Earnings release
Page 1
1 Hong Kong Exchanges and Clearing Limited, The Stock Exchange of Hong Kong Limited and Hong Kong Securities Clearing Company 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. (A company controlled through weighted voting rights and incorporated in the Cayman Islands with limited liability) (Stock Codes: 0020 (HKD Counter) and 80020 (RMB Counter)) INTERIM RESULTS ANNOUNCEMENT FOR THE SIX MONTHS ENDED JUNE 30, 2026 The Board is pleased to announce the unaudited consolidated results of the Group for the six months ended June 30, 2026. These interim results have been prepared in accordance with International Accounting Standard 34 “Interim Financial Reporting ”, and reviewed by the Auditor, and by the Audit Committee. KEY HIGHLIGHTS Financial Summary Six months ended June 30 2026 2025 Amount As a percentage of revenue Amount As a percentage of revenue Year- over-year change (RMB in thousands, except for percentages) Revenue 2,910,946 100.0 2,358,198 100.0 23.4% Gross Profit 1,206,236 41.4 907,800 38.5 32.9% Profit/(Loss) for the period 617,308 21.2 (1,489,270) (63.2) -141.5% EBITDA 1,388,061 47.7 (847,692) (35.9) -263.7% Non-IFRS Measures: Adjusted EBITDA 384,855 13.2 (540,339) (22.9) -171.2% Adjusted net loss (385,898) (13.3) (1,181,917) (50.1) -67.3%
Page 2
2 BUSINESS REVIEW AND OUTLOOK Overall Performance During the first half of 2026, the shift from “model intelligence ” to “agentic intelligence ” significantly accelerated. At the same time, as long-horizon task reasoning, multimodal understanding, real-time interaction, and Agent Harness technologies continue to advance, AI is evolving from a tool that answers questions and generates content into intelligent systems that can understand goals, call tools, collaborate between humans and machines seamlessly, and continuously execute complex, long-running tasks. Industry competition has, as such, spilled over to encompass the deep integration of models, compute, data, tools, workflows, and safety governance into an end-to-end system capability – and the ability to deliver verifiable business outcomes in real-world scenarios. Against this backdrop, we have drawn on our long-standing expertise in foundation models, AI infrastructure, joint model-infrastructure optimization, and industry applications to further build a system-level AI capability stack defined by “one model system, one Token Factory, and one Agent Harness ”: a native multimodal model system that continuously pushes the boundaries of intelligence and extends agents ’ ability to handle multimodal, long-horizon tasks; a Token Factory that drives the scaling and efficiency of model inference, scheduling, optimization, and delivery; and a unified Agent Harness that supports diverse product form factors, translating model and compute capabilities into complete, reliable task delivery. Our view is that the shift brought by agents is not merely an upgrade in capability – it is a systemic migration in AI ’s audience, customer structure, and ultimately, how value is measured. • As the barrier to entry continues to fall, AI is expanding from an efficiency tool for a small number of professionals into a service for everyday individuals and “One-Person Companies ” (OPCs); • In the enterprise market, service demand is shifting from back-office digital and IT functions, to front-line business units, while the customer base is broadening from leading enterprises to a growing number of small and medium-sized businesses; • At the same time, as AI evolves from “a model being called ” to “an executor that delivers results, ” its value measurement will gradually shift from Token consumption to Task delivery – Token efficiency determines the cost of intelligence, while Task delivery determines customer value. Building on this judgment, our strategic focus is evolving from providing models and compute to delivering trustworthy multimodal agent capabilities at scale. During the reporting period, this strategic upgrade has already begun to materialize as healthy growth in revenue and gross profit, an increase in Recurring Revenue (RR), a continued significant narrowing of core operating losses, and continuous expansion in user scale.
Page 3
3 Financial and Operating Performance During the reporting period, we generated revenue of RMB2,910.9 million, up 23.4% year over year. Generative AI revenue reached RMB2,327.2 million, increasing 28.2% YoY and rising to 79.9% of our total revenue – reflecting that the “one model system, one Token Factory, one Agent Harness ” framework is accelerating its conversion into scaled model services, Token services, and agent applications. Computer Vision revenue came in at RMB496.8 million, up 13.9% YoY, returning to healthy growth following optimization of its operating strategy and continuing to serve as a key entry point into industry verticals and overseas markets. Overseas revenue surged 127.0% YoY – well above our overall growth rate – as we extended our system-level AI capabilities to new regions and customers, supported by mature Computer Vision capabilities, unified multimodal models, and a localized delivery system. • We are introducing Recurring Revenue (RR). RR for the first half of 2026 was RMB1,144.7 million, up 124.4% from RMB510.1 million in the first half of 2025, representing 39.3% of our revenue – a substantial increase from 21.6% in the same period last year. RR refers to revenue generated from existing contracts during the reporting period that demonstrates recurring characteristics over the contract term. The rapid growth in RR reflects how customer relationships are evolving from one-off project delivery towards ongoing services and deeper workflows, and signals that the sustainability and predictability of the Company ’s revenue are further enhanced. • We achieved gross profit of RMB1,206.2 million in the first half of 2026, up 32.9% YoY, with gross margin expanding to 41.4%, an increase of 2.9 percentage points YoY. This performance reflects our ongoing efforts to optimize our business mix and project quality, as well as continuous improvements in delivery quality and efficiency through joint model- infrastructure optimization, higher Token production efficiency, product standardization, and reuse of underlying capabilities. As this unified system-level capability stack serves more customers, users, and tasks, earlier investments made in technology and infrastructure are beginning to yield economies of scale. • During the reporting period, we continued to improve operating efficiency alongside healthy growth in revenue and gross profit, with core operating losses narrowing further and significantly. Non-IFRS adjusted net loss was RMB385.9 million, narrowing at an accelerating pace of 67.3% YoY. Benefiting from the continued improvement in our core business performance, as well as a significant increase in the fair value of self-incubated companies under the “1+X” strategy and externally invested AI ecosystem companies, we recorded net profit of RMB617.3 million under IFRS in the first half of 2026. This marks our first IFRS profit since being publicly listed. These results mark a key milestone – evidence that the commercial flywheel, powered by our system-level, end-to-end AI capabilities, is now beginning to turn and drive financial outcomes: customers and tasks continuously generate data and feedback that drive iteration in models, workflows, and product experience; improvements in model capability and Token production efficiency further lower task delivery costs, expand the boundaries of product application, and enable us to serve a broader base of enterprises, individuals, and OPCs. As value measurement gradually shifts from Token consumption towards Task delivery, we are driving a mutually reinforcing growth flywheel across technology capabilities, customer value, recurring revenue, and operating efficiency.
Page 4
4 System-Level AI Capabilities and Business Performance The competitive focus in AI is shifting from single-point metrics like model parameters and Token costs, towards the ability to complete complex tasks and create measurable value. We are breaking through model capability boundaries with our native multimodal unified architecture, building system-level AI capabilities through multimodal and long-horizon agents, and driving our products and services to evolve from per-Token call pricing towards high-value task delivery. One Model System: Continuously Pushing the Boundaries of Intelligence with Native Multimodal and Long-Horizon Agents Our technology roadmap is clear and coherent: unifying understanding and generation, seamlessly connecting perception, reasoning, and execution, and ultimately driving the convergence of digital and physical intelligence. Our goal is not merely to build models that answer questions, but intelligent systems that can continuously act autonomously and complete tasks. Along this path, we have continuously iterated on the SenseNova multimodal foundation model series, achieving a series of important breakthroughs. Native Unified Understanding and Generation: Laying the Foundation of Model Architecture Earlier this year, we introduced the NEO-Unify native multimodal unified architecture, departing from the mainstream separate encoder-decoder architecture by processing text and images within a unified representation space, and driving multimodal AI to evolve from functional stitching towards native fusion. SenseNova U1, released in April, unified understanding, reasoning, and generation within a single model, validating the feasibility of this roadmap; the subsequently released U1.5 further seamlessly integrated visual understanding, reasoning, generation, and editing, validating the architecture ’s scalability. In visual perception and reasoning, within a single model, SenseNova Vision unifies tasks such as structured visual understanding, dense geometric prediction, image segmentation, and multi-view 3D geometry, outperforming Google DeepMind ’s Vision Banana on the vast majority of metrics and matching, or even surpassing, specialized models across multiple core tasks. SenseNova U1.5- Lite scored 68.3 on the VBVR-Pro visual reasoning benchmark, surpassing Gemini 3 Pro ’s 56.4 and GPT Image 2 ’s 50.7, demonstrating the potential of the unified architecture to extend towards general visual and spatial intelligence. In controllable generation, the unified understanding and generation reduces information loss from cross-model transitions, enabling the U1 series to directly understand complex creative intent. U1.5-Lite achieves native 4K generation at a lightweight scale of just 8B-MoT, with overall performance matching Nano Banana 2; the model can precisely follow long-form, multi-constraint instructions to organize text, layout, and visual style, generate high-density complex infographics, and precisely modify specified content while keeping other parts and overall structure stable – driving image creation to evolve from one-off generation towards controllable, deliverable professional workflows. Building on our native multimodal technology innovations, we have actively promoted the development of an open-source ecosystem, further expanding our technology influence. As of the date of this announcement, less than four months after the U1 series went open-source, the three projects – SenseNova U1, SenseNova Vision, and SenseNova Skills – have collectively received over 11,000 stars on GitHub.
Page 5
5 Multimodal Long-Horizon Agents: Building System-Level High-Value Delivery Capabilities Real-world work consists of diverse information, tools, and steps. We are driving the deep integration of native multimodal and long-horizon agents, building system-level AI capabilities that cover perception, planning, execution, verification, and delivery – upgrading AI from a “tool” that provides answers to a “deliverer ” of end-to-end tasks. This capability has already been validated first in office productivity and content creation scenarios. In office scenarios, SenseNova 6.8-Flash-Lite, a multimodal agent released in August, brings documents, charts, web pages, videos, and application interfaces into a single task trajectory, capable of spanning hundreds of steps over several hours and dynamically orchestrating over a dozen specialized sub-agents to autonomously complete workflows from information-collection and data-analysis to report and presentation-generation. In long-chain tasks such as information search, its Token consumption is on average approximately 60% lower than that of text-only agents; it achieved across-the-board improvements across six benchmarks covering agent execution, planning and reasoning, and multimodal content generation, and led DeepSeek-V4-Flash, Gemini 3.5 Flash, and GPT-5.6 Luna in office benchmarks such as ClawEval, ArtifactBench, and TeamBench – achieving simultaneous optimization of quality and delivery cost. In content creation scenarios, SenseNova U1-Pro integrates intent understanding, creative reasoning, image generation, editing, and verification into a complete workflow through scaling and Agentic Generation. In third-party and customer evaluations, U1-Pro ’s overall generation quality matches GPT Image 2 and leads domestic multimodal models, with standout performance in dimensions such as creative reasoning and reliable text generation – driving AI to evolve from generating single images towards delivering professional content outcomes. Across both scenarios, a closed loop of “architecture innovation – system capability – high-value delivery – application feedback ” has begun to take shape. In the near future, we expect to release a larger-scale multimodal agent, SenseNova 6.8-Flash, further expanding capabilities in complex tasks and multi-agent collaboration. How AI value is measured will likewise shift from how many Tokens are consumed to whether tasks can be completed at lower cost, with higher quality, and with greater reliability. Towards Physical AGI: Driving the Convergence of Digital and Physical Intelligence System-level AI in the digital world is not the end goal. In the long term, AGI also needs to understand three-dimensional space and the laws of physics, perceive its environment, predict changes, and interact continuously – we define this goal as Physical AGI. Pursuing this long- term vision, we are extending accumulated expertise in native multimodal, long-horizon agents, and visual intelligence further into spatial understanding and physical interaction. Our SenseNova SI-8B model has achieved leading levels among open-source models of the same scale across eight spatial intelligence benchmarks, laying the foundation for models to understand spatial relationships and the three-dimensional world.
