INFORMATION

The Group determines operating segments on the basis of the internal organisational structure, management requirements and internal report system, and determines reporting segments on the basis of operation segments, and discloses information of the segments.

An operating segment is a component of our Group that meets all the following conditions: (1) it engages in business activities from which it may earn revenues and incur expenses; (2) its financial performance is regularly reviewed by the Group’s Management to make decisions about resource to be allocated to the segment and assess its performance; (3) the Group is able to obtain its accounting information regarding financial position, financial performance and cash flows. Two or more operating segments may be aggregated into a single operating segment if the segments have similar economic characteristics and satisfy certain conditions. As the management of the Group believes that the Group has only one operating segment, the Group does not disclose any segment information.

During 2025, more than 90% of the Group’s non-current assets, excluding financial instruments and deferred income tax assets, were situated in Chinese Mainland.

36 -

4. SIGNIFICANT CHANGES IN ACCOUNTING POLICIES

In 2025, the Ministry of Finance issued several Q&As on the Implementation of Accounting Standards for Business Enterprises (hereinafter referred to as the “Implementation Q&As”). The accounting policies consistently adopted by the Group are consistent with the principles set out in the aforesaid Implementation Q&As, which have no material impact on the financial statements of the Group and the Company.

In addition, in December 2025, the Ministry of Finance issued Interpretation No. 19 of Accounting Standards for Business Enterprises, which became effective on January 1, 2026. The Group and the Company expect that the adoption of such Interpretation will have no material impact on their financial statements.

5. REVENUE AND COST OF SALES

	2025		2024
	Revenue	Cost of sales	Revenue	Cost of sales
Sales of goods	1,956,765	1,226,686	1,247,606	890,994
Provision of services	44,234	20,500	57,755	40,358
	2,000,999	1,247,186	1,305,361	931,352

The Group’s revenue and cost of sales breakdown based on the geographical regions and the timing of delivery of goods and provision of services are as follows:

| | 2025
Chinese
Mainland | 2025
Hong Kong
and overseas | 2025
Total |
| --- | --- | --- | --- |
| Revenue | | | |
| Including: Recognised at a point in time | 1,482,824 | 473,941 | 1,956,765 |
| Recognised over time | 42,999 | 1,235 | 44,234 |
| | 1,525,823 | 475,176 | 2,000,999 |
| | 2024
Chinese
Mainland | 2024
Hong Kong
and overseas | 2024
Total |
| Revenue | | | |
| Including: Recognised at a point in time | 820,442 | 427,164 | 1,247,606 |
| Recognised over time | 57,033 | 722 | 57,755 |
| | 877,475 | 427,886 | 1,305,361 |

In 2025, no customers contributed more than 10% of the total revenue of the Group.

In 2024, the revenue contributed by two customers which individually contributed more than 10% of the total revenue of the Group amounted to RMB436,945,000 in total, representing 33% of the Group's revenue.

6. EXPENSES BY NATURE

The cost of sales, selling expenses, general and administrative expenses, research and development expenses presented in the income statement classified by nature are sets out below:

	2025	2024
Raw materials and low-value consumables used	889,601	791,903
Employee benefit expenses	815,316	695,335
Share-based payments	99,300	245,326
Advertising and promotion expenses	129,095	135,445
Depreciation and amortisation	86,522	85,147
Transportation expenses	72,479	67,192
Changes in finished goods, work in progress, delivered goods and contract fulfilment costs	73,269	(54,566)
Sales commission	46,081	42,572
Travelling expenses	44,533	41,976
Subcontracting expenses	3,464	37,390
Outsourcing labour costs	35,812	34,309
Warehousing and storage expenses	29,330	32,978
Office expenses	32,789	29,622
Entertainment expenses	19,730	26,219
Software tools and consumables	37,671	15,262
Professional service fees	15,053	10,728
Repair and maintenance expenses	60,237	10,588
Installation costs	23,694	7,752
Scrap of inventories	3,687	5,813
Auditors’ remuneration	5,255	5,933
– Audit services	5,120	5,520
– Non-audit services	135	413
Others	38,385	36,472
	2,561,303	2,303,396

38 -
INCOME TAX EXPENSES

	2025	2024
Current income tax calculated based on tax law and related regulations	15,445	11,618
Deferred income tax	(302)	(953)
	15,143	10,665

The Group’s statutory tax rate in the PRC is 25%, while some PRC subsidiaries of the Group are subject to preferential tax rates of 5% and 15%. Subsidiaries in Hong Kong are subject to an income tax rate of 16.5%.

