商汤(00020) - 2022 - 年度业绩

Annual Results Announcement Summary Financial Highlights In fiscal year 2022, SenseTime Group's revenue decreased by 19.0% to RMB 3,808.5 million, with gross profit down 22.4% to RMB 2,542.3 million; annual loss significantly narrowed by 64.5% to RMB 6,093.0 million, but adjusted EBITDA and adjusted net loss both expanded 2022 Fiscal Year Key Financial Data (thousand RMB) | Indicator | 2022 Amount | 2021 Amount | YoY Change | | :------------- | :--------- | :--------- | :------- | | Revenue | 3,808,510 | 4,700,263 | -19.0% | | Gross Profit | 2,542,267 | 3,277,638 | -22.4% | | Annual Loss | (6,092,990)| (17,177,050)| -64.5% | | Adjusted EBITDA | (4,213,444)| (878,970) | 379.4% | | Adjusted Net Loss | (4,736,199)| (1,418,348)| 233.9% | Operating Data In 2022, the number of Smart Business customers decreased by 22.2%, Smart Living customers increased by 23.3%, and Smart Auto customers remained flat, while the cumulative number of Smart City service cities grew by 15.7% 2022 Fiscal Year Operating Data | Indicator | 2022 | 2021 | YoY Change | | :----------------- | :----- | :----- | :------- | | Smart Business Customers | 717 | 922 | -22.2% | | Smart Living Customers | 317 | 257 | 23.3% | | Smart Auto Customers | 34 | 34 | 0.0% | | Cumulative Smart City Service Cities | 162 | 140 | 15.7% | Business Review and Outlook Core Technology Breakthroughs In 2022, SenseTime achieved significant progress in Artificial General Intelligence (AGI), continuously investing in high-performance computing infrastructure and large model R&D, and making breakthroughs in five key technical areas: computer vision, AIGC, natural language processing, speech recognition and synthesis, decision intelligence, and AI chips - The company actively invests in the AGI field, focusing on large computing power and large model R&D, maintaining a leading position in high-performance computing systems and large model training capabilities[4] - In computer vision, the company successfully developed the world's largest general vision model with 32 billion parameters, widely applied in autonomous driving, industrial quality inspection, and medical imaging[5] - In AI Content Generation (AIGC), multiple text-to-image large models were trained, supporting 6K high-definition image generation and demonstrating unique advantages in understanding traditional Chinese culture[5] Large-scale AI Infrastructure and High-Performance Computing SenseTime's SenseCore AI large-scale infrastructure continues to make breakthroughs, achieving 5.0 exaFLOPS computing power output based on 27,000 GPUs, supporting simultaneous training of up to 20 ultra-large models with hundreds of billions of parameters, providing reliable computing power for large model R&D - SenseCore AI large-scale infrastructure achieves 5.0 exaFLOPS computing power output, supporting simultaneous training of up to 20 ultra-large models with hundreds of billions of parameters[4] Computer Vision The company achieved significant results in computer vision, developing the world's largest general vision model with 32 billion parameters, applied in autonomous driving, industrial quality inspection, and medical imaging, and winning the 2022 Waymo Challenge - Successfully developed the world's largest general vision model with 32 billion parameters, applied in autonomous driving, industrial quality inspection, and medical imaging[5] - BEVFormer++ perception algorithm based on large vision models won the championship in the main track of the 2022 Waymo Challenge[5] AI Content Generation (AIGC) The company trained multiple text-to-image large models in AIGC, demonstrating excellent text-to-image understanding, image quality, and inference speed, with a unique advantage in understanding traditional Chinese culture, supporting 6K high-definition image generation - Achieved training of multiple text-to-image large models, demonstrating excellent text-to-image understanding, image quality, and inference speed[5] - Possesses unique advantages in understanding traditional Chinese culture, supporting 6K high-definition image generation with more delicate light and shadow processing[5] Natural Language Processing (NLP) The company achieved significant breakthroughs in natural language processing, successfully developing a pre-trained large model with hundreds of billions of parameters, significantly improving text generation and human-computer interaction performance, with a market launch planned for mid-2023 - Successfully developed a pre-trained large model with hundreds of billions of parameters, significantly improving text generation and human-computer interaction performance[6] - This model and its capabilities are planned to be launched to the market by mid-2023[6] Speech Recognition and Synthesis The company achieved technological leadership in speech recognition and synthesis, with its speech recognition model achieving high accuracy and low-latency speech-to-text capabilities, and its speech synthesis model generating voices with different tones and timbres, supporting digital humans and smart cockpits - Speech recognition model achieves high accuracy and low-latency