Page 6
6 Building on this foundation, we are co-developing the Kairos world model with ecosystem partner Ace Robotics. The newly released Kairos 3.1 integrates understanding, generation, and prediction of the physical world within a unified architecture. Its 8B model achieves 125-millisecond inference latency at BF16 precision on an industry-leading platform, enabling real-time on- device operation on robots, and adapts to different robot forms through a “one brain, multiple forms ” approach. Through real-world tasks, physical actions, and environmental feedback, we are exploring the formation of an R&D closed loop of “understand the world – simulate the future – drive actions – obtain feedback – iterate models, ” driving the unified multimodal model to gradually extend from digital intelligence towards physical intelligence. One Token Factory: Upgrading from Compute Provisioning to Scaled Intelligence Production With foundation model training continuing to evolve, inference demand growing rapidly, and agents gradually entering real-world workflows, how the value of AI infrastructure is measured is shifting from pure compute scale to the ability to continuously, stably, and efficiently produce high-quality Tokens. At the same time, Tokens are not merely homogeneous compute outputs – their value depends on the intelligence level of the underlying model, as well as the accuracy, reliability, response efficiency, and unit cost. SenseCore is evolving into a production-grade Token Factory. Through joint model-infrastructure optimization, as well as capabilities in model adaptation, inference optimization, heterogeneous scheduling, energy management, and global delivery, it not only expands Token output and lowers unit cost, but also continuously enhances the reasoning, understanding, generation, and execution capabilities that Tokens carry. As agents become capable of completing more complex long-horizon tasks, the Token Factory ’s long-term goal will likewise evolve from “producing Tokens ” towards “completing tasks ”: the measurement of infrastructure efficiency will gradually extend from per-Token cost to the completion quality, success rate, and cost per task, while the measurement of commercial value will likewise gradually evolve from billing by Token usage towards pricing by tasks and delivered outcomes.
Page 7
7 Powering SenseNova Foundation Models: Advancing Technical Frontiers and Large-Scale Commercial Adoption We continue to upgrade SenseCore, focusing on the training, post-training, inference, and deployment of frontier language and multimodal models, with key breakthroughs in ultra-long- context parallel training, large-scale reinforcement learning acceleration, heterogeneous mixed inference, and domestic hardware/software adaptation: • Long-context Agent model training. SenseCore leverages the interconnect and memory characteristics of advanced compute to implement sequence parallelism, topology-aware multi-dimensional parallel strategies, and compute-communication co-optimization, tripling training speed compared to pre-optimization levels. • RL post-training acceleration. In the reinforcement learning post-training phase, SenseCore decouples training from Rollout tasks and leverages heterogeneous inference resources across both domestic and advanced compute to accelerate sample generation, further improving post-training efficiency. • Image and video generation efficiency. For image and video generation, our proprietary LightX2V inference framework combines step-distillation technology with multi-level compute optimization, boosting generation efficiency by 6x and laying the foundation for large-scale commercial content generation and creation. • Model iteration and deployment acceleration. Backed by a unified model R&D and deployment platform, SenseCore has compressed the journey for our models from R&D validation to large-scale deployment down to just one week. While strengthening our internal system capabilities, we also actively promote collaboration between our in-house models and the upstream ecosystem. The open-source SenseNova U1 model has achieved Day 0 (launch day) adaptation with 10+ domestic chip manufacturers { ready the moment it goes open source, accelerating technology iteration and large-scale commercial use of our in-house models.
Page 8
8 Advancing the Domestic Compute Ecosystem and Large-Scale Commercial Adoption We are firmly committed to advancing the domestic compute ecosystem, continuously refining a full-stack adaptation system covering models, frameworks, operators, toolchains, and hardware, and conducting joint optimization with domestic chip and industry chain partners, to convert different types of compute into effective Token supply for real-world businesses. Through low- level operator optimization, multi-device parallel tuning, PD (Prefill/Decode) separation, and heterogeneous hybrid inference, we can achieve an MFU (Model Floating-point Utilization) on mainstream domestic chips of up to 2.5 times the manufacturer ’s baseline level, and have successfully validated the commercial value of domestic compute in scenarios including AI4S (AI for Science) and video generation, driving domestic compute to evolve from technical validation towards large-scale commercial services. We are partnering with Hong Kong Science Park to build the largest domestic AI compute center in Hong Kong, targeting a compute scale of 40,000 PFLOPS by 2030 and establishing a self-reliant and controllable compute anchor point for the Greater Bay Area, Southern China. In addition, we have deployed the first overseas domestic compute cluster in Saudi Arabia, replicating China ’s leading AI infrastructure construction and operation experience overseas, and continuously expanding the global footprint of Token services. Continuously Reducing Unit Intelligence Cost Through Compute-Energy Synergy We continue to integrate compute management, data center operations, and energy management systems, establishing a coordinated mechanism that spans from power load forecasting to compute scheduling and operations decision-making. During the reporting period, the Compute- Energy Synergy Agent achieved 96% load forecasting accuracy, and based on forecasting results, coordinated energy storage scheduling, data center operations and maintenance, and compute task allocation to continuously optimize data center Power Usage Effectiveness (PUE) and electricity consumption structure. These measures drove an approximately 8% reduction in average unit electricity cost, with cumulative electricity cost savings exceeding RMB12 million in the first half of 2026. Annual carbon reduction reaches 24,000 tons per 10,000 PFLOPS of compute, and the overall expected annual carbon reduction is on track for a step-change increase – several-fold higher than the 4,000 tons achieved in full-year 2025. While reducing energy costs and carbon emissions, compute-energy synergy also provides a replicable operational foundation for large- scale Token Factories to continuously reduce unit intelligence production costs. Rapid Growth in Token Services As of the date of this announcement, the total compute scale under our operation reached 48,000 PFLOPS. In July 2026, SenseCore ’s daily average Token service volume exceeded 2.4 trillion, an approximately 22-fold increase year over year. In addition to serving the SenseNova foundation model, our Token Factory has provided services to four external foundation model providers, and – leveraging domestic heterogeneous inference and joint model-infrastructure optimization capabilities – has built differentiated service capabilities in high-complexity computing domains such as AI4S, video generation, embodied intelligence, and world models, as well as in city-level intelligent scenarios.
Page 9
9 One Agent Harness: Delivering Complete Tasks to Enterprises and Individuals For enterprise and individual users, we offer agent products in various forms, including office productivity, content generation, and intelligent interaction; while these products serve different users and scenarios, they are all underpinned by the same Agent Harness. This system connects the unified multimodal model with the Token Factory, converting model capabilities and compute into complete tasks that are executable, manageable, and deliverable. Specifically, the Agent Harness is responsible for understanding user goals, decomposing tasks, calling knowledge and tools, orchestrating workflows, and centrally managing identity and permissions, long-term memory, execution evaluation, audit tracing, and safety governance. Through diversified front-end products and unified back-end capabilities, we can rapidly reuse the same task delivery capability across different industries and user scenarios, shortening product development and project deployment cycles, and continuously improving the quality, reliability, and delivery efficiency of task completion. Enterprise Agent Services: Deep into Office and Content Production Workflows, Delivering Business Outcomes In the enterprise market, we integrate our unified AI agent capabilities into customer workflows through various delivery forms, including public cloud services, APIs, enterprise on-premise deployment, and FDE (Forward Deployed Engineer) services. In the first half of 2026, we served over 1,000 enterprise customers, with over 100 new customers added during the period. Our customers span over 20 key industries, including education, government, marketing, logistics, banking, and insurance. Among these, Raccoon is our agent product built for office scenarios, and has served leading enterprises across industries including Lenovo, Ping An Technology, the three major telecom operators, JD.com, and Kylin Software. For example, through our partnership with Kylin Software, Raccoon has been integrated into the domestic operating system, connecting the operating system entry point, AI application distribution, and enterprise customer delivery chain; and through on-premise deployment in JD.com ’s supply chain business, Raccoon has built an end-to-end intelligent service system around actual workflows, validating its delivery capability in complex environments, high-security requirements, and mission-critical business systems. In content production and creation, Seko has advanced from single-purpose generation to end-to- end delivery covering intent understanding, script creation, and storyboard production. Currently, Seko’s daily video output reaches 10,000 minutes, with active users averaging over one hour of creation time per day, and it ranked first in per-capita visit duration on the web in the QbitAI AI Application Monthly Report for June 2026. Content created by Seko has accumulated an estimated 1.5 billion views across social media short-video platforms, with 10 short dramas each surpassing 100 million views. These figures demonstrate that Seko is evolving from a single-point video generation tool into a professional platform supporting continuous creation and large-scale content production, and the content it creates also demonstrates broad market reach potential.
Page 10
10 As the barrier to AI adoption continues to fall and capabilities evolve from single-point Q&A to end-to-end task delivery, the user base of enterprise AI agents is also expanding beyond traditional enterprise customers to broader groups such as OPCs. In the first half of 2026, the Raccoon OPC Capability Challenge Season attracted over 650,000 participants, Raccoon ’s peak daily active users surpassed 700,000, and over 8,000 real-scenario creations were accumulated. Building on this, we are advancing the “Hundred Industries, Thousand Enterprises Co-construction Plan ” ( “ϵ ྌ ”), covering over 40 cities, and participating in the development of a nine-level OPC capability standard and a talent incubation center. The first batch of the Seko Industrial Base (“Seko ପุਿή ”) has also attracted over 20 OPC startup teams to join, supporting micro teams to enter professional content production and commercialization chains by providing models, compute, creation tools, and industry resources. As a result, we are extending our unified agent capabilities from serving large organizations to a broader range of enterprises, teams, and individuals, continuously expanding the boundaries of AI productivity services. Personal Assistants: Bringing Enterprise-Grade AI Capabilities to Help Users Achieve a Better Personal Life In the consumer market, we have built the Kapi series – our Personal CXO product matrix of agent-powered personal assistants. Surpassing 45 million cumulative users, Kapi has evolved from single-point tools into proactive life agents spanning health, finance, and personal creation, helping users navigate everyday life with greater ease and confidence. Kapi Cam, leveraging the powerful multimodal capabilities of the SenseNova foundation model, has launched an AI photography agent that proactively understands scenes and creative intent, helping users effortlessly capture better production. Kapi Money has undergone a complete revamp, launching a full-scenario financial agent that further addresses complex financial needs such as income and expense insights, budget planning, and multi-account management. Our new addition, Kapi Health, has quickly gained user recognition since its launch, with its retention rate reaching the leading level of comparable products. We believe the value of personal agents should not stop at improving efficiency – it should also help users achieve personal goals and enjoy a better life experience. In the long term, we hope the Kapi series of personal assistant agents can become personal life assistants that truly understand users and provide continuous companionship. Further Strengthening Our Leadership in the Computer Vision Market Computer Vision is a cornerstone of our business – deeply embedded across industry verticals and overseas markets, and serving as a critical gateway for our system-level AI capabilities to integrate into customer workflows. During the reporting period, our Computer Vision business covered over 20 countries and regions, having cumulatively served over 4,500 customers domestically and overseas, with a recurring customer revenue contribution rate of 67% (meaning 67% of current- period revenue came from customers who made purchases in previous year), reflecting our long- established customer base, industry understanding, and large-scale delivery capability. Backed by our complete AI product and service portfolio, we grow with our customers – expanding from a single application to broader business workflows. We deliver unified multimodal models, agents, and compute services supporting use cases from analysis curation and internal training to digital content generation and cybersecurity optimization, continuously deepening and broadening customer service.