LOSSES PER SHARE

Basic losses per share is calculated by dividing the consolidated net loss attributable to ordinary shareholders of the Company by the weighted average number of outstanding ordinary shares of the Company:

	2025	2024
Consolidated net loss attributable to ordinary shareholders of the Company (RMB’000)	703,191	1,123,590
Weighted average number of outstanding ordinary shares of the Company (thousands)	454,862	421,322
Basic losses per share (RMB)	1.55	2.67

The weighted average number of outstanding ordinary shares of the Company does not include treasury shares held by the Company, which have been deducted in the calculation of basic losses per share. As there were no dilutive potential ordinary shares in 2025 and 2024, diluted losses per share equal to basic losses per share.

ACCOUNTS RECEIVABLES

	December 31, 2025	December 31, 2024
Accounts receivables	1,841,662	1,311,563
Less: Provision for bad debts	(539,438)	(398,065)
	1,302,224	913,498

(a) The ageing analysis of accounts receivables by invoice date is as follows:

	December 31, 2025	December 31, 2024
Within 6 months	890,612	613,709
6 months to 1 year	167,170	102,805
1 to 2 years	272,544	251,584
2 to 3 years	169,068	262,588
Over 3 years	342,268	80,877
	1,841,662	1,311,563

OTHER RECEIVABLES

	December 31, 2025	December 31, 2024
Deposits and guarantees receivables (a)	116,953	8,859
Receivables from equity transfer	10,543	10,783
Advances to employees	1,920	2,298
Receivables relating to asset disposals	-	3,500
Others	7,427	10,299
	136,843	35,739
Less: Provision for bad debts	(8,497)	(9,304)
	128,346	26,435

(a) As at December 31, 2025, the Group's deposits and guarantees receivable primarily consisted of performance guarantee deposit and quality guarantee deposit paid by the Group for the acquisition of the relevant shares in Zhejiang Fenglong Electric Co., Ltd. (浙江鋒龍電氣股份有限公司), as well as deposits for construction in progress and lease deposits, etc.

(b) There is no situation in the Group where funds were classified as other receivables due to centralized fund management.

40 -

11. SHORT-TERM BORROWINGS

	December 31, 2025	December 31, 2024
Secured, pledged and guaranteed borrowings (a)	15,022	138,188
Pledged borrowings	–	25,083
Guaranteed borrowings (b)	507,836	580,094
Credit borrowings	100,000	50,000
	622,858	793,365

(a) As at December 31, 2025, the secured, pledged and guaranteed bank borrowings of the Group of RMB15,022,000 were pledged by the patent rights of a wholly-owned subsidiary of the Group, and guaranteed by the third parties. As at December 31, 2024, the secured, pledged and guaranteed bank borrowings of the Group of RMB138,188,000 were secured by certain land use rights of the Group as collateral, and pledged by 100% equity interest of a subsidiary Shenzhen UBTECH Technology Industrial Co., Ltd. and 16 patent rights independently developed by the Group as pledges, and guaranteed by the Company.

(b) As at December 31, 2025, guaranteed bank borrowings of RMB507,836,000 (December 31, 2024: RMB580,094,000) were obtained by the Group, which are guaranteed by the Company and other subsidiaries.

12. ACCOUNTS PAYABLES

	December 31, 2025	December 31, 2024
Payables for purchase of raw materials	648,212	408,405
Payables for outsourcing labour costs	51,186	138,396
Others	2,046	1,007
	701,444	547,808

The ageing analysis of accounts payables based on invoice dates is as follows:

	December 31, 2025	December 31, 2024
Within 6 months	373,226	424,338
6 months to 1 year	154,778	46,310
1 to 2 years	121,580	62,814
Over 2 years	51,860	14,346
	701,444	547,808

As at December 31, 2025, accounts payables over one year amounted to RMB173,440,000 (December 31, 2024: RMB77,160,000), primarily representing payables for certain procurement relating to logistics projects. Due to the relatively long project periods, a portion of the payments will only be finally settled upon completion of acceptance inspection.