speech-to-text capabilities[6] - Speech synthesis model can generate voices with different tones and timbres according to user needs, supporting digital humans and smart cockpits[6] Reinforcement Learning and Decision Intelligence The company made significant progress in reinforcement learning, successfully developing efficient decision intelligence models applied in robot control, game AI, and autonomous driving decision systems - Successfully developed efficient decision intelligence models, applied in robot control, game AI, and autonomous driving decision systems[6] AI Chips and Hardware Acceleration The company achieved important results in AI hardware, successfully developing multiple high-performance, low-power AI inference chips and AI ISP chips, providing hardware foundation for AI business, and collaborating with semiconductor manufacturers to optimize AI model operating efficiency - Successfully developed multiple high-performance, low-power AI inference chips and AI ISP chips, demonstrating powerful computing capabilities in data centers and edge computing scenarios[6] - Collaborates with semiconductor manufacturers to continuously optimize the operating efficiency of AI models on hardware[6] SenseCore AI Infrastructure and R&D Investment SenseTime's SenseCore AI large-scale infrastructure continues to expand, deploying 27,000 GPUs to achieve 5.0 exaFLOPS computing power, providing services to industry partners in an AIaaS model to support large model production, with R&D investment of RMB 3,795.8 million in 2022, a 24.0% increase year-on-year, and significantly improved R&D efficiency - SenseCore AI large-scale infrastructure has deployed 27,000 GPUs, achieving 5.0 exaFLOPS computing power output, providing services to industry partners in an AIaaS model[9] - In 2022, R&D investment excluding share-based compensation expenses was RMB 3,795.8 million, a 24.0% increase year-on-year[8][14] - The number of models developed per person per year increased by 90% to 9.35, and the cumulative number of commercial models increased by 93% to 67,000[14] SenseCore AIaaS Capabilities SenseCore provides Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Model as a Service (MaaS), offering high-performance computing resources, user-friendly development tools, a rich library of pre-trained models, flexible deployment, professional support, and high-quality data processing capabilities - SenseCore provides IaaS, PaaS, and MaaS capabilities, supporting large-scale distributed training at the thousand-card level, with a thousand-card parallel efficiency of 91.5%[9][10] - Open-source platforms like OpenMMLab and four others, containing over 2,500 pre-trained models and over 400 algorithms, attracting global developers[11] - Supports one-click cross-platform deployment of AI models to cloud, edge devices, or hybrid environments, with adaptation and optimization for 30 chips and over 100 platforms[12] R&D Investment and Efficiency In 2022, the company's R&D expenditure excluding share-based compensation expenses was RMB 3,795.8 million, a 24.0% increase year-on-year, with an R&D team of 3,466 people, accounting for 68% of total employees, and significant improvements in per capita R&D efficiency and commercial model count - In 2022, R&D expenditure excluding share-based compensation expenses was RMB 3,795.8 million, a 24.0% increase year-on-year[14] - The R&D team comprises 3,466 people, accounting for 68% of total employees[14] - The number of models developed per person per year increased by 90% to 9.35, and the cumulative number of commercial models increased by 93% to 67,000[14] Performance of Business Segments In 2022, SenseTime comprehensively upgraded its product matrix across Smart Living, Smart Auto, Smart Business, and Smart City business lines, leveraging SenseCore AI large-scale infrastructure and large model capabilities, achieving business structure transformation, with significant revenue growth in Smart Living and Smart Auto, and revenue decline in Smart Business and Smart City - Revenue from Smart Living and Smart Auto businesses significantly increased, while revenue from Smart Business and Smart City businesses decreased[7][15] - Smart Living revenue increased by 129.9%, and Smart Auto revenue increased by 58.9%[7][8] - Smart Business revenue decreased by 25.2%, and Smart City revenue decreased by 48.8%[8] Smart Living The Smart Living segment's revenue in 2022 increased by 129.9% to RMB 955.1 million, reaching a new high and increasing its proportion of group revenue to 25.1%, with commercialization breakthroughs in AIGC, AI sensors, AI ISP chips, and smart healthcare, and digital human products becoming market leaders Smart Living Business Performance | Indicator | 2022 Amount (million RMB) | 2021 Amount (million RMB) | YoY Growth | | :--------- | :------------------------ | :------------------------ | :------- | | Revenue | 955.1 | 415.4 | 129.9% | | Revenue Share | 25.1% | 8.8% | Increase | | Customer Count | 317 | 257 | 23.3% | | Revenue Per Customer | - | - | 86.4% | - AIGC multi-modal full-stack capabilities empower over 200 APP applications, with offline scenario