Page 11
11 According to IDC ’s “China AI Software Market ” report, we have ranked first in China ’s Computer Vision market for ten consecutive years. In July 2026, we were awarded the National Science and Technology Progress Award (Second-Class). The National Science and Technology Award, established by the State Council, is one of the highest-level and most authoritative national awards in China ’s science and technology innovation field, further reflecting our long-established technical strength and industry contribution in the visual intelligence domain. Ecosystem Strategy: Connecting Digital and Physical Intelligence Guided by our long-term technology roadmap and industry collaboration priorities, we have consistently invested ahead of the curve in world models and embodied intelligence, intelligent terminals, smart healthcare, chips, and related domains. Our ecosystem companies include those incubated and nurtured in-house, as well as upstream and downstream ecosystem partners with whom we collaborate through strategic equity investments. These ecosystem companies provide us with real-world scenarios, industry feedback, productization capabilities, and commercialization channels, while we support ecosystem partners through our models, Token Factory, and engineering capabilities to shorten their R&D and deployment cycles – forming a mutually empowering industry network. We take a deliberate approach to ecosystem building – selectively investing in and continuously nurturing areas characterized by long-term technological scarcity, scenario value, and scaling potential. We believe that in emerging fields with significant room for demand expansion, the maturation of high-quality supply not only meets existing demand but also lowers adoption barriers, expands application boundaries, and stimulates new market demand. As these ecosystem companies gradually complete the “0 to 1 ” phase of technology and product validation and move towards the “1 to 100 ” phase of large-scale commercial application, their industrial and commercial value is being progressively unlocked. During the reporting period, Ace Robotics, SenseCare, and other partners closed fresh rounds of institutional financing – a strong signal that capital markets and industry partners are increasingly backing their technology roadmaps, industry directions, and commercialization progress. During the reporting period, a number of AI ecosystem companies in our strategic equity portfolio completed public listings. The fair value appreciation of these investments materially contributed to our first IFRS profit – a clear sign that the financial value of our ecosystem assets is beginning to crystallize. The long-term value of our ecosystem strategy goes well beyond investment returns. It positions us at the forefront of frontier domains – embodied intelligence, intelligent terminals, next-generation compute platforms – and creates a persistent pipeline of real-world tasks, data, and feedback that strengthens our unified multimodal models and system-level delivery capabilities. In turn, we are building a virtuous cycle where technology capabilities, industry scenarios, and ecosystem companies reinforce one another, aligning financial returns with strategic value.
Page 12
12 Strategic Outlook Looking ahead, AI will evolve from generating content and responding to commands towards understanding complex goals, continuously executing long-horizon tasks, and producing measurable actions and outcomes in both the digital and physical worlds. As capability boundaries continue to expand, competition in AI will likewise shift from individual models or products towards the deep system-level integration of models, Token production, Agent Harness, and real- world scenarios. We will continue to evolve our system-level AI capabilities around the “one model system, one Token Factory, one Agent Harness ” framework, along the following directions: • Driving multimodal agents to continuously evolve through real-world tasks. We will continue to advance the deep integration of native multimodal models and long-horizon agents, enhancing unified understanding, generation, and reasoning capabilities across text, images, video, spatial, and physical world information, with a focus on breakthroughs in complex task decomposition, long-horizon execution, autonomous tool orchestration, dynamic error correction, long-term memory, and multi-agent collaboration. Through data, execution results, and user feedback generated by real-world tasks, we will continuously feed back into model training, evaluation, and workflow optimization, gradually forming an R&D closed loop of “model improvement – task delivery – feedback accumulation – model re-evolution. ” • Driving the Token Factory to evolve from scaled intelligence production towards Task delivery. We will continue to deepen the large-scale commercialization of domestic compute, heterogeneous inference optimization, and compute-energy synergy, significantly lowering the production cost of high-quality Tokens and enhancing the reasoning, understanding, generation, and action capabilities they carry. As agents become capable of handling more complex long-horizon tasks, the efficiency measurement of the Token Factory will gradually extend from per-Token cost to the completion quality, success rate, and cost per task, while the measurement of commercial value will likewise gradually evolve from billing by Token usage towards pricing by tasks and delivered outcomes. • Driving multi-agent collaboration deeper into personal and enterprise workflows. We will continue to refine the Agent Harness, enabling multiple agents with different capabilities and specializations to share context, collaborate on planning, cross-validate each other, and jointly complete complex tasks. Through diverse product forms including work, content, interaction, and industry agents, as well as FDE and accumulated industry knowledge, we will drive AI to evolve from single-point tools and capability demonstrations towards systematic delivery covering “understanding – planning – execution – validation – delivery, ” providing complete, reliable, and sustainable productivity for enterprises, individuals, and OPCs. These directions converge on our core judgment for the next phase of AI: beyond raising the ceiling of model intelligence, the AI of the future must enter more scenarios at lower cost, orchestrating complex tasks, and continuously learning and evolving from real-world outcomes. We will leverage the synergistic advantages of the “one model system, one Token Factory, one Agent Harness ” framework, connecting tasks, actions, and feedback across the digital and physical worlds, continuously expanding the audience and application boundaries of AI capabilities, and driving technology capabilities, customer value, and business growth to form a self-reinforcing long-term cycle.
Page 13
13 MANAGEMENT DISCUSSION AND ANALYSIS Six months ended June 30, 2026 compared to six months ended June 30, 2025 The following table sets forth the comparative figures for the six months ended June 30, 2026 and 2025: Six months ended June 30 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Revenue 2,910,946 2,358,198 Cost of sales (1,704,710) (1,450,398) Gross profit 1,206,236 907,800 Selling expenses (270,933) (272,285) Administrative expenses (517,995) (597,033) Research and development expenses (1,755,585) (2,118,533) Net impairment gains/(losses) on financial assets and contract assets 35,178 (142,732) Other income 120,799 128,974 Other gains – net 2,127,455 640,616 Operating profit/(loss) 945,155 (1,453,193) Finance income 99,343 128,124 Finance costs (81,392) (96,379) Finance income – net 17,951 31,745 Share of losses of investments accounted for using the equity method (269,751) (13,212) Fair value losses of preferred shares (20,063) (18,623) Profit/(Loss) before income tax 673,292 (1,453,283) Income tax expenses (55,984) (35,987) Profit/(Loss) for the period 617,308 (1,489,270) Profit/(Loss) is attributable to: Equity holders of the Company 607,420 (1,477,949) Non-controlling interests 9,888 (11,321) 617,308 (1,489,270) Non-IFRS measures: Adjusted EBITDA 384,855 (540,339) Adjusted net losses (385,898) (1,181,917)
Page 14
14 Revenue Our revenue increased by 23.4% to RMB2,910.9 million in the six months ended June 30, 2026, compared to RMB2,358.2 million in the six months ended June 30, 2025, primarily driven by continued expansion of Generative AI. The following table sets out a breakdown of our revenue streams in absolute amounts and as percentages of our total revenue for the years indicated: Six months ended June 30 2026 2025 RMB million % RMB million % (Unaudited) (Unaudited) Generative AI 2,327.2 79.9 1,815.5 77.0 Computer Vision 496.8 17.1 436.0 18.5 X Businesses 86.9 3.0 106.7 4.5 Total 2,910.9 100.0 2,358.2 100.0 Our Generative AI revenue increased by 28.2% to RMB2,327.2 million in the six months ended June 30, 2026, compared to RMB1,815.5 million in the six months ended June 30, 2025 since market continues to expand steadily, underpinned by rising demand for AI agents and cloud services. Our Computer Vision revenue increased by 13.9% to RMB496.8 million in the six months ended June 30, 2026, compared to RMB436.0 million in the six months ended June 30, 2025. Benefiting from domestic demand recovery and sustained growth in overseas markets, computer vision is entering a second growth phase through multimodal vision agents. Our X Businesses revenue decreased by 18.6% to RMB86.9 million in the six months ended June 30, 2026, compared to RMB106.7 million in the six months ended June 30, 2025 due to deconsolidation of some X Businesses. Over time, we would expect our X Business composition to evolve as we incubate more X Businesses or as existing X Businesses attract external investors and become deconsolidated from our financial statements. As a result, year-over-year comparison will become less meaningful going forward for this particular revenue line.
Page 15
15 Cost of sales Our cost of sales increased by 17.5% from RMB1,450.4 million in the six months ended June 30, 2025 to RMB1,704.7 million in the six months ended June 30, 2026. The following table sets out a breakdown of our cost of sales by nature in absolute amounts and as percentages of our cost of sales for the years indicated: Six months ended June 30 2026 2025 RMB million % RMB million % (Unaudited) (Unaudited) Hardware costs and subcontracting service fees 1,152.4 67.6 1,244.8 85.7 Depreciation and amortization 296.5 17.4 87.5 6.0 AIDC operation costs 232.7 13.7 79.8 5.5 Employee benefit expenses 15.9 0.9 30.0 2.1 Other expenses 7.2 0.4 8.3 0.7 Total 1,704.7 100.0 1,450.4 100.0 The increase in cost of sales was primarily attributable to the increase in AIDC-related depreciation and operational costs, which grew faster than total cost of sales due to our continued expansion of generative AI-related services. Gross Profit and Gross Margin Our gross profit increased by 32.9% from RMB907.8 million in the six months ended June 30, 2025 to RMB1,206.2 million in the six months ended June 30, 2026, outpacing revenue growth rate for the same period. Our gross margin increased from 38.5% in the six months ended June 30, 2025 to 41.4% in the six months ended June 30, 2026, mainly due to the contribution of Generative AI business. Research and Development Expenses Our research and development expenses decreased by 17.1% from RMB2,118.5 million in the six months ended June 30, 2025 to RMB1,755.6 million in the six months ended June 30, 2026, primarily due to the decrease of employee benefit expenses. Selling Expenses Our selling expenses slightly decreased by 0.5% from RMB272.3 million in the six months ended June 30, 2025 to RMB270.9 million in the six months ended June 30, 2026, primarily due to the decrease of employee benefit expenses partially offset by the increase of marketing and travelling expenses.
Page 16
16 Administrative Expenses Our administrative expenses decreased by 13.2% from RMB597.0 million in the six months ended June 30, 2025 to RMB518.0 million in the six months ended June 30, 2026, primarily due to decline in workplace operational costs, including depreciation and amortization, rental fees, property management and administrative expenses. Net Impairment Gains/(Losses) on Financial Assets and Contract Assets Our net impairment losses on financial assets of RMB142.7 million in the six months ended June 30, 2025 have reversed into net gains amounting to RMB35.2 million in the six months ended June 30, 2026. This favorable turnaround is primarily attributable to our ongoing disciplined efforts in collecting outstanding trade receivables. Other Income Our other income decreased by 6.4% from RMB129.0 million in the six months ended June 30, 2025 to RMB120.8 million in the six months ended June 30, 2026 primarily due to a slight decrease in government grants. Other gains – Net We had net other gains of RMB2,127.5 million in the six months ended June 30, 2026 compared to RMB640.6 million in the six months ended June 30, 2025. The net other gains in the six months ended June 30, 2026 primarily consist of gains on disposal of subsidiaries amounting to RMB521.0 million and net fair value unrealized gains on financial assets at fair value through profit or loss amounting to RMB1,248.7 million after offset by net asset value of investment funds attributable to limited partners. Finance Income – Net Net finance income decreased by 43.2% from RMB31.7 million in the six months ended June 30, 2025 to RMB18.0 million in the six months ended June 30, 2026 primarily due to slight decrease in net interest income. Fair Value Losses of Preferred Shares Fair value losses of preferred shares increased from RMB18.6 million in the six months ended June 30, 2025 to RMB20.1 million in the six months ended June 30, 2026 resulting from changes in the fair value change of preferred shares issued by a subsidiary of the Company. Income tax expenses We had income tax expenses of RMB56.0 million in the six months ended June 30, 2026 compared to RMB36.0 million in the six months ended June 30, 2025 primarily attributable to enhanced profitability.