41 -

13. LONG-TERM BORROWINGS

	December 31, 2025	December 31, 2024
Secured, pledged and guaranteed borrowings	500,096	744,273
Less: Long-term borrowings repayable within one year	(60,696)	(35,643)
	439,400	708,630

As at December 31, 2025, long-term borrowings of RMB500,096,000 (December 31, 2024: RMB744,273,000) were mainly secured by certain land use rights of the Group as collateral, 100% equity interest of a subsidiary, Shenzhen UBTECH Technology Industrial Co., Ltd. (深圳市優必選科技實業有限公司), and 16 patent rights developed by the Group as pledges, and guaranteed by our Company, carrying floating interest rate, payable every three months, and the principal should be fully repaid before June 18, 2031.

14. SHARE CAPITAL

	December 31, 2024	Issuance of new shares in the current year	December 31, 2025
Share capital	431,623	71,778	503,401
	December 31, 2023	Issuance of new shares in the current year	December 31, 2024
Share capital	417,851	13,772	431,623

In February 2025, the Company issued a total of 10,155,099 shares to placees through private placing at the price of HK$90.00 per share, raising a total of RMB807,900,000, of which RMB10,155 thousand was included in share capital and RMB797,745 thousand was included in share premium.

In July 2025, the Company issued a total of 30,155,450 shares to placees through private placing at the price of HK$82.00 per share, raising a total of RMB2,194,565,000, of which RMB30,155 thousand was included in share capital and RMB2,164,410 thousand was included in share premium.

In December 2025, the Company issued a total of 31,468,000 shares to placees through private placing at the price of HK$98.80 per share, raising a total of RMB2,777,369,000, of which RMB31,468 thousand was included in share capital and RMB2,745,901 thousand was included in share premium.

15. DIVIDENDS

On March 31, 2026, the Board of the Company did not recommend the distribution of dividends for the year 2025 (2024: nil).

42 -

CORPORATE GOVERNANCE AND OTHER INFORMATION

Compliance with the Corporate Governance Code

The Company recognises the importance of good corporate governance for enhancing the management of the Company as well as preserving the interests of the Shareholders as a whole. The Company has adopted corporate governance practices based on the principles and code provisions as set out in the Corporate Governance Code (the "CG Code") set out in Appendix C1 to the Rules Governing the Listing of Securities on the Stock Exchange (the "Listing Rules") as its own code of corporate governance practices.

The Board is of the view that during the year ended December 31, 2025, the Company has complied with all the applicable code provisions as set out in the CG Code, except for Code Provision C.2.1 as disclosed below. The Board will continue to review and monitor the code of corporate governance practices of the Company with an aim to maintaining a high standard of corporate governance.

Pursuant to Code Provision C.2.1 of the CG Code, the roles of chairman and chief executive officer should be separate and should not be performed by the same individual. Mr. Zhou Jian is both the chairman of the Board and the chief executive officer of the Company. Since establishment of our Group in 2012, Mr. Zhou has been a key leadership figure of our Group and has been deeply involved in the formulation of business strategies for and the determination of the overall direction of our Group. Taking into account the continuation of management and the implementation of our business strategies, our Directors (including our independent non-executive Directors) consider that it would be most suitable for Mr. Zhou to hold both the positions of chief executive officer and the chairman of the Board, and that the existing arrangements are beneficial to the management of our Group and are in the interests of the Company and the Shareholders as a whole. The balance of power and authority is ensured by the operation of the senior management and our Board, both of which comprises experienced and high-calibre individuals. Our Board comprises four executive Directors (including Mr. Zhou), three non-executive Directors and four independent non-executive Directors, and therefore has a strong independence element in its composition.

In light of the above, the Board considers that the deviation from Code Provision C.2.1 of the CG Code is appropriate in the circumstances of the Company. The Board will continue to review the current structure from time to time and consider the appropriate move to take when appropriate.