coverage increasing to 15 million square meters[17] - Digital human products ranked first in technical evaluations for tenders and were recognized as a market leader by Frost & Sullivan[17] - AI sensors and AI ISP chips achieved technological and commercial breakthroughs, with AI-ISP chip power consumption 50% lower than similar products, expected to be mass-produced in 2023[18] - Smart healthcare business revenue achieved multi-fold growth, with SenseCare Smart Diagnosis and Treatment Platform obtaining 22 domestic and international medical device certifications and deployed in over a hundred hospitals[19] - Consumer AI product "SenseRobot Go" chess robot quickly topped Tmall and JD's smart robot category in transaction volume since its launch[20] Smart Auto The Smart Auto segment's revenue in 2022 increased by 58.9% to RMB 292.7 million, increasing its proportion of group revenue to 7.7%, with over 500,000 units delivered, over 8 million new designated vehicle models, multiple industry-first innovative functions, successful mass production of L2+ and L2++, significantly improved per-vehicle value, and the launch of a vehicle-road-cloud integrated product matrix Smart Auto Business Performance | Indicator | 2022 Amount (million RMB) | 2021 Amount (million RMB) | YoY Growth | | :------- | :------------------------ | :------------------------ | :------- | | Revenue | 292.7 | 184.2 | 58.9% | | Revenue Share | 7.7% | 3.9% | Increase | - Over 500,000 units mass-produced and delivered in 2022, with over 8 million new designated vehicle models, covering over 30 car manufacturers and 80 models[22] - Achieved mass production of multiple smart cockpit and smart driving functions, including high-speed navigation and urban navigation for L2+ and L2++ smart driving solutions, significantly increasing per-vehicle value[23] - Launched AI cloud platform – SenseAuto Vehicle-Road Collaboration Platform, integrating a comprehensive "vehicle-road-cloud" product matrix, improving traffic efficiency by over 20%[24] Smart Business The Smart Business segment's revenue in 2022 decreased by 25.2% to RMB 1,464.3 million, reducing its proportion of group revenue to 38.4%, primarily due to the impact of the pandemic and macro economy, with the company focusing on AI cloud business transformation, rapid growth in SenseCore external service revenue, and assisting leading clients in digital energy and industrial sectors with digital transformation Smart Business Business Performance | Indicator | 2022 Amount (million RMB) | 2021 Amount (million RMB) | YoY Change | | :------- | :------------------------ | :------------------------ | :------- | | Revenue | 1,464.3 | 1,957.9 | -25.2% | | Revenue Share | 38.4% | 41.7% | Decrease | | Customer Count | 717 | 922 | -22.2% | | Revenue Per Customer | - | - | -3.8% | - SenseCore external service revenue rapidly broke through, accounting for over 20% of Smart Business's total revenue, expected to significantly increase in 2023[26] - Assists leading enterprises in digital energy and industrial sectors with digital transformation, providing automated inspection services for China's largest power grid company, improving efficiency by 5-6 times[27] Smart City The Smart City segment's revenue in 2022 decreased by 48.8% to RMB 1,096.4 million, reducing its proportion of group revenue to 28.8%, primarily due to the impact of the pandemic, with the company actively adjusting its business, focusing on serving key clients, optimizing revenue quality, and continuously expanding the deployment of SenseFoundry City Ark Smart City Business Performance | Indicator | 2022 Amount (million RMB) | 2021 Amount (million RMB) | YoY Change | | :------- | :------------------------ | :------------------------ | :------- | | Revenue | 1,096.4 | 2,142.8 | -48.8% | | Revenue Share | 28.8% | 45.6% | Decrease | - The company was recognized in the Leader category of IDC MarketScape China City Intelligent Computing Platform Vendor Assessment, ranking first in strategic capabilities[28] - SenseFoundry City Ark expanded its coverage to 22 new cities, with a cumulative deployment in 162 cities, equipped with approximately 35,000 AI models[28] Environmental, Social and Governance (ESG) Environmental Protection SenseTime actively practices green development, setting carbon peaking by 2025, operational carbon neutrality by 2030, and net-zero emissions by 2050 targets, optimizing energy consumption through AIDC smart management systems, with AI large-scale infrastructure power consumption over 10% lower than the industry average, reducing approximately 35,600 tons of CO2 emissions annually - Established carbon neutrality targets: carbon peaking by 2025, operational carbon neutrality by 2030, and net-zero emissions by 2050[30] - AI large-scale infrastructure power consumption is over 10% lower than the average level of China's medium and large data center industry, saving approximately 45 million kWh of power and reducing approximately 35,600 tons of CO2 emissions annually[30] - Plans to optimize energy consumption per unit of computing power by 30% by 2025 and launch solutions like smart virtual power plants to