Page 17
17 Profit/(Loss) for the period As a result of the foregoing, we had a profit of RMB617.3 million in the six months ended June 30, 2026, compared with a loss of RMB1,489.3 million in the six months ended June 30, 2025. The turnaround from a net loss to net profit this year was primarily driven by a combination of factors, including improved revenue and gross profit performance, effective control and reduction of operating cost, and fair value increase of our various strategic investments. Non-IFRS Measures To supplement our consolidated results which are prepared and presented in accordance with IFRS, we also use EBITDA/adjusted EBITDA and adjusted net loss as additional financial measures, which are not required by, or presented in accordance with, IFRS. We believe that these non-IFRS measures facilitate comparisons of operating performance from year to year and company to company by eliminating potential impacts of items that our management does not consider to be indicative of our operating performance such as certain non-cash items and certain impact of financing and investment activities. We believe that these measures provide useful information to investors and others in understanding and evaluating our consolidated results of operations in the same manner as they help our management. However, our presentation of the EBITDA/adjusted EBITDA and adjusted net loss may not be comparable to similarly titled measures presented by other companies. The use of these non-IFRS measures has limitations as an analytical tool, and you should not consider them in isolation from, or as substitute for analysis of, our results of operations or financial condition as reported under IFRS. The following table sets out EBITDA/adjusted EBITDA and a reconciliation from profit/(loss) before income tax for the periods to EBITDA/adjusted EBITDA for the periods indicated: Six months ended June 30 2026 2025 RMB million RMB million (Unaudited) (Unaudited) Profit/(Loss) before income tax 673.3 (1,453.3) Add: Finance income, net (18.0) (31.7) Depreciation and amortization 732.7 637.3 EBITDA 1,388.1 (847.7) Add: Fair value losses of preferred shares 20.1 18.6 Share-based compensation expenses 225.4 153.3 Fair value unrealized (gains)/losses on financial assets at fair value through profit or loss (4,446.3) 155.3 Net asset value of investment funds attributable to limited partners 3,197.6 (19.8) Adjusted EBITDA 384.9 (540.3)
Page 18
18 The following table reconciles our adjusted net profit/(loss) for the periods presented to the most directly comparable financial measure calculated and presented in accordance with IFRS, which is net loss for the periods: Six months ended June 30 2026 2025 RMB million RMB million (Unaudited) (Unaudited) Net profit/(losses) for the period 617.3 (1,489.3) Add: Fair value losses of preferred shares 20.1 18.6 Share-based compensation expenses 225.4 153.3 Fair value unrealized (gains)/losses on financial assets at fair value through profit or loss (4,446.3) 155.3 Net asset value of investment funds attributable to limited partners 3,197.6 (19.8) Adjusted net losses (385.9) (1,181.9) Trade, Other Receivables and Prepayments Our trade, other receivables and prepayments increased from RMB4,304.1 million as of December 31, 2025 to RMB5,290.8 million as of June 30, 2026, which primarily comprise trade receivables and other receivables. Trade Receivables The following table sets out a breakdown of our trade receivables as of the dates indicated: As at June 30, As at December 31, 2026 2025 RMB million RMB million (Unaudited) Trade receivables – Due from related parties 355.9 368.5 – Due from third parties 6,683.5 5,526.9 Provision for impairment (3,276.5) (3,384.9) Total 3,762.9 2,510.4 Our net trade receivables increased as of June 30, 2026, compared to that as of December 31, 2025.
Page 19
19 Trade receivables ’ aging analysis based on date of revenue recognition is as follows: As at June 30, As at December 31, 2026 2025 RMB million RMB million (Unaudited) Up to 6 months 2,165.9 1,896.2 6 months to 1 year 1,233.6 309.7 1 to 2 years 261.7 235.1 2 to 3 years 116.9 118.7 More than 3 years 3,261.3 3,335.6 7,039.4 5,895.3 LIQUIDITY AND FINANCIAL RESOURCES We have historically funded our cash requirements principally from capital contribution from shareholders. We had cash and cash equivalents of RMB9,877.2 million and term deposits of RMB3,124.0 million as of June 30, 2026, compared to the balance of RMB10,887.1 million and RMB2,282.7 million as of December 31, 2025. The following table sets forth a summary of our cash flows for the years indicated: Six months ended June 30 2026 2025 RMB million RMB million (Unaudited) (Unaudited) Net cash used in operating activities (763.7) (639.5) Net cash (used in)/generated from investing activities (3,744.9) 699.7 Net cash generated from financing activities 3,668.8 2,158.0 Net (decrease)/increase in cash and cash equivalents (839.8) 2,218.2 Cash and cash equivalents at the beginning of the year 10,887.1 8,888.0 Exchange changes on cash and cash equivalents (170.0) 62.2 Cash and cash equivalents at the end of the period 9,877.2 11,168.4 Net Cash Used in Operating Activities Net cash used in operating activities represents the cash used in our operations plus the income tax paid. Cash used in our operations primarily comprises our loss before income tax adjusted by non-cash items and changes in working capital.
Page 20
20 For the six months ended June 30, 2026, net cash used in operating activities was RMB763.7 million, which was primarily attributable to our loss before income tax, as adjusted by (i) depreciation and amortization, fair value gains on financial assets at fair value through profit or loss, provision for impairment of financial assets and contract assets and gains on disposal of subsidiaries; and (ii) changes in operating assets and liabilities. Net cash used in operating activities increased during the period as the reduction in operating losses was offset by an increase in working capital requirements, primarily due to a higher trade receivables balance resulting from, among other factors, the timing of customer collections. Net Cash (Used in)/Generated from Investing Activities For the six months ended June 30, 2026, net cash used in investing activities was RMB3,744.9 million, which was mainly attributable to purchase of property, plant and equipment and acquisition of investments in financial assets at fair value through profit and loss. Net Cash Generated from Financing Activities For the six months ended June 30, 2026, net cash generated from financing activities was RMB3,668.8 million, which was mainly attributable to net proceeds from placing ordinary shares and capital injection by limited partners of investment fund controlled by the Group. Borrowings As of December 31, 2025 and June 30, 2026, we had total borrowings of RMB5,724.6 million and RMB6,076.4 million, respectively. Repayment analysis and the currency denomination of bank borrowings of the Group as at June 30, 2026 are set out in note 15 to the consolidated financial statements. The Group maintains a prudent approach in its treasury management with interest rate exposure maintained principally on a floating rate basis. The Group did not use any interest rate swap contracts or other financial instruments to hedge against its interest rate risk. The Group will continue to monitor interest rate risk exposure and will consider hedging significant interest rate risk exposure should the need arises. Exposure to Exchange Rate Fluctuation Foreign exchange risk arises when future commercial transactions or recognized assets and liabilities are denominated in a currency that is not the respective Group entities ’ functional currency. The Company ’s functional currency is USD. The Company ’s primary subsidiaries were incorporated in Mainland China, Hong Kong, Singapore and Middle East. These subsidiaries considered RMB, HKD, SGD and SAR as their functional currencies, respectively. We are primarily exposed to fluctuations in the HKD/RMB and HKD/USD exchange rates. We did not engage in hedging activities designed and intended to manage foreign exchange rate risk as of June 30, 2026. However, we will continue to monitor foreign exchange rate movements and will take necessary measures to mitigate the impact of exchange rate fluctuations.
Page 21
21 Employees, Training and Remuneration Policies As at June 30, 2026, the Group had 2,668 employees. The number of employees employed by the Group varies from time to time depending on needs. The Group formulates the remuneration package for its employees based on the overall remuneration standard in the market, industry practice and the Group ’s remuneration strategy. In addition to salary, in-house training programmes and employee benefits, employees may receive year-end performance incentives depending on their individual performance, which includes cash incentives or share options. Use of Proceeds from Placing of Class B Shares under General Mandate December 2025 Placing On December 18, 2025, the Company entered into a placing agreement (the “December 2025 Placing Agreement ”) with Guotai Junan Securities (Hong Kong) Limited and CMB International Capital Limited in relation to the placing of 1,750,000,000 new Class B Shares (the “December 2025 Placing Shares ”) at HK$1.80 per December 2025 Placing Share (the “December 2025 Placing Price ”) to not less than six placees who and whose ultimate beneficial owner(s) (where applicable) shall be independent third parties (the “December 2025 Placing ”). All the conditions set out in the December 2025 Placing Agreement were fulfilled and the December 2025 Placing was completed on December 29, 2025, where a total of 1,750,000,000 new Class B Shares (with an aggregate nominal value of US$43.75), representing approximately 4.40% of the number of the then issued Class B Shares and approximately 4.33% of the number of the then existing issued Shares as enlarged by the allotment and issue of such placing shares, have been successfully placed to no fewer than six independent placees (which were individual, professional, corporate and/or institutional investors) at the placing price of HK$1.80 pursuant to the terms and conditions of the December 2025 Placing Agreement. The closing price of the Class B Share as quoted on the Stock Exchange on December 17, 2025, being the date on which the December 2025 Placing Price was fixed, was HK$1.97. For details, please refer to the announcements of the Company dated December 18, 2025 and December 29, 2025 respectively. The Directors considered that the December 2025 Placing represents a suitable financing option for the Company to raise further funding to support the Group ’s continuous development and business growth taking into account the then recent market conditions, which is in the interest of the Company and its Shareholders as a whole. The net proceeds from the December 2025 Placing, after deducting the placing commission and other relevant costs and expenses of the December 2025 Placing, amounted to approximately HK$3,146 million, representing a net issue price of approximately HK$1.80 per December 2025 Placing Shares. The Company intends to use the net proceeds to support its core business development, which includes building an industry-leading AI Cloud and expanding its AI infrastructure (SenseCore) with increased adoption of domestic chips; funding the R&D and commercialization of Generative AI and its Multimodality Large Model; exploring AI integration in innovative vertical fields such as finance, education and marketing, and for general working capital purposes covering operating costs, talent development, branding and sales initiatives. As at June 30, 2026, all of such net proceeds had been utilised by the Group.