Compliance with the Model Code

The Company has adopted the Model Code as set out in Appendix C3 to the Listing Rules as its code of conduct regarding dealings in the securities of the Company by the Directors and the supervisors of the Company (the “Supervisors”), and the Group’s employees who, because of his/her office or employment, are likely to possess inside information in relation to the Group or the Company’s securities. Specific enquiries have been made to all Directors and Supervisors and the Directors and Supervisors have confirmed that they have complied with the required standards set out in the Model Code during the year ended December 31, 2025.

No incident of non-compliance of the Model Code by the employees was noted by the Company for the year ended December 31, 2025.

Purchase, sale or redemption of the Company’s listed securities

References are made to the Company’s announcements dated January 19, 2025, January 24, 2025, and March 31, 2025 in relation to the purchase of H shares by the trustee of the Company under the H share incentive scheme from the secondary market (the “Share Purchase”). On January 24, 2025 and March 31, 2025, Yunnan International Trust Co., Ltd.* (雲南國際信託有限公司), the trustee of the H share incentive scheme, purchased 419,200 H shares and 396,750 H shares, respectively, from the market in accordance with the terms and conditions of the H share incentive scheme, representing approximately 0.097% and 0.090% of the total issued shares of the Company on the respective dates. These shares are held in trust for the benefit of the eligible participants of the H share incentive scheme, with total considerations of approximately HK$30.7 million and HK$31.3 million, respectively. For the avoidance of doubt, the Company itself did not purchase, sell, or redeem any H shares under the Share Purchase.

Save as disclosed above, during the year ended December 31, 2025 and up to the date of this announcement, there was no purchase, sale or redemption of any listed securities of the Company by the Company or any of its subsidiaries. As of 31 December, 2025, the Company did not hold any treasury shares (as defined under the Listing Rules).

Material legal proceedings

The Group was not involved in any material legal proceeding during the year ended December 31, 2025.

43 -

Use of proceeds from the Global Offering

The net proceeds from the Global Offering and the partial exercise of the over-allotment option amounted to approximately HK$931.21 million (the “Proceeds from the Initial Public Offering of H Shares”). As at December 31, 2025, the Group had fully utilized the Proceeds from the Initial Public Offering of H Shares, amounting to approximately HK$931.21 million.

As disclosed in the announcement of the Company dated May 15, 2024, the Board, based on the principle of prudence, after careful consideration and detailed evaluation of the progress of execution of the future plans and business strategy of the Group, resolved to change the use of the Proceeds from the Initial Public Offering of H Shares. The utilization of the Proceeds from the Initial Public Offering of H Shares up to December 31, 2025 was as below:

Use of proceeds	Proposed use of proceeds according to the prospectus (HK$' million)	Unutilized amount as of April 30, 2024 (HK$' million)	New reallocation of the unutilized net proceeds (HK$' million)	Unutilized amount as of December 31, 2024 (HK$' million)	Used amount during the Reporting Period (HK$' million)	Unutilized amount as of December 31, 2025 (HK$' million)
Further advance our R&D capabilities to enhance our core technologies and products and services offerings	425.76	394.80	425.76	212.66	212.66	-
Repayment of bank loans	175.74	141.60	263.43	-	-	-
Enhance our R&D infrastructure to improve our R&D capabilities and efficiencies	71.56	70.16	71.56	46.41	46.41	-
Enhance brand awareness and market penetration	86.06	78.39	23.00	-	-	-
Further optimize our management and operational efficiency	56.16	56.16	-	-	-	-
General working capital	115.93	34.32	147.46	-	-	-
Total	931.21	775.43	931.21	259.07	259.07	-