empower carbon neutrality in the energy industry[30] AI Governance and Information Security Protection SenseTime highly values AI governance, publishing the "AI Governance White Paper on Balanced Development" and proposing a vision for "responsible and assessable" AI, embedding ethical principles throughout the product lifecycle, and receiving multiple trusted AI practice awards, while SenseAuto obtained ISO/SAE 21434:2021 automotive information security process certification, becoming one of the few companies in the industry to obtain all three certifications - Published the "AI Governance White Paper on Balanced Development," proposing a vision for developing "responsible and assessable" AI[31] - AI training platform (AutoLink) received the "Trusted AI Platform Application Benchmark Case" award, and its ethical governance system and algorithm evaluation tools won the "Trusted AI Practice Excellent Case" award[31] - SenseAuto obtained ISO/SAE 21434:2021 automotive information security process certification, becoming one of the few companies in the industry to obtain ISO 26262, ASPICE CL2, and ISO/SAE 21434 certifications[32] Strategic Outlook Core Strategic Directions SenseTime will pursue Artificial General Intelligence (AGI) as its core strategic goal, aiming for significant breakthroughs in AGI technology within the next few years, focusing on large computing power, multi-modal large models, AI-specific hardware and chips, and open research collaboration, while establishing and executing the "AI for All" development goal - Artificial General Intelligence (AGI) is the core strategic goal, aiming for significant breakthroughs in AGI technology within the next few years[33] - Continuously expand SenseCore AI large-scale infrastructure capabilities, invest in high-performance computing infrastructure, and strengthen system scalability[34] - Multi-modal large model R&D is a strategic investment direction, building capabilities in model architecture innovation, dataset construction, knowledge fusion and reasoning, application scenario expansion, and interpretability and security[35] Large Computing Power SenseTime will use SenseCore AI large-scale infrastructure as the core platform for large model production, continuously investing in high-performance computing infrastructure, deeply researching parallel computing and distributed system technologies, strengthening system scalability, shortening model training time, and accelerating iteration cycles - SenseCore AI large-scale infrastructure serves as the core platform, with continuous investment in high-performance computing infrastructure, including high-speed networks, large-scale data storage, and powerful computing nodes[34] - Deeply research parallel computing and distributed system technologies, strengthening system scalability, shortening model training time, and accelerating iteration cycles[34] Multi-modal Large Models SenseTime will strategically invest in multi-modal large model R&D, aiming to process various data types and enhance data understanding and generation capabilities, focusing on model architecture innovation, large-scale multi-modal dataset construction, knowledge fusion and reasoning, application scenario expansion, and model interpretability and security - Optimize model architectures that combine various data types, build large-scale multi-modal datasets, and enhance the model's ability to process multi-modal data[35] - Optimize multi-modal knowledge representation and fusion methods, actively collaborate with ecosystem partners to advance multi-modal large model applications in smart assistants, autonomous driving, and medical diagnosis[35] AI-specific Hardware and Chips and Open Research Collaboration SenseTime will collaborate with leading global chip design and manufacturing companies to promote the development of AI hardware and chips, achieving stronger AI computing capabilities and higher energy efficiency ratios, while actively cooperating with global research institutions, universities, and enterprises to jointly explore new ways to optimize computing power and advance AI computing development - Collaborate with leading global chip design and manufacturing companies to promote the development of AI hardware and chips, achieving stronger AI computing capabilities and higher energy efficiency ratios[36] - Actively cooperate with global research institutions, universities, and enterprises to jointly explore new ways to optimize computing power and advance AI computing development[36] AI for All Development Goal SenseTime is committed to making AI technology benefit all humanity by reducing AI technology costs and barriers, developing low-cost, high-performance software and hardware solutions, and providing user-friendly APIs and development tools, while promoting AI technology development through open source and collaboration, focusing on diversity and inclusion, and collaborating with governments and industry partners to formulate AI policies and regulatory frameworks to ensure compliant and sustainable AI technology development - Committed to reducing AI technology costs and barriers, developing