Page 22
22 Details of the use of proceeds are set out as follows: Intended purposes of placing net proceeds Planned proportion of the net proceeds Planned use of the net proceeds Actual use of net proceeds during the Reporting Period Net proceeds unused as of June 30, 2026 Expected timeline for fully utilising the net proceeds (%) (Approximately HK$ million) (Approximately HK$ million) (Approximately HK$ million) Supporting the development of the Company ’s core business 30.0 943.7 943.7 – NA Supporting the R&D of Generative AI, productization of the Company ’s Multimodality Large Model 30.0 943.7 943.7 – NA Exploring the integration and application of AI in innovative fields 20.0 629.1 629.1 – NA General working capital 20.0 629.1 629.1 – NA Total 100.0 3,145.7 3,145.7 – April 2026 Placing On April 16, 2026, the Company entered into a placing agreement (the “April 2026 Placing Agreement ”) with The Hongkong and Shanghai Banking Corporation Limited, in relation to the placing of 1,700,000,000 new Class B Shares (the “April 2026 Placing Shares ”) at HK$1.91 per April 2026 Placing Share (the “April 2026 Placing Price ”) to not less than six placees who and whose ultimate beneficial owner(s) (where applicable) shall be independent third parties (the “April 2026 Placing ”). All the conditions set out in the April 2026 Placing Agreement were fulfilled and the April 2026 Placing Agreement was completed on April 24, 2026, where total of 1,700,000,000 new Class B Shares (with an aggregate nominal value of US$42.50) representing approximately 4.25% of the then number of issued Class B Shares completion of the April 2026 Placing Agreement and approximately 4.03% of the number of then existing issued Shares as enlarged by the allotment and issue of the April 2026 Placing Shares. The closing price of the Class B Shares as quoted on the Stock Exchange on April 16, 2026, being the date on which the April 2026 Placing Price was fixed, was HK$2.09 per Class B Shares. For details, please refer to the announcements of the Company dated April 17, 2026 and April 21, 2026 respectively. The Directors considered that the April 2026 Placing represents a suitable financing option for the Company to raise further funding to support the Group ’s continuous development and business growth taking into account the then recent market conditions, which is in the interest of the Company and its Shareholders as a whole.
Page 23
23 The net proceeds from the April 2026 Placing, after deducting the placing commission and other relevant costs and expenses of the April 2026 Placing, amounted to approximately HK$3,230 million, representing a net issue price of approximately HK$1.90 per April 2026 Placing Shares. The Company intends to use the net proceeds to expand the scale of SenseCore, the Company ’s AI infrastructure, with focus on the scaling up of the domestic AI supercomputing clusters empowered by domestic chips, as well as to enhance the offerings of our domestic AI cloud stack, providing infrastructure for AI token plan; support the Company ’s R&D of Generative AI, with a core focus on the continuous innovation of our SenseNova model series so as to lead globally, including but not limited to the development of our native multimodality large model built on the proprietary NEO architecture, as well as to scale up the offerings of our AI tokens plan to realize commercialization; explore the innovative integration and application of AI technology in vertical scenarios, including but not limited to embedding native AI into education, software as a service, software embedded hardware, and for general working capital. Details of the use of proceeds are set out as follows: Intended purposes of placing net proceeds Planned proportion of the net proceeds Planned use of the net proceeds Actual use of net proceeds during the Reporting Period Net proceeds unused as of June 30, 2026 Expected timeline for fully utilising the net proceeds (%) (Approximately HK$ million) (Approximately HK$ million) (Approximately HK$ million) Expand the scale of SenseCore, the Company ’s AI infrastructure 40.0 1,291.9 1,291.9 – NA Supporting the R&D of Generative AI 30.0 968.9 534.2 434.7 December 31,2026 Exploring the innovative integration and application of AI technology in vertical scenarios 20.0 646.0 646.0 – NA General working capital 10.0 323.0 161.0 162.0 December 31,2026 Total 100.0 3,229.8 2,633.1 596.7
Page 24
24 Gearing Ratio As of June 30, 2026, our gearing ratio was -8.6%, which represented a net cash position. Our gearing ratio is calculated as net debt divided by total capital at the end of each financial year. Net debt equals to our total borrowings, lease liabilities and preferred share liabilities less our cash and cash equivalents. Total capital is calculated as total equity plus net debt. Contingent Liabilities As of June 30, 2026, we did not have any material contingent liabilities. Significant Investments held As of June 30, 2026, we did not hold any significant investments in the equity interest of other companies. Material Acquisitions and Disposals of Subsidiaries, Associates and Joint Ventures For the six months ended June 30, 2026, we did not have any material acquisitions or disposals of subsidiaries, associates and joint ventures. Pledge of Assets As of June 30, 2026, we pledged certain equipments, buildings and land use rights with carrying amount of RMB4,042.3 million for borrowings. Future Plans for Material Investments and Capital Assets As of June 30, 2026, we have no specific future plan for material investments and acquisition of capital assets. The Group will continue to identify new investment opportunities in companies with principal businesses related to the Group ’s core business with a view to create synergies with the Group’s existing core business and improve the Group ’s service and products to its customers. EVENTS AFTER THE REPORTING PERIOD Save as otherwise disclosed in this announcement, there were no other significant events that may affect the Group since the end of the Reporting Period.
Page 25
25 FINANCIAL INFORMATION INTERIM CONDENSED CONSOLIDATED INCOME STATEMENT FOR THE SIX MONTHS ENDED 30 JUNE 2026 Six months ended 30 June 2026 2025 Notes RMB’000 RMB’000 (Unaudited) (Unaudited) Revenue 4(a) 2,910,946 2,358,198 Cost of sales 5 (1,704,710) (1,450,398) Gross profit 1,206,236 907,800 Selling expenses 5 (270,933) (272,285) Administrative expenses 5 (517,995) (597,033) Research and development expenses 5 (1,755,585) (2,118,533) Net impairment gains/(losses) on financial assets and contract assets 35,178 (142,732) Other income 120,799 128,974 Other gains – net 6 2,127,455 640,616 Operating profit/(loss) 945,155 (1,453,193) Finance income 99,343 128,124 Finance costs (81,392) (96,379) Finance income – net 17,951 31,745 Share of losses of investments accounted for using the equity method (269,751) (13,212) Fair value losses of preferred shares (20,063) (18,623) Profit/(loss) before income tax 673,292 (1,453,283) Income tax expenses 7 (55,984) (35,987) Profit/(loss) for the period 617,308 (1,489,270) Profit/(loss) is attributable to: Equity holders of the Company 607,420 (1,477,949) Non-controlling interests 9,888 (11,321) 617,308 (1,489,270) Earnings/(losses) per share for profit/(loss) attributable to equity holders of the Company Basic and diluted earnings/(losses) per share (RMB) 8 0.01 (0.04) The above interim condensed consolidated income statement should be read in conjunction with the accompanying notes.
Page 26
26 INTERIM CONDENSED CONSOLIDATED STATEMENT OF COMPREHENSIVE INCOME FOR THE SIX MONTHS ENDED 30 JUNE 2026 Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Profit/(loss) for the period 617,308 (1,489,270) Other comprehensive (loss)/income Item that may be reclassified to profit or loss Exchange differences on translation of foreign operations 1,187,925 336,717 Item that will not be reclassified to profit or loss Exchange differences on translation of foreign operations (1,691,815) (206,859) Revaluation gains upon transfer from property, plant and equipment to investment properties – 3,558 Effects of changes in credit risk for financial liabilities designated as at fair value through profit or loss (4,633) (3,848) Other comprehensive (loss)/income for the period, net of taxes (508,523) 129,568 Total comprehensive income/(loss) for the period 108,785 (1,359,702) Total comprehensive income/(loss) for the period is attributable to: Equity holders of the Company 98,927 (1,348,381) Non-controlling interests 9,858 (11,321) 108,785 (1,359,702) The above interim condensed consolidated statement of comprehensive loss should be read in conjunction with the accompanying notes.
Page 27
27 INTERIM CONDENSED CONSOLIDATED BALANCE SHEET AS AT 30 JUNE 2026 As at 30 June As at 31 December 2026 2025 Notes RMB’000 RMB’000 (Unaudited) Assets Non-current assets Property, plant and equipment 10 7,817,212 5,883,256 Right-of-use assets 232,161 282,692 Investment properties 2,507,845 2,512,134 Intangible assets 114,885 121,279 Contract assets 11,525 15,791 Investments accounted for using the equity method 1,504,694 1,426,743 Deferred income tax assets 658,544 654,554 Financial assets at fair value through profit or loss 12 12,780,648 7,510,261 Long-term receivables 38,879 35,118 Restricted cash 27,846 21,601 Other non-current assets 1,333,006 1,658,231 27,027,245 20,121,660 Current assets Inventories 410,748 248,115 Contract assets 2,728 3,230 Trade, other receivables and prepayments 11 5,290,769 4,304,083 Financial assets at fair value through profit or loss 12 741,776 571,969 Derivative financial instrument 13 – 3,097 Restricted cash 74,008 483,707 Term deposits 3,124,021 2,282,664 Cash and cash equivalents 9,877,244 10,887,050 19,521,294 18,783,915 Total assets 46,548,539 38,905,575 Equity Equity attributable to equity holders of the Company Share capital 7 7 Other reserves 84,767,221 81,472,449 Currency translation reserves 3,421,123 3,924,983 Accumulated losses (58,071,060) (58,678,480) 30,117,291 26,718,959
Page 28
28 As at 30 June As at 31 December 2026 2025 Notes RMB’000 RMB’000 (Unaudited) Non-controlling interests 144,302 341,944 Total equity 30,261,593 27,060,903 Liabilities Non-current liabilities Borrowings 15 3,814,128 3,710,335 Lease liabilities 146,376 176,010 Deferred income tax liabilities 10,352 16,758 Contract liabilities 34,257 39,361 Deferred revenue 217,573 209,500 Other financial liabilities 5,467,735 1,959,550 9,690,421 6,111,514 Current liabilities Borrowings 15 2,262,316 2,014,250 Trade and other payables 14 2,601,507 2,090,358 Lease liabilities 58,387 62,358 Contract liabilities 316,280 228,914 Deferred revenue 82,830 84,681 Current income tax liabilities 63,412 27,217 Preferred share liabilities 1,211,793 1,225,380 6,596,525 5,733,158 Total liabilities 16,286,946 11,844,672 Total equity and liabilities 46,548,539 38,905,575 The above interim condensed consolidated balance sheet should be read in conjunction with the accompanying notes.
Page 29
29 INTERIM CONDENSED CONSOLIDATED STATEMENT OF CHANGES IN EQUITY FOR THE SIX MONTHS ENDED 30 JUNE 2026 Equity attributable to equity holders of the Company Share capital Other reserves Currency translation reserves Accumulated losses Total Non- controlling interests Total equity Notes RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 (Unaudited) As at 1 January 2026 7 81,472,449 3,924,983 (58,678,480) 26,718,959 341,944 27,060,903 Comprehensive loss Profit for the period – – – 607,420 607,420 9,888 617,308 Effects of changes in credit risk for financial liabilities designated as at fair value through profit or loss – (4,633) – – (4,633) – (4,633) Exchange differences on translation of foreign operations – – (503,860) – (503,860) (30) (503,890) Total comprehensive loss – (4,633) (503,860) 607,420 98,927 9,858 108,785 Transactions with equity holders Placing of ordinary shares * 2,831,055 – – 2,831,055 – 2,831,055 Treasury shares issued to employees – 35,474 – – 35,474 – 35,474 Transactions with non-controlling interests – 257,500 – – 257,500 (257,500) – Recognition of financial instruments with preferred rights at amortised cost – (50,000) – – (50,000) 50,000 – Share-based compensation expenses 5 – 225,376 – – 225,376 – 225,376 Total transactions with equity holders – 3,299,405 – – 3,299,405 (207,500) 3,091,905 As at 30 June 2026 7 84,767,221 3,421,123 (58,071,060) 30,117,291 144,302 30,261,593
Page 30
30 Equity attributable to equity holders of the Company Share capital Other reserves Currency translation reserves Accumulated losses Total Non- controlling interests Total equity Notes RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 (Unaudited) As at 1 January 2025 6 76,069,902 4,303,417 (56,912,551) 23,460,774 180,931 23,641,705 Comprehensive loss Loss for the period – – – (1,477,949) (1,477,949) (11,321) (1,489,270) Effects of changes in credit risk for financial liabilities designated as at fair value through profit or loss – (3,848) – – (3,848) – (3,848) Revaluation gains upon transfer from property, plant and equipment to investment properties – 3,558 – – 3,558 – 3,558 Exchange differences on translation of foreign operations – – 129,858 – 129,858 – 129,858 Total comprehensive loss – (290) 129,858 (1,477,949) (1,348,381) (11,321) (1,359,702) Transactions with equity holders Treasury shares issued to employees – 35,552 – – 35,552 – 35,552 Transactions with non-controlling interests – 40,000 – – 40,000 (40,000) – Recognition of financial instruments with preferred rights at amortised cost – (62,500) – – (62,500) 62,500 – Share-based compensation expenses 5 – 153,250 – – 153,250 – 153,250 Total transactions with equity holders – 166,302 – – 166,302 22,500 188,802 As at 30 June 2025 6 76,235,914 4,433,275 (58,390,500) 22,278,695 192,110 22,470,805 * represents that the amount is less than RMB1,000 for the period. The above interim condensed consolidated statement of changes in equity should be read in conjunction with the accompanying notes.