44 -

Use of Proceeds from placings of new H Shares under the General Mandate

The Company has conducted several rounds of placing under the General Mandate, namely (i) the placing of 5,060,000 new H Shares at the placing price of HK$86.18 per H Share under the placing agreement entered into between the Company, Guotai Junan Securities (Hong Kong) Limited, CLSA Limited and TradeGo Markets Limited (collectively, the “Placing Agents”) dated October 23, 2024, which was completed on October 29, 2024, and the closing price of the shares was approximately HK$111.90 on October 23, 2024 (being the date of the agreement of the October 2024 Placing) (the “October 2024 Placing”); (ii) the placing of 7,000,000 new H Shares at the placing price of HK$83.88 per H Share under the placing agreement entered into between the Company and the Placing Agents dated November 20, 2024, which was completed on November 26, 2024 (the “November 2024 Placing”), and the closing price of the shares was approximately HK$103.30 on November 20, 2024 (being the date of the agreement of the November 2024 Placing); (iii) the placing of 10,155,099 new H Shares at the placing price of HK$90.00 per H Share under the placing agreement entered into between the Company and the Placing Agents dated February 20, 2025, which was completed on February 27, 2025 (the “February 2025 Placing”), the closing price of the Shares on 20 February 2025 (being the date of the placing agreement in respect of the February 2025 Placing) was approximately HK$94.70; (iv) the placing of 30,155,450 new H Shares at the placing price of HK$82.00 per H Share under the placing agreement entered into between the Company and the Placing Agents dated July 22, 2025 (the “July 2025 Placing”), which was completed on July 28, 2025, and the closing price of the shares was approximately HK$85.10 on July 22, 2025 (being the date of the agreement of the July 2025 Placing); and (v) the placing of 31,468,000 new H Shares at the placing price of HK$98.80 per H Share under the placing agreement entered into between the Company and the Placing Agents dated November 25, 2025, which was completed on December 2, 2025 (the “November 2025 Placing”; the closing price of the shares was approximately HK$110.60 on November 25, 2025 (being the date of the agreement of the November 2025 Placing), collectively with the October 2024 Placing, the November 2024 Placing, the February 2025 Placing, and the July 2025 Placing, the “Placings”). For further details, please refer to the announcements of the Company dated October 23, 2024, October 29, 2024, November 20, 2024, November 26, 2024, February 20, 2025, February 24, 2025, February 27, 2025, July 22, 2025, July 28, 2025, November 25, 2025 and December 2, 2025.

45 -

Reasons for and benefits of the Placings

The Group is principally engaged in design, production, commercialization, sales and marketing and research development (R&D) of smart service robotic products and services. The Group's offerings ranging from consumer-level robots and appliances, enterprise-level smart service robotic products and services tailored for education, logistics and other sectors, are equipped to different extents with smart features that sense, interact, analyze and process human instructions and external environment such as mapping, temperature measurement and facial recognition.

The Directors consider that the Placings represent opportunities to replenish capital for continuous optimization and iteration and future large-scale commercialization in the global market of the products of the Group (mainly embodied intelligent humanoid robots) while broadening its Shareholders and capital base.

The Directors are of the view that the Placings would strengthen the financial position of the Group and provide more efficient funding support to the Group.

The Directors consider that the Placing aims to integrate the upstream and downstream supply chains, consolidate and strengthen the Group's leading position in the humanoid robot industry chain, thereby strengthening the Group's strategic layout in industrial manufacturing application scenarios, seizing the historic opportunity of the implementation of humanoid robots in industrial manufacturing application scenarios, and enhancing the Company's core competitiveness, technological strength, sales revenue and profitability.

The Directors consider that the terms of each of the placement agreements under the respective Placings (including the respective placing prices) are fair and reasonable and are in the interests of the Company and the Shareholders as a whole.

The net proceeds from the October 2024 Placing, the November 2024 Placing, the February 2025 Placing, the July 2025 Placing and the November 2025 Placing amounted to approximately HK$407.73 million, HK$551.88 million, HK$876.32 million, HK$2,410.26 million, and HK$3,055.71 million, respectively. The planned use of the proceeds from the Placings is consistent with the plans previously disclosed by the Company. The utilization of the proceeds from the October 2024 Placing, the November 2024 Placing, the February 2025 Placing, the July 2025 Placing and the November 2025 Placing up to December 31, 2025 was as below:

46 -

47 -

Use of proceeds		Net proceeds to be applied (HK$' million)	Utilized net proceeds as of December 31, 2024 (HK$' million)	Unutilized net proceeds as of December 31, 2024 (HK$' million)	Utilized net proceeds during the Reporting Period (HK$' million)	Utilized net proceeds as of December 31, 2025 (HK$' million)	Unutilized net proceeds as of December 31, 2025 (HK$' million)
October 2024 Placing
Business operation and development		254.49	220.95	33.54	33.54	254.49	-
- Working capital and general corporate purposes		253.97	220.43	33.54	33.54	253.97	-
(a)	Salaries, social insurance, housing fund, and other human resource expenses	56.74	47.34	9.40	9.40	56.74	-
(i)	For R&D personnel	13.47	9.52	3.95	3.95	13.47	-
(ii)	For selling and marketing personnel	16.12	13.55	2.57	2.57	16.12	-
(iii)	For management personnel	21.72	20.34	1.38	1.38	21.72	-
(iv)	For production personnel	5.43	3.93	1.50	1.50	5.43	-
(b)	Payment to suppliers and service providers	179.33	156.57	22.76	22.76	179.33	-
(i)	Procurement of materials for core business	132.76	115.82	16.94	16.94	132.76	-
(ii)	Procurement of services for core business	13.01	11.43	1.58	1.58	13.01	-
(iii)	Traveling and administrative services	7.99	7.88	0.11	0.11	7.99	-
(iv)	Storage and logistics services	14.20	13.28	0.92	0.92	14.20	-
(v)	Marketing and business development services	10.46	7.29	3.17	3.17	10.46	-
(vi)	Human resources related services	0.91	0.87	0.04	0.04	0.91	-
(c)	Other operating expenses	17.90	16.52	1.38	1.38	17.90	-
- Offshore and onshore investments (Note 1)		0.52	0.52	-	-	0.52	-
Repayment of amounts due under the credit facilities		153.24	153.24	-	-	153.24	-
Total		407.73	374.19	33.54	33.54	407.73	-

48 -

Use of proceeds		Net proceeds to be applied (HK$' million)	Utilized net proceeds as of December 31, 2024 (HK$' million)	Unutilized net proceeds as of December 31, 2024 (HK$' million)	Utilized net proceeds during the Reporting Period (HK$' million)	Utilized net proceeds as of December 31, 2025 (HK$' million)	Unutilized net proceeds as of December 31, 2025 (HK$' million)
November 2024 Placing
Business operation and development		192.92	-	192.92	192.92	192.92	-
- Working capital and general corporate purposes		191.94	-	191.94	191.94	191.94	-
(a)	Salaries, social insurance, housing fund, and other human resource expenses	46.72	-	46.72	46.72	46.72	-
(i)	For R&D personnel	12.98	-	12.98	12.98	12.98	-
(ii)	For selling and marketing personnel	13.99	-	13.99	13.99	13.99	-
(iii)	For management personnel	14.34	-	14.34	14.34	14.34	-
(iv)	For production personnel	5.41	-	5.41	5.41	5.41	-
(b)	Payment to suppliers and service providers	117.45	-	117.45	117.45	117.45	-
(i)	Procurement of materials for core business	75.75	-	75.75	75.75	75.75	-
(ii)	Procurement of services for core business	15.10	-	15.10	15.10	15.10	-
(iii)	Traveling and administrative services	8.82	-	8.82	8.82	8.82	-
(iv)	Storage and logistics services	9.12	-	9.12	9.12	9.12	-
(v)	Marketing and business development services	7.40	-	7.40	7.40	7.40	-
(vi)	Human resources related services	1.26	-	1.26	1.26	1.26	-
(c)	Other operating expenses	27.77	-	27.77	27.77	27.77	-
- Offshore and onshore investments (Note 1)		0.98	-	0.98	0.98	0.98	-
Repayment of amounts due under the credit facilities		358.96	-	358.96	358.96	358.96	-
Total		551.88	-	551.88	551.88	551.88	-