low-cost, high-performance software and hardware solutions, and providing user-friendly APIs and development tools[36] - Through open-source projects, shared resources, and collaborative research, jointly advance AI technology development with global AI researchers and developers[36] - Collaborate with governments, industry partners, and academic organizations to jointly discuss and formulate AI technology policies and regulatory frameworks, ensuring compliance and sustainable development[37] Business Segment Upgrade Strategy SenseTime will leverage SenseCore large-scale infrastructure and large model capabilities to comprehensively upgrade its Smart Living, Smart Auto, Smart Business, and Smart City business segments, promoting AI applications through continuous advancement of SenseCore external services, deepening AIGC multi-modal capabilities, innovating smart auto functions, focusing on AI cloud business transformation, and optimizing Smart City cash flow, with commercialization as the key driver - Continuously advance SenseCore external service business, empowering large model R&D enterprises and research institutions, as well as key industry enterprises in AIGC, autonomous driving, and AI for Science[38] - Smart Living will deepen AIGC multi-modal full-stack capabilities, promote digital human product applications, and continuously innovate AI SDK, AI sensor, and AI ISP chip product portfolios[38] - Smart Auto will continuously innovate and mass-produce smart driving and smart cockpit functions, deepen the vehicle-road-cloud integrated product matrix, and maintain industry leadership[38] - Smart City and Smart Business will focus on AI cloud business transformation, increasing AIaaS revenue, assisting digital energy and industrial clients with digital transformation, and optimizing Smart City business cash flow[38] Management Discussion and Analysis Revenue Analysis In 2022, revenue decreased by 19.0% year-on-year to RMB 3,808.5 million, primarily due to declines in Smart City and Smart Business revenue, partially offset by growth in Smart Living and Smart Auto revenue, with Smart Living revenue growing by 129.9%, Smart Auto revenue by 58.9%, and Smart Business and Smart City revenues decreasing by 25.2% and 48.8% respectively Revenue Breakdown by Source (million RMB) | Business Segment | 2022 Amount | 2022 Share | 2021 Amount | 2021 Share | YoY Change | | :------- | :--------- | :--------- | :--------- | :--------- | :------- | | Smart Business | 1,464.3 | 38.4% | 1,957.9 | 41.7% | -25.2% | | Smart City | 1,096.4 | 28.8% | 2,142.8 | 45.6% | -48.8% | | Smart Living | 955.1 | 25.1% | 415.4 | 8.8% | 129.9% | | Smart Auto | 292.7 | 7.7% | 184.2 | 3.9% | 58.9% | | Total | 3,808.5 | 100.0% | 4,700.3 | 100.0% | -19.0% | - Smart Business revenue decline was primarily due to the impact of the COVID-19 pandemic and lockdown measures, leading to delayed customer AI spending and hindered on-site product deployment[42] - Smart Living revenue growth was mainly driven by increased R&D revenue from AI ISP chips, growth in healthcare customer revenue, and the successful international expansion of the SenseMARS platform[43] Cost of Sales, Gross Profit and Gross Margin In 2022, cost of sales decreased by 11.0% year-on-year to RMB 1,266.2 million, primarily due to lower hardware costs and subcontracting service fees, with gross profit decreasing by 22.4% to RMB 2,542.3 million, and gross margin declining from 69.7% in 2021 to 66.8% in 2022 Cost of Sales Breakdown by Nature (million RMB) | Nature | 2022 Amount | 2022 Share | 2021 Amount | 2021 Share | | :----------------- | :--------- | :--------- | :--------- | :--------- | | Hardware Costs and Subcontracting Service Fees | 1,118.9 | 88.4% | 1,265.4 | 89.0% | | Server Operations and Cloud Service Fees | 13.8 | 1.1% | 1.5 | 0.1% | | Employee Benefit Expenses | 72.1 | 5.7% | 80.1 | 5.6% | | Depreciation and Amortization | 39.4 | 3.1% | – | – | | Other Expenses | 22.0 | 1.7% | 75.6 | 5.3% | | Total | 1,266.2 | 100.0% | 1,422.6 | 100.0% | - The decrease in cost of sales is consistent with the decline in revenue, reflecting a reduction in hardware costs and subcontracting service fees[45] - Gross margin decreased from 69.7% in 2021 to 66.8% in 2022, primarily due to reduced revenue[46] Operating Expenses Analysis In 2022, R&D expenses increased by 11.1% to RMB 4,014.3 million, mainly due to increased AIDC infrastructure investment and professional service fees; selling expenses increased by 32.2% to RMB 900.8 million, due to an expanded sales and marketing team and increased brand promotion activities; administrative expenses decreased by 31.8% to RMB 1,567.7 million, primarily due to reduced share-based compensation expenses R&D Expenses In 2022, R&D expenses increased by 11.1% year-on-year to RMB 4,014.3 million, primarily due to increased depreciation and amortization from AIDC infrastructure investment, and increased professional service and consulting fees related to AI chip R&D work - R&D expenses increased by 11.1% to RMB 4,014.3 million[47] - Primarily due to increased depreciation and amortization from AIDC infrastructure investment, and increased professional