Page 31
31 INTERIM CONDENSED CONSOLIDATED STATEMENT OF CASH FLOWS FOR THE SIX MONTHS ENDED 30 JUNE 2026 Six months ended 30 June 2026 2025 Notes RMB’000 RMB’000 (Unaudited) (Unaudited) Cash flows from operating activities Cash used in operations (734,461) (628,372) Income tax paid (29,270) (11,113) Net cash used in operating activities (763,731) (639,485) Cash flows from investing activities Purchase of property, plant and equipment (2,292,070) (1,389,008) Purchase of intangible assets (44,993) (70,932) Proceeds from disposal of property, plant and equipment 113,874 240,871 Net cash outflow from disposal of subsidiaries (130,099) (138,542) Increase in investments accounted for using the equity method (27,500) (4,250) Dividend received from financial assets at fair value through profit and loss 7,322 6,789 Acquisition of debt and equity investments 12(a), (b) (1,386,885) (244,950) Acquisition of wealth management products 12(c) (6,873,206) (6,788,000) Disposal of debt and equity investments 12(a), (b) 428,549 341,044 Redemption of wealth management products and derivative financial instruments 12(c) 6,722,740 5,655,073 Net (increase)/decrease in investments in term deposits (872,085) 2,936,461 Decrease in restricted cash for derivative instruments 351,447 – Proceeds from disposal of associates 174,488 3,536 Interest received from banks 83,488 151,578 Net cash (used in)/generated from investing activities (3,744,930) 699,670
Page 32
32 Six months ended 30 June 2026 2025 Notes RMB’000 RMB’000 (Unaudited) (Unaudited) Cash flows from financing activities Proceeds from borrowings 1,734,305 3,882,748 Repayments of borrowings (1,382,104) (1,633,071) Interest paid (71,160) (89,736) Distribution to limited partners (78,488) – Principal elements of lease payments (22,794) (66,743) Interests elements of lease payments (3,704) (5,189) Capital injection by limited partners of investment fund controlled by the Group 597,857 – Capital injection by non-controlling shareholder 50,000 62,500 Net proceeds from placing of ordinary shares 2,831,055 – Proceeds from exercise of share options 13,849 7,522 Net cash generated from financing activities 3,668,816 2,158,031 Net (decrease)/increase in cash and cash equivalents (839,845) 2,218,216 Cash and cash equivalents at beginning of period 10,887,050 8,887,988 Effect of foreign exchange rates changes (169,961) 62,171 Cash and cash equivalents at end of period 9,877,244 11,168,375 The above interim condensed consolidated statement of cash flows should be read in conjunction with the accompanying notes.
Page 33
33 NOTES TO THE INTERIM CONDENSED CONSOLIDATED FINANCIAL INFORMATION FOR THE SIX MONTHS ENDED 30 JUNE 2026 1 GENERAL INFORMATION SenseTime Group Inc. (the “Company ”) was incorporated in the Cayman Islands on 15 October 2014 as an exempted company with limited liability under the Companies Law (Cap. 22, Law 3 of 1961 as consolidated and revised) of the Cayman Islands. The address of the Company ’s registered office is Vistra (Cayman) Limited, of P.O. Box 31119 Grand Pavilion, Hibiscus Way, 802 West Bay Road, Grand Cayman, KY1-1205, Cayman Islands. The Company is an investment holding company. The principal activities of the Company and its subsidiaries including the structured entities (collectively, the “Group”), are the sale of advanced artificial intelligence (“AI”) software, sale of AI software platform and related services, sale of software-embedded hardware and related services, AIDC service as well as research and development activities in relation to AI technology mainly in the People ’s Republic of China (the “PRC”), Northeast Asia, Southeast Asia and other geographical areas. The Company is a leading AI software company with customers across a broad spectrum of industries. The Group does not have ultimate holding company and controlling shareholder. On 30 December 2021, the Company has successfully listed on the Main Board of the Stock Exchange of Hong Kong Limited. The interim condensed consolidated financial information is presented in thousands of Renminbi ( “RMB’000”), unless otherwise stated. This interim condensed consolidated financial information was approved for issue by the board of directors on 26 August 2026. 2 BASIS OF PREPARATION This interim condensed consolidated financial information for the six months ended 30 June 2026 has been prepared in accordance with International Accounting Standard IAS 34 Interim Financial Reporting . The interim condensed consolidated financial information should be read in conjunction with the consolidated financial statements of the Group for the year ended 31 December 2025, which have been prepared in accordance with IFRS Accounting Standards (IFRS), as set out in the Company ’s annual report dated 26 March 2026 except for the adoption of amended standard as disclosed in Note 3.
Page 34
34 3 NEW STANDARDS AND INTERPRETATIONS Except as described below, the accounting policies applied are consistent with those of the annual financial statements for the year ended 31 December 2025, as described in those annual financial statements. Taxes on income for the interim period are accrued using the estimated tax rates that would be applicable to expected total annual assessable profit. (a) New or amended standards adopted by the Group A number of amended standards became applicable for the current reporting period. The following amendment to standards have been adopted by the Group for the first time for the financial year beginning on 1 January 2026: IFRS 7 and IFRS 9 (Amendment) Amendments to the Classification and Measurement of Financial Instruments IFRS 7 and IFRS 9 Contracts Referencing Nature-dependent Electricity Annual Improvements to IFRS Annual Improvements to IFRS Accounting Standards The adoption of amended standards did not have any material impact on the amounts recognised in prior periods and are not expected to significantly affect the current or future periods. (b) New standards, amendments and interpretations not yet adopted The followings new standards and amendments to standards have not come into effect for the financial year beginning 1 January 2026 and have not been early adopted by the Group in preparing the interim condensed consolidated financial statements. According to the preliminary assessment made by the directors of the Group, no significant impact on the financial performance and positions of the Group is expected when they become effective. Effective for annual periods Beginning on or after IFRS 19 and Amendment Subsidiaries without Public Accountability: Disclosures 1 January 2027 IFRS 18 Presentation and Disclosure in Financial Statements 1 January 2027 IAS 21 (Amendment) Transaction to a Hyperinflationary Presentation Currency 1 January 2027 IAS 28 (Amendment) Amendment to the Fair Value Option 1 January 2027 IFRS 10 and IAS 28 (Amendment) Sale or Contribution of Assets between an investor and its Associate or Joint Venture To be determined Amendment to illustrative Examples on IFRS 7, IFRS 18, IAS 1, IAS 8, IAS 36 and IAS 37 Disclosure about Uncertainties in the Financial Statements To be determined
Page 35
35 4 SEGMENT INFORMATION The Company develops software and hardware products for different industry verticals based on the same AI infrastructure platform and model training framework. The technologies and nature of the products of different business lines are substantially similar. The executive directors of the Company, who are the chief operating decision makers, for the purpose of resource allocation and assessment of performance did not discrete operating segment financial information and the executive directors reviewed the financial results of the Group as a whole. Therefore, no further information about the operating segment is presented. (a) Revenue by geographical areas The Company is domiciled in the Cayman Islands while the Group mainly operates its businesses in four principal geographical areas of the world. The following table shows the Group ’s total consolidated revenue by location of the customers during the six months ended 30 June 2026: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Mainland China 2,498,124 2,176,347 Northeast Asia 47,440 61,356 Southeast Asia 220,448 21,043 Others* 144,934 99,452 2,910,946 2,358,198 * Other geographical areas mainly represented Hong Kong China and Middle East. (b) Non-current assets by geographical areas The total of the non-current assets including property, plant and equipment, right-of-use assets, investment properties and intangible assets as at 30 June 2026, broken down by the location of the assets, is as follows: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Mainland China 10,468,374 8,567,687 Northeast Asia 26,954 32,380 Southeast Asia 13,937 12,463 Others 162,838 186,831 10,672,103 8,799,361
Page 36
36 5 EXPENSES BY NATURE The expenses charged to cost of sales, selling expenses, administrative expenses and research and development expenses are analysed below: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Employee benefit expenses (a) 1,165,792 1,486,608 Hardware costs and project subcontracting service fees 1,152,404 1,244,813 Depreciation and amortization 732,720 637,336 Server operation and cloud based service fees 729,160 574,032 Professional service and other consulting fees 216,421 212,122 Marketing, conference and travelling expenses 89,251 67,963 Utilities, property management and administrative expenses 64,764 88,008 Research and development tools and consumables 51,278 7,942 Data labelling fees 19,775 79,290 Taxes and surcharges 15,346 22,537 Other expenses 12,312 17,598 4,249,223 4,438,249 (a) During the six months ended 30 June 2026, employee benefits expenses included share-based compensation expenses of approximately RMB225,376,000 (for the six months ended 30 June 2025: RMB153,250,000). 6 OTHER GAINS – NET Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Fair value gains/(losses) on financial assets at fair value through profit or loss (Note 12(d)) 4,588,084 (155,264) Fair value changes on net asset value of investment funds attributable to limited partners (3,197,634) 19,784 Gains on disposal of subsidiaries (i) 521,024 937,640 Gains on disposal of investments accounted for using the equity method and FVPL (ii) 80,397 – Net foreign exchange gains/(losses) 113,330 (168,990) Gains/(losses) on disposal of property, plant and equipment 393 (10,177) Others 21,861 17,623 2,127,455 640,616 (i) During the six months ended 30 June 2026, the Group disposed its interests in certain subsidiaries to third parties and lost control over these subsidiaries and they became joint ventures of the Group. The Group recognised net gains on disposal of subsidiaries of RMB521,024,000 based on the fair value of the joint ventures of RMB459,600,000 on disposal date and the carrying amount of net liabilities of these subsidiaries amounted to RMB61,424,000 on disposal date.