Use of proceeds	Net proceeds to be applied (HK$' million)	Utilized net proceeds during the Reporting Period (HK$' million)	Utilized net proceeds as of December 31, 2025 (HK$' million)	Unutilized net proceeds as of December 31, 2025 (HK$' million)	Expected timeline of use of unused amount
February 2025 Placing
Business operation and development	586.38	558.68	558.68	27.70	By March 31, 2026
- Working capital and general corporate purposes	466.61	443.45	443.45	23.16	By March 31, 2026
(a) Salaries, social insurance, housing fund, and other human resource expenses (Note 2)	103.26	97.96	97.96	5.30	By March 31, 2026
(b) Payment to suppliers and service providers (Note 3)	335.78	319.60	319.60	16.18	By March 31, 2026
(c) Other operating expenses	27.57	25.89	25.89	1.68	By March 31, 2026
- Project construction and renovation	92.62	88.08	88.08	4.54	By March 31, 2026
- Offshore and onshore investments (Note 1)	27.15	27.15	27.15	-	N/A
Repayment of amounts due under the credit facilities	289.94	275.44	275.44	14.50	By March 31, 2026
Total	876.32	834.12	834.12	42.20	By March 31, 2026

Use of proceeds	Net proceeds to be applied (HK$' million)	Utilized net proceeds during the Reporting Period (HK$' million)	Utilized net proceeds as of December 31, 2025 (HK$' million)	Unutilized net proceeds as of December 31, 2025 (HK$' million)	Expected timeline of use of unused amount
July 2025 Placing
Business operation and development	1,933.91	841.62	841.62	1,092.29	By September 30, 2026
- Working capital and general corporate purposes	1,765.82	794.46	794.46	971.36	By September 30, 2026
(a) Salaries, social insurance, housing fund, and other human resource expenses (Note 2)	573.66	205.22	205.22	368.44	By September 30, 2026
(b) Payment to suppliers and service providers (Note 3)	1,085.58	530.98	530.98	554.60	By September 30, 2026
(c) Other operating expenses	106.58	58.26	58.26	48.32	By September 30, 2026
- Project construction and renovation	146.35	25.42	25.42	120.93	By September 30, 2026
- Offshore and onshore investments (Note 1)	21.74	21.74	21.74	-	N/A
Repayment of amounts due under the credit facilities	476.35	449.14	449.14	27.21	By September 30, 2026
Total	2,410.26	1,290.76	1,290.76	1,119.50	By September 30, 2026

Use of proceeds	Net proceeds to be applied (HK$' million)	Utilized net proceeds during the Reporting Period (HK$' million)	Utilized net proceeds as of December 31, 2025 (HK$' million)	Unutilized net proceeds as of December 31, 2025 (HK$' million)	Expected timeline of use of unused amount
November 2025 Placing
Investment in or acquisition of potential upstream or downstream targets in the Group's business value chain, or integration with related industries or establish joint operating entities	2,291.78	112.03	112.03	2,179.75	By the end of 2027
Business operation and development	458.36	109.89	109.89	348.47	By the end of 2027
- Working capital and general corporate purposes	319.47	-	-	319.47	By the end of 2027
(a) Salaries, social insurance, housing fund, and other human resource expenses (Note 2)	64.00	-	-	64.00	By the end of 2027
(b) Payment to suppliers and service providers (Note 3)	250.47	-	-	250.47	By the end of 2027
(c) Other operating expenses	5.00	-	-	5.00	By the end of 2027
- Project construction and renovation	29.00	-	-	29.00	By the end of 2027
- Offshore and onshore investments (Note 1)	109.89	109.89	109.89	-	N/A
Repayment of amounts due under the credit facilities	305.57	-	-	305.57	By the end of 2027
Total	3,055.71	221.92	221.92	2,833.79	By the end of 2027

Notes:

(1) The offshore and onshore investments mainly comprise the fees to be incurred for identification of potential target companies and relevant investment amount.
(2) "Salaries, social insurance, housing fund, and other human resource expenses" mainly include the relevant expenses for R&D personnel, selling and marketing personnel, management personnel and production personnel. The exact allocation will be determined based on the actual situation.
(3) "Payment to suppliers and service providers" mainly include the relevant payment for procurement of materials for core business, procurement of services for core business, traveling and administrative services, storage and logistics services, marketing and business development services, and human resources related services. The exact allocation will be determined based on the actual situation.