service and consulting fees related to AI chip R&D work[47] Selling Expenses In 2022, selling expenses increased by 32.2% year-on-year to RMB 900.8 million, primarily due to increased employee compensation expenses from an expanded sales and marketing team, and increased domestic and international marketing and promotional activities - Selling expenses increased by 32.2% to RMB 900.8 million[48] - Primarily due to increased employee compensation expenses from an expanded sales and marketing team, and increased marketing, conference, and travel expenses[48] Administrative Expenses In 2022, administrative expenses decreased by 31.8% year-on-year to RMB 1,567.7 million, primarily due to reduced share-based compensation expenses for the management team and administrative staff - Administrative expenses decreased by 31.8% to RMB 1,567.7 million[49] - Primarily due to reduced share-based compensation expenses for the management team and administrative staff[49] Other Financial Items In 2022, net impairment loss on financial and contract assets significantly increased by 222.2% to RMB 1,667.7 million, mainly due to increased impairment provisions for trade receivables; other income decreased by 11.3% to RMB 447.3 million, primarily due to reduced government grants; net other losses were RMB 1,367.3 million, mainly including foreign exchange losses and fair value losses on financial assets; net finance income increased to RMB 205.1 million; and annual loss narrowed to RMB 6,093.0 million Net Impairment Loss on Financial and Contract Assets In 2022, net impairment loss on financial and contract assets increased by 222.2% to RMB 1,667.7 million from RMB 517.6 million in 2021, primarily due to increased impairment provisions for trade receivables, with the total amount of trade receivables also increasing - Net impairment loss on financial and contract assets increased by 222.2% to RMB 1,667.7 million[50] - The percentage of impairment provisions for trade receivables increased from 16.2% to 33.1%, primarily due to COVID-19 related disruptions and customer budget constraints[50] Other Income In 2022, other income decreased by 11.3% to RMB 447.3 million from RMB 504.3 million in 2021, primarily due to reduced government grants - Other income decreased by 11.3% to RMB 447.3 million, primarily due to reduced government grants[51] Net Other Losses In 2022, net other losses were RMB 1,367.3 million, compared to RMB 400.0 million in 2021, mainly comprising net foreign exchange losses of RMB 453.4 million and fair value losses on financial assets measured at fair value through profit or loss of RMB 885.1 million - In 2022, net other losses were RMB 1,367.3 million, mainly comprising net foreign exchange losses of RMB 453.4 million and fair value losses on financial assets measured at fair value through profit or loss of RMB 885.1 million[52] - Foreign exchange losses were primarily due to unrealized foreign exchange losses from the 8.5% depreciation of RMB against HKD[52] Net Finance Income In 2022, net finance income increased to RMB 205.1 million from RMB 154.0 million in 2021, primarily due to increased interest income earned on financial assets held for cash management purposes - Net finance income increased to RMB 205.1 million, primarily due to increased interest income earned on financial assets[53] Fair Value Loss on Preferred Shares and Other Financial Liabilities In 2021, fair value loss on preferred shares and other financial liabilities was RMB 13,525.5 million, mainly referring to changes in the fair value of preferred shares issued by the company; in 2022, this loss significantly decreased to RMB 7.2 million, primarily due to the automatic conversion of preferred shares into ordinary shares after the company's initial public offering - In 2021, fair value loss on preferred shares and other financial liabilities was RMB 13,525.5 million, primarily related to the increase in the company's valuation[54] - In 2022, this loss significantly decreased to RMB 7.2 million, primarily due to the automatic conversion of preferred shares into ordinary shares after the company's initial public offering[54] Income Tax Credit / (Expense) In 2022, income tax credit was RMB 239.8 million, compared to an income tax expense of RMB 35.5 million in 2021, primarily due to an increase in deductible tax losses - In 2022, income tax credit was RMB 239.8 million, primarily due to an increase in deductible tax losses[55] Annual Loss In 2022, the annual loss was RMB 6,093.0 million, a significant narrowing from RMB 17,177.1 million in 2021, primarily due to the combined impact of R&D investment, impairment losses on financial assets, fair value losses, and net foreign exchange losses - In 2022, the annual loss was RMB 6,093.0 million, a significant narrowing from RMB 17,177.1 million in 2021[56] - The loss was primarily affected by R&D investment, net impairment loss on financial and contract assets, fair value losses on financial assets measured at fair value through profit or loss, and net foreign exchange losses[56] Non-IFRS Measures The company uses adjusted EBITDA and adjusted net loss as non-IFRS financial