Page 37
37 (ii) During the six months ended 30 June 2026, the Group disposed its interests in certain associates and joint ventures to third parties. The Group recognised net gain amounted RMB38,994,000 based on cash consideration of RMB173,550,000 and the carrying amount of these companies amounted RMB134,556,000 on disposal date. During the six months ended 30 June 2026, the Group recognised gains of RMB41,943,000 arising from the disposal of certain FVPL financial assets. 7 INCOME TAX EXPENSE (i) Cayman Islands The Company was redomiciled in the Cayman Islands in 2014 as an exempted company with limited liability, and is exempted from Cayman Islands income tax under the current tax laws of the Cayman Islands. In addition, no Cayman Islands withholding tax is imposed upon any payments of dividends. (ii) British Virgin Islands Under the current laws of the British Virgin Islands, entities incorporated in British Virgin Islands are not subject to tax on their income or capital gains. (iii) Hong Kong China Entities incorporated in Hong Kong are subject to Hong Kong profits tax of which the tax rate is 8.25% for assessable profits in the first HKD2 million and the remaining assessable profits are taxed at 16.5% (for the six months ended 30 June 2025: 16.5%). (iv) Singapore Entities incorporated in Singapore are subject to income tax at a rate of 17% for the six months ended 30 June 2026 (for the six months ended 30 June 2025: 17%). (v) Malaysia Malaysia income tax rate is 24% during the six months ended 30 June 2026 (for the six months ended 30 June 2025: 24%). In the case that the paid-up capital is Malaysian Ringgit ( “MYR”) 2.5 million or less, and the gross income from business is not more than MYR50 million, the income tax rate on the first MYR0.15 million chargeable income is 15%, the income tax rate on the next MYR0.45 million chargeable income is 17% and the part in excess of MYR0.6 million is 24%. (vi) Saudi Arabia Enterprises incorporated in Saudi Arabia are subject to income tax rate of 20% for the six months ended 30 June 2026 (for the six months ended 30 June 2025: 20%) on foreign ownership portion. (vii) The United Arab Emirates Enterprises incorporated in the United Arab Emirates are subject to UAE corporate tax at a rate of 9% where the taxable income exceeding AED0.375 million for the six months ended 30 June 2026 (for the six months ended 30 June 2025: 9%). (viii) PRC corporate income tax ( “CIT”) The income tax provision of the Group in respect of its operations in the Mainland China was subject to statutory tax rate of 25% on the assessable profits for the six months ended 30 June 2026 (for the six months ended 30 June 2025: 25%), based on the existing legislation, interpretations and practices in respect thereof.
Page 38
38 Some of the subsidiaries were qualified as “High and New Technology Enterprises ” ( “HNTEs”) under the relevant PRC laws and regulations. Accordingly, these entities were entitled to a preferential income tax rate of 15% in 2026. (ix) PRC Withholding Tax ( “WHT”) According to the applicable PRC tax regulations, dividends distributed by a company established in the PRC to a foreign investor with respect to profits derived after 1 January 2008 are generally subject to a 10% WHT. If a foreign investor incorporated in Hong Kong China meets the conditions and requirements under the double taxation treaty arrangement entered into between the PRC and Hong Kong China, the relevant withholding tax rate will be reduced from 10% to 5%. During the six months ended 30 June 2026, no deferred income tax liability on WHT was accrued as at the end of each reporting period because the subsidiaries of the Group were primarily loss making in these periods (for the six months ended 30 June 2025: nil). Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Current income tax 69,494 9,996 Deferred income tax (13,510) 25,991 Income tax expenses 55,984 35,987 8 EARNINGS/(LOSSES) PER SHARE Basic The basic earnings/(losses) per share is calculated by dividing the profit/(loss) attributable to equity holders of the Company by the weighted average number of ordinary shares (excluding treasury shares) issued during the six months ended 30 June 2026: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Profit/(loss) attributable to equity holders of the Company 607,420 (1,477,949) Weighted average number of ordinary shares in issue (thousand) 40,687,080 36,109,541 Basic earnings/(losses) per share (expressed in RMB per share) 0.01 (0.04)
Page 39
39 Diluted Diluted earnings/(losses) per share is calculated by adjusting the weighted average number of ordinary shares outstanding to assume conversion of all dilutive potential ordinary shares. The Company has three categories of dilutive potential ordinary shares: preferred shares issued by a subsidiary of the Company, restricted share units (“RSUs”) and share options. The Company had no potential dilutive ordinary shares outstanding during the six months ended 30 June 2026 as all preferred shares issued by a subsidiary of the Company, RSUs and share options were anti-dilutive, therefore the diluted earnings per share is equal to the basic earnings per share for the period presented. Accordingly, diluted earnings per share for the six months ended 30 June 2026 is the same as basic earnings per share of the respective period (six months ended 30 June 2025: same as basic loss per share of the respective period). 9 DIVIDENDS No dividend had been declared or paid by the Company during the six months ended 30 June 2026 (for the six months ended 30 June 2025: nil). 10 PROPERTY, PLANT AND EQUIPMENT Buildings and facilities (a) Property improvement Large-scale electronic equipment (b) Computers and related equipment Office equipment and furniture Transportation equipment and vehicles Other equipment Construction in progress Total RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 RMB’000 At 31 December 2025 Cost 2,121,427 556,526 5,768,789 738,926 58,364 20,201 31,591 304,194 9,600,018 Accumulated depreciation (373,586) (387,193) (2,404,026) (484,456) (41,256) (14,759) (11,486) – (3,716,762) Net book amount 1,747,841 169,333 3,364,763 254,470 17,108 5,442 20,105 304,194 5,883,256 (Unaudited) Six months ended 30 June 2026 Opening net book amount 1,747,841 169,333 3,364,763 254,470 17,108 5,442 20,105 304,194 5,883,256 Additions – 2,748 2,420,040 89,347 841 280 1,364 180,334 2,694,954 Internal transfer – 22,734 337,323 (718) 291 – – (359,630) – Disposals – (5,161) (102,828) (1,212) (72) (1,542) – (2,520) (113,335) Disposals of subsidiaries – – (1,197) (323) (1) – (3) – (1,524) Depreciation charge (38,784) (17,089) (511,012) (68,632) (3,764) (1,429) (2,225) – (642,935) Currency translation differences (933) (653) (1,346) (168) (74) (67) (706) 743 (3,204) Closing net book amount 1,708,124 171,912 5,505,743 272,764 14,329 2,684 18,535 123,121 7,817,212 At 30 June 2026 Cost 2,120,460 576,194 8,409,252 741,498 59,034 16,911 32,244 123,121 12,078,714 Accumulated depreciation (412,336) (404,282) (2,903,509) (468,734) (44,705) (14,227) (13,709) – (4,261,502) Net book amount 1,708,124 171,912 5,505,743 272,764 14,329 2,684 18,535 123,121 7,817,212 (a) As at 30 June 2026, certain buildings with carrying amount of RMB1,224,295,000 (31 December 2025: RMB1,537,922,000) were pledged as collaterals for the Group ’s borrowings (Note 15). (b) As at 30 June 2026, certain large-scale electronic equipment with carrying amount of RMB250,356,000 were pledged as collaterals for the Group ’s borrowings (Note 15).
Page 40
40 (c) During the six months ended 30 June 2026, depreciation charges were expensed off in the following categories in the interim condensed consolidated income statement: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Cost of sales 294,279 87,095 Selling expenses 2,436 7,115 Administrative expenses 60,991 95,991 Research and development expenses 285,229 314,672 642,935 504,873 11 TRADE, OTHER RECEIVABLES AND PREPAYMENTS As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Notes receivables 54,260 56,724 Provision for impairment (50,257) (49,019) 4,003 7,705 Trade receivables – Due from related parties 355,921 368,451 – Due from third parties 6,683,451 5,526,893 7,039,372 5,895,344 Less: provision for impairment (3,276,511) (3,384,897) 3,762,861 2,510,447 Other receivables – Refundable deposits 59,051 30,040 – Loans to related parties 21,769 28,287 – Payments on behalf of customers 365,326 369,348 – Others 204,475 346,757 650,621 774,432 Less: provision for impairment (306,755) (311,743) 343,866 462,689 Prepayments 638,465 895,462 Input VAT to be deducted 541,574 427,780 Total trade, other receivables and prepayments 5,290,769 4,304,083
Page 41
41 The aging analysis of the notes receivables based on date of revenue recognition is as follows: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Bank notes receivables – Up to 6 months 5,260 7,624 Commercial notes receivables – Up to 6 months – 100 – Over 1 year 49,000 49,000 54,260 56,724 The credit terms given to trade customers are determined on an individual basis with normal credit period mainly around 90 to 270 days. The aging analysis of the trade receivables based on date of revenue recognition is as follows: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Up to 6 months 2,165,910 1,896,217 6 months to 1 year 1,233,603 309,726 1 to 2 years 261,666 235,122 2 to 3 years 116,919 118,721 More than 3 years 3,261,274 3,335,558 7,039,372 5,895,344
Page 42
42 12 FINANCIAL ASSETS AT FAIR VALUE THROUGH PROFIT OR LOSS (i) Classification of financial assets at fair value through profit or loss The Group classified the following financial assets at FVPL: – Debt investments that do not qualify for measurement at either amortised cost or fair value through other comprehensive income; – Equity investments that are held for trading; and – Equity investments for which the entity has not elected to recognise fair value gains and losses through other comprehensive income. Financial assets mandatorily measured at FVPL include the following: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Non-current assets Debt investments (a) – Unlisted entities 4,342,457 3,893,508 – Investments in bonds 1,304,849 1,330,278 – Fund 1,669,758 1,001,587 Equity investments (b) – Listed entities 5,032,628 865,737 – Unlisted entities 430,956 419,151 12,780,648 7,510,261 Current assets Wealth management products (c) 741,776 571,969 13,522,424 8,082,230 (a) Debt investments The movement of the debt investments during the six months ended 30 June 2026 is as follows: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) At beginning of the period 6,225,373 5,214,360 Additions 1,316,403 194,371 Disposals (32,838) (2,261) Disposal of subsidiaries (73,565) – Transfer to equity investments (431,862) – Fair value changes 402,832 142,830 Currency translation differences (89,279) (16,425) At end of the period 7,317,064 5,532,875
Page 43
43 The Group made investments in various industry companies in the form of convertible redeemable preferred shares, ordinary shares with preferential rights and convertible loans. The Group has the right to require and demand the investees to redeem all of the investments held by the Group at guaranteed predetermined amount upon redemption events which are out of control of the investees. Hence these investments are accounted for as debt instruments and are measured as financial assets at fair value through profit or loss. In addition, the Group also made investments in certain investment funds as a limited partner, these investments were included in debt investments, depending on the investment contract terms. (b) Equity investments The movement of the equity investments during the six months ended 30 June 2026 is as follows: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) At beginning of the period 1,284,888 1,149,136 Additions 70,482 50,579 Disposals (395,711) (338,783) Transfer from debt investments (a) 431,862 – Fair value changes 4,168,315 (284,812) Currency translation differences (96,252) (1,746) At end of the period 5,463,584 574,374 The fair values of the listed securities are determined based on the closing price quoted in active market or that with the adjustment of lack of marketability discount. The fair values of the unlisted securities are measured using a valuation technique with unobservable inputs. (c) Wealth management products The wealth management products were non-principal protected with maturity of less than 1 year. The movement of the wealth management products during the six months ended 30 June 2026 is as follows: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) At beginning of the period 571,969 701,087 Additions 6,873,206 6,788,000 Redemptions (6,719,643) (5,655,073) Disposal of subsidiary – (50,000) Fair value changes 16,937 11,875 Currency translation differences (693) – At end of the period 741,776 1,795,889
Page 44
44 (d) Amounts recognised in the interim condensed consolidated income statement During the six months ended 30 June 2026, the following gains/(losses) were recognised in the interim condensed consolidated income statement: Six months ended 30 June 2026 2025 RMB’000 RMB’000 (Unaudited) (Unaudited) Fair value gains/(losses) on investments in: – Debt investments 402,832 142,830 – Equity investments 4,168,315 (284,812) – Derivative financial instruments – (25,157) – Wealth management products 16,937 11,875 4,588,084 (155,264) 13 DERIVATIVE FINANCIAL INSTRUMENTS As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Assets: Foreign currency swap contract – 3,097
Page 45
45 14 TRADE AND OTHER PAYABLES As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Notes payables – 29 Trade payables – Third parties 1,207,607 679,482 – Related parties 11,732 2,522 Other payables – Third parties 542,912 624,129 – Related parties 66,574 52,181 Payables on purchase of property, plant and equipment and intangible assets 541,912 357,103 Accrued taxes other than income tax 10,279 60,923 Staff salaries and welfare payables 195,483 287,596 VAT payables related to contract liabilities 24,008 21,482 Accrued warranty expenses 1,000 4,911 2,601,507 2,090,358 Aging analysis of the notes and trade payables based on purchase date at the end of 30 June 2026 are as follows: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Up to 6 months 1,085,357 540,971 6 months to 1 year 9,877 13,612 1 to 2 years 24,881 27,098 More than 2 years 99,224 100,352 1,219,339 682,033
Page 46
46 15 BORROWINGS As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) Non-Current Bank borrowing – secured and guaranteed 222,271 – Bank borrowing – secured 3,481,144 3,611,939 Bank borrowing – guaranteed 823,120 1,156,745 Bank borrowing – unsecured and unguaranteed 39,400 39,400 Less: current portion of non-current borrowings (751,807) (1,097,749) 3,814,128 3,710,335 Current Short-term bank borrowing – guaranteed 942,073 330,500 Short-term borrowing – unsecured and unguaranteed 565,043 582,266 Current portion of non-current borrowings 751,807 1,097,749 Interest payable 3,393 3,735 2,262,316 2,014,250 6,076,444 5,724,585 Secured and guaranteed bank borrowings As at 30 June 2026, the Group had a non-current bank borrowing with carrying amount of RMB222,271,000 which was pledged by large-scale electronic equipment with a carrying amount of RMB250,356,000 (Note 10(b)). This non-current bank borrowing was also guaranteed by Shanghai SenseTime. Secured bank borrowings As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB1,836,004,000 (31 December 2025: RMB1,932,039,000) which was pledged by equity interest of Shanghai Yuqin Information Technology Co., Ltd. ( “Shanghai Yuqin ”) and joint liability guarantee from Shanghai Yuqin and Shanghai SenseTime. In addition, certain buildings (Note 10(a)) with a carrying amount of RMB922,649,000 and land use right with a carrying amount of RMB59,781,000 (31 December 2025: buildings with a carrying amount of RMB953,089,000 and land use right with a carrying amount of RMB60,455,000) respectively were also pledged as collaterals for this bank borrowing. As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB1,645,140,000 which was pledged by certain building with a carrying amount of RMB301,646,000 (Note 10(a)) in property, plant and equipment and RMB2,507,845,000 in investment properties (31 December 2025: bank borrowing with carrying amount of RMB1,650,140,000 which was pledged by certain buildings with a carrying amount of RMB584,833,000 in property, plant and equipment and RMB2,512,134,000 in investment properties) respectively as a collateral for the Group ’s borrowings. Guaranteed bank borrowings As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB139,000,000 for public rental housing, which was guaranteed by a state-owned property developer before the property registration is ready. After that, this borrowing will be guaranteed by Shanghai SenseTime and be pledged by the public rental housing itself as a collateral (31 December 2025: borrowings with carrying amount of RMB140,000,000).