52 -

Audit committee

The audit committee of the Board (the “Audit Committee”) currently comprises three independent non-executive Directors, being Dr. Dong Xiuqin, Dr. He Jia and Dr. Yao Xin. Dr. Dong Xiuqin is the chairperson of the Audit Committee.

The Audit Committee has reviewed the consolidated annual results of the Group for the year ended December 31, 2025 with the management and the auditor of the Company. The Audit Committee considered that the consolidated annual results of the Group for the year ended December 31, 2025 are in compliance with the applicable accounting standards, laws and regulations. The Audit Committee has also discussed matters with respect to the accounting policies and practices adopted by the Company and issues in relation to internal control, risk management and financial reporting with the management of the Company.

Scope of work of the auditor

The figures in respect of the Group’s consolidated balance sheet, consolidated income statement and the related notes thereto for the year ended December 31, 2025 as set out in the preliminary results announcement have been agreed by the Group’s auditor, PricewaterhouseCoopers Zhong Tian LLP, to the amounts set out in the Group’s audited consolidated financial statements for the year. The work performed by PricewaterhouseCoopers Zhong Tian LLP in this respect did not constitute an assurance engagement and consequently no opinion or assurance conclusion have been expressed by PricewaterhouseCoopers Zhong Tian LLP on the preliminary results announcement.

EVENTS AFTER THE REPORTING PERIOD

On January 22, 2026, the Company entered into a partnership agreement (the “Partnership Agreement”) with Liuzhou Industrial Holding Fund and Liuzhou Industrial Guidance Fund, pursuant to which the parties agreed to establish a partnership (the “Partnership”) to conduct investments within the intelligent robot industry chain with a focus on the existing industrial base of Liuzhou City. Pursuant to the terms of the Partnership Agreement, the total capital contribution amount to be paid by all partners of the Partnership shall be RMB660 million, of which the Company shall contribute RMB300 million. The purpose of the Partnership is to invest in upstream and downstream sectors, core components, and intelligent infrastructure, with the intelligent robot industry chain as its core and focusing on the existing industrial base of Liuzhou City. It aims to create a new highland for the artificial intelligence and robotics industries. The Partnership is jointly established by all partners, and through portfolio investment, it seeks to generate reasonable economic returns for its partners. For further details, please refer to the Company’s announcement dated January 22, 2026.

On March 10, 2026, the acquisition of 29.99% equity interest in Zhejiang Fenglong Electric Co., Ltd. (“Target Company”) and partial offer to acquire 13.02% equity interest in the Target Company have been duly passed at the 2026 first extraordinary general meeting held by the Company. All of the conditions precedent to the Share Transfer Agreement in relation to the payment of the first instalment have been satisfied, including that the SZSE has issued a Share Transfer Agreement Confirmation of Listed Company in respect of the transfer of the Sale Shares on March 10, 2026. In addition, on March 11, 2026, China Securities Depository has handled and registered all the sale shares to the Company’s A-share securities account and has provided to the Company with a Securities Transfer Registration Confirmation on March 12, 2026. As a result, as of March 12, 2026, the Group holds 29.99% equity interest in the Target Company and becomes a controlling shareholder of the Target Company (taking into account the giving up of voting rights in relation to the pre-acceptance shares by the vendors as detailed in the circular dated February 16, 2026). As at the date of this announcement, the partial offer has not yet been completed. The Company will publish further announcements in due course in accordance with the Listing Rules to update, among other things, the latest progress in relation to the partial offer, if necessary. For further details, please refer to the Company’s announcement dated March 10, 2026 and March 12, 2026.

Saved as disclosed above, there are no events causing significant impact on the Group from the balance sheet date to the date of this announcement.

53 -

54 -

FINAL DIVIDEND

The Board did not recommend the payment of any final dividend for the year ended December 31, 2025 (2024: nil).

PUBLICATION OF ANNUAL RESULTS ANNOUNCEMENT AND ANNUAL REPORT

This announcement is published on the websites of the Stock Exchange (www.hkexnews.hk) and the Company (www.ubtrobot.com).

The annual report of the Company for the year ended December 31, 2025 containing all the information required by the Listing Rules will be despatched to the Shareholders and published on the aforementioned websites of the Stock Exchange and the Company in due course.

By order of the Board