metrics to eliminate the impact of non-cash items and financing and investing activities, aiding in the understanding and evaluation of operating performance, with adjusted EBITDA at RMB (4,213.4) million and adjusted net loss at RMB (4,736.2) million in 2022 - Adjusted EBITDA and adjusted net loss are used as non-IFRS metrics to eliminate the impact of non-cash items and financing and investing activities[57] Reconciliation of Non-IFRS Measures (million RMB) | Indicator | 2022 Amount | 2021 Amount | | :------------- | :--------- | :--------- | | Adjusted EBITDA | (4,213.4) | (879.0) | | Adjusted Net Loss | (4,736.2) | (1,418.4) | Trade, Other Receivables and Prepayments In 2022, total trade, other receivables and prepayments increased to RMB 6,071.7 million; net trade receivables remained relatively stable, but their aging deteriorated, with turnover days increasing to 494 days, mainly due to delayed customer payments affected by the COVID-19 pandemic; other receivables decreased to RMB 211.1 million - Total trade, other receivables and prepayments increased from RMB 5,775.9 million in 2021 to RMB 6,071.7 million in 2022[60] Trade Receivables In 2022, net trade receivables remained relatively stable, but the aging of the total amount deteriorated, with trade receivables turnover days increasing from 319 days in 2021 to 494 days, primarily due to delayed customer payments affected by the COVID-19 pandemic Trade Receivables Details (million RMB) | Indicator | 2022 Amount | 2021 Amount | | :------------- | :--------- | :--------- | | Gross Trade Receivables | 7,796.3 | 6,064.1 | | Impairment Provision | (2,578.5) | (980.4) | | Net Trade Receivables | 5,217.8 | 5,083.7 | Trade Receivables Aging Analysis (million RMB) | Aging | 2022 Amount | 2021 Amount | | :----------- | :--------- | :--------- | | Up to 6 months | 1,995.6 | 2,659.6 | | 6 months to 1 year | 746.1 | 1,048.8 | | 1 to 2 years | 3,244.9 | 1,402.8 | | 2 to 3 years | 1,180.0 | 852.1 | | Over 3 years | 629.7 | 100.8 | | Total | 7,796.3| 6,064.1| - Trade receivables turnover days increased from 319 days in 2021 to 494 days in 2022, primarily due to longer payment cycles in the Smart City business and the impact of the COVID-19 pandemic[64] Other Receivables In 2022, other receivables decreased to RMB 211.1 million from RMB 329.2 million in 2021, primarily due to a decrease in balances paid on behalf of customers - Other receivables decreased to RMB 211.1 million, primarily due to a decrease in balances paid on behalf of customers[65] Liquidity and Financial Resources As of December 31, 2022, the company held cash and cash equivalents of RMB 7,962.8 million and time deposits of RMB 6,212.9 million; net cash used in operating activities was RMB 3,084.5 million, net cash used in investing activities was RMB 9,298.2 million, and net cash generated from financing activities was RMB 3,329.1 million; total borrowings increased to RMB 3,228.6 million Cash Flow Summary (million RMB) | Cash Flow Type | 2022 Amount | 2021 Amount | | :------------------- | :--------- | :--------- | | Net Cash Used in Operating Activities | (3,084.5) | (2,485.4) | | Net Cash Used in Investing Activities | (9,298.2) | (1,548.1) | | Net Cash Generated from Financing Activities | 3,329.1 | 9,378.5 | | Cash and Cash Equivalents at Year-End | 7,962.8 | 16,529.5 | - As of December 31, 2022, the total balance of bank deposits, cash on hand, structured deposits, and fair value of bond investments was RMB 16,602.7 million[8] - Total borrowings increased from RMB 339.9 million in 2021 to RMB 3,228.6 million in 2022[71] Net Cash Used in Operating Activities In 2022, net cash used in operating activities was RMB 3,084.5 million, primarily from loss before income tax, adjusted for non-cash items such as impairment provisions for financial assets, depreciation and amortization, and fair value losses - In 2022, net cash used in operating activities was RMB 3,084.5 million[68] Net Cash Used in Investing Activities In 2022, net cash used in investing activities was RMB 9,298.2 million, primarily due to purchases of property, plant and equipment, acquisitions of debt and equity investments, and net increase in time deposits - In 2022, net cash used in investing activities was RMB 9,298.2 million[69] - Primarily due to purchases of property, plant and equipment, acquisitions of debt and equity investments, and net increase in time deposits[69] Net Cash Generated from Financing Activities In 2022, net cash generated from financing activities was RMB 3,329.1 million, primarily due to proceeds from borrowings and net proceeds from the over-allotment option of the initial public offering, partially offset by net changes in deposits for bank borrowings - In 2022, net cash generated from financing activities was RMB 3,329.1 million[70] - Primarily due to proceeds from borrowings and net proceeds from the over-allotment option of the initial public offering[70] Borrowings As of December 31, 2022, total borrowings increased to RMB 3,228.6 million; the company maintains a prudent approach to capital management, with interest rate risk primarily calculated at floating rates, and continuously