Page 47
47 As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB409,380,000 which were guaranteed by SenseTime Group Limited (31 December 2025: bank borrowings with carrying amount of RMB578,725,000). As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB109,100,000 which were guaranteed by Beijing SenseTime (31 December 2025: bank borrowings with carrying amount of RMB109,200,000). As at 30 June 2026, the Group had non-current bank borrowings with carrying amount of RMB165,640,000 which were guaranteed by Shanghai SenseTime (31 December 2025: bank borrowings with carrying amount of RMB328,820,000). As at 30 June 2026, the Group had current bank borrowings with carrying amount of RMB931,573,000 which were guaranteed by SenseTime Group Limited (31 December 2025: RMB320,000,000 which were guaranteed by SenseTime Group Limited). As at 30 June 2026, the Group had current bank borrowings with carrying amount of RMB10,500,000 which were guaranteed by Shanghai SenseTime (31 December 2025: RMB10,500,000 which were guaranteed by Shanghai SenseTime). The exposure of the Group ’s borrowings to interest rate changes and the contractual repricing dates of the borrowings at the end of the reporting period are as follows: As at 30 June As at 31 December 2026 2025 RMB’000 RMB’000 (Unaudited) 6 months or less 870,123 1,277,936 Between 6 and 12 months 1,388,800 732,579 Between 1 and 2 years 831,483 570,970 Between 2 and 5 years 1,335,255 1,460,725 Over 5 years 1,647,390 1,678,640 6,073,051 5,720,850 The aggregate principal amounts of bank borrowing and applicable interest rates are as follows: As at 30 June 2026 As at 31 December 2025 Amount Interest rate Amount Interest rate RMB’000 Per annum RMB’000 Per annum (Unaudited) (Unaudited) RMB bank borrowings 6,073,051 1.48%~2.80% 5,720,850 1.48%~3.80% As at 30 June 2026, the carrying amounts of bank borrowings approximated their fair values since the interest payable on those borrowings is either close to current market rates or the borrowings are of a short-term nature. As at 30 June 2026 the weighted average effective interest rate for borrowings was 2.40% (31 December 2025: 2.52%). 16 EVENTS AFTER THE BALANCE SHEET DATE There were no material subsequent events during the period from 1 July 2026 to the approval date of the interim financial information by the Board.
Page 48
48 OTHER INFORMATION Purchase, Sale or Redemption of the Company ’s Listed Securities During the Reporting Period, neither the Company nor any of its subsidiaries or consolidated affiliated entities purchased, sold or redeemed any of the Company ’s securities listed on the Stock Exchange (including any sale of treasury shares). Compliance with the Corporate Governance Code The Company is committed to achieving high standards of corporate governance with a view to safeguarding the interests of the Shareholders. The principles of the Company ’s corporate governance are to promote effective internal control measures, to enhance transparency of the work of the Board, and to strengthen accountability to all the Shareholders. During the Reporting Period, the Company has complied with the corporate governance requirements under the Corporate Governance Code set out in Appendix C1 to the Listing Rules save for code provision C.2.1 of Part 2 of the Corporate Governance Code as discussed below. Pursuant to code provision C.2.1 of Part 2 of the Corporate Governance Code, companies listed on the Stock Exchange are expected to comply with, but may choose to deviate from the requirement that the responsibilities between the chairman and the chief executive officer should be segregated and should not be performed by the same individual. The Company does not have a separate chairman and chief executive officer and Dr. Xu Li currently performs these two roles. The Board believes that vesting the roles of both executive chairman of the Board and chief executive officer in the same person has the benefit of ensuring consistent leadership within the Group and enables more effective and efficient overall strategic planning for the Group. The Board considers that the balance of power and authority for the present arrangement will not be impaired and this structure will enable the Company to make and implement decisions promptly and effectively. The Board will continue to review and consider splitting the roles of executive chairman of the Board and the chief executive officer of the Company at a time when it is appropriate by taking into account the circumstances of the Group as a whole.
Page 49
49 Interim Dividend The Board has resolved not to declare an interim dividend for the six months ended June 30, 2026 (June 30, 2025: Nil). Directors ’ Securities Transactions The Board has adopted the Model Code as the code of conduct regulating Directors ’ dealings in securities of the Company. In response to specific enquiries made by the Board, all Directors confirmed that they have complied with the provisions of the Model Code during the Reporting Period. Audit Committee The Audit Committee (comprising two independent non-executive Directors, Mr. Lyn Frank Yee Chon (being the chairman of the Audit Committee) and Mr. Chiu Duncan; and the non-executive Director, Ms. Fan Yuanyuan) has reviewed the unaudited interim condensed consolidated financial information of the Group for the Reporting Period. The Audit Committee has also discussed matters in relation to the accounting policies and practices adopted by the Company, risk management, internal control and financial reporting with senior management members and the Auditor. Publication of the Interim Results and Interim Report This interim results announcement has been published on the website of the Stock Exchange at www.hkexnews.hk and the website of the Company at www.sensetime.com. The interim report of the Company containing all the information required by the Listing Rules will be published on the websites of the Stock Exchange and the Company and will be despatched to the Shareholders (if requested) in due course.
Page 50
50 DEFINITIONS In this announcement, unless the context otherwise requires, the following expression shall have the meanings set out below: “Audit Committee ” the audit committee of the Company “Auditor ” PricewaterhouseCoopers, the independent auditor of the Company “Board ” the board of Directors of the Company “Class A Share(s) ” class A ordinary shares of the share capital of the Company with a par value of US$0.000000025 each “Class B Share(s) ” class B ordinary shares of the share capital of the Company with a par value of US$0.000000025 each “Company ” SenseTime Group Inc. ( 商湯集團股份有限公司 ), an exempted company incorporated under the laws of Cayman Islands with limited liability, the issued Class B Shares of which are listed on the Main Board of the Stock Exchange (stock codes: 0020 (HKD Counter) and 80020 (RMB Counter)) “Director(s) ” director(s) of the Company “Dr. Wang Xiaogang ” Dr. Wang Xiaogang (the Company ’s co-founder, executive Director, chief scientist and a WVR Beneficiary “Dr. Xu Li ” Dr. Xu Li (ͭ), the Company ’s co-founder, executive Chairman of the Board, executive Director, chief executive officer and a WVR Beneficiary “Group ”, “SenseTime ” or “we” the Company and its subsidiaries and consolidated affiliated entities “HKD” or “HK$” Hong Kong Dollars, the lawful currency of Hong Kong “Hong Kong ” Hong Kong Special Administrative Region of the PRC “IFRS” International Financial Reporting Standards “Listing Rules ” the Rules Governing the Listing of Securities on the Stock Exchange “Model Code ” the Model Code for Securities Transactions by Directors of Listed Issuers, as set out in Appendix C3 of the Listing Rules “PRC” or “China” the People ’s Republic of China “Reporting Period ” the six months ended June 30, 2026
Page 51
51 “RMB” Renminbi, the lawful currency of the PRC “SAR” Saudi Riyal, the lawful currency of Saudi Arabia “SGD” Singapore Dollars, the lawful currency of Singapore “Share(s) ” the Class A Shares and Class B Shares in the share capital of our Company, as the context so requires “Shareholder(s) ” the holder(s) of Share(s) “Stock Exchange ” The Stock Exchange of Hong Kong Limited “USD” or “US$” United States Dollars, the lawful currency of the United States “WVR Beneficiary(ies) ” has the meaning ascribed to it under the Listing Rules and unless the context otherwise requires, refers to Dr. Xu Li and Dr. Wang Xiaogang, being holders of the Class A Shares, entitling each to weighted voting rights By order of the Board SenseTime Group Inc. 商湯集團股份有限公司 Dr. Xu Li Executive Chairman Chief Executive Officer Hong Kong, August 26, 2026 As at the date of this announcement, the executive Directors are Dr. Xu Li, Dr. Wang Xiaogang, Dr. Lin Dahua, Mr. Yang Fan and Mr. Wang Zheng; the non-executive Director is Ms. Fan Yuanyuan; and the independent non-executive Directors are Prof. Xue Lan, Mr. Lyn Frank Yee Chon and Mr. Chiu Duncan. This announcement contains certain forward-looking statements. These forward-looking statements are based on information currently available to the Group or the current belief, expectations and assumptions of the Board. These forward-looking statements are subject to risks, uncertainties and other factors beyond the Company ’s control which may cause actual results or performance to differ materially from those expressed or implied in such forward-looking statements. In light of the risks and uncertainties, the inclusion of forward-looking statements in this announcement should not be regarded as representations by the Board or the Company that the plans and objectives will be achieved, and shareholders and investors of the Company should not place undue reliance on such statements.