monitors exchange rate movements - Total borrowings increased from RMB 339.9 million in 2021 to RMB 3,228.6 million in 2022[71] - The company maintains a prudent approach to capital management, with interest rate risk primarily calculated at floating rates and no interest rate swap contracts used for hedging[71] Exchange Rate Fluctuation Risk The company is primarily exposed to fluctuations in exchange rates between HKD and RMB, and HKD and USD; currently, it does not engage in foreign currency exchange risk hedging activities but will continue to monitor and take necessary measures to mitigate exchange rate impacts - Primarily exposed to fluctuations in exchange rates between HKD and RMB, and HKD and USD[72] - Currently does not engage in foreign currency exchange risk hedging activities but will continue to monitor and take necessary measures to mitigate exchange rate impacts[72] Other Information Employees, Training and Remuneration Policy As of December 31, 2022, the Group had 5,098 employees; the company formulates employee remuneration plans based on market compensation, industry practices, and remuneration strategies, providing year-end performance bonuses (cash or share options) based on individual performance, in addition to salaries, training courses, and employee benefits - As of December 31, 2022, the Group had 5,098 employees[73] - Employee remuneration plans combine market compensation, industry practices, and the Group's remuneration strategy, offering year-end performance bonuses (cash or share options)[73] Use of Proceeds The company received net proceeds of approximately HKD 6,351.0 million from the global offering; as of December 31, 2022, approximately HKD 5,447.2 million had been utilized, mainly for enhancing R&D capabilities, expanding business, seeking strategic investment and acquisition opportunities, and working capital and general corporate purposes; unused net proceeds are held as bank deposits - Net proceeds from the global offering were approximately HKD 6,351.0 million[74] Use of Proceeds Details (approx. million HKD) | Intended Use | Planned Proportion (%) | Planned Use | Unused at 2021 Year-End | Actual Use in 2022 | Unused at 2022 Year-End | Expected Full Utilization Time | | :------------------- | :----------- | :------- | :------------- | :------------- | :------------- | :--------------- | | Enhancing R&D Capabilities | 60.0 | 3,810.6 | 3,810.6 | 3,601.3 | 209.3 | Before end of 2023 | | Business Expansion | 15.0 | 952.7 | 952.7 | 258.2 | 694.5 | Before end of 2023 | | Strategic Investments and Acquisitions | 15.0 | 952.7 | 952.7 | 952.7 | – | Not applicable | | Working Capital and General Corporate Purposes | 10.0 | 635.0 | 635.0 | 635.0 | – | Not applicable | | Total | 100.0 | 6,351.0| 6,351.0 | 5,447.2 | 903.8 | | Capital Gearing Ratio As of December 31, 2022, the company's capital gearing ratio was -11.7%, representing a net cash position, calculated as net debt divided by total capital - As of December 31, 2022, the capital gearing ratio was -11.7%, representing a net cash position[76] Contingent Liabilities, Material Investments and Acquisitions/Disposals As of December 31, 2022, the company had no material contingent liabilities, material investments, or material acquisitions or disposals concerning subsidiaries, associates, and joint ventures - As of December 31, 2022, the company had no material contingent liabilities[77] - As of December 31, 2022, the company held no material investments[77] - For the year ended December 31, 2022, the company had no material acquisitions or disposals concerning subsidiaries, associates, and joint ventures[77] Future Plans for Material Investments and Capital Assets As of December 31, 2022, the company had no specific future plans for material investments and acquisitions of capital assets; the Group will continue to identify new investment opportunities, investing in companies related to its principal and core businesses to achieve synergies and improve services and products - As of December 31, 2022, the company had no specific future plans for material investments and acquisitions of capital assets[79] - The Group will continue to identify new investment opportunities, investing in companies related to its principal and core businesses to achieve synergies and improve services and products[79] Repurchase of Listed Securities During the reporting period and up to the announcement date, the company repurchased a total of 38,475,000 Class B shares on the Stock Exchange for a total consideration of HKD 75,956,740, all of which have been cancelled; concurrently, Class A shareholders converted 8,644,928 Class A shares into Class B shares to maintain the differential voting rights ratio 2022 Class B Share Repurchase Details | Repurchase Month | Number of Class B Shares Repurchased | Price Paid Per Class B Share (HKD) | Total Consideration (HKD) | | :------- | :-------------- | :-------------------------- | :------------- | | July | 6,700,000 | Highest 2.1, Lowest 2.1 | 14,070,000 | | September | 31,775,000 | Highest 2, Lowest 1.89 | 61,886,740 | | Total | 38,475,000 | | 75,956,740 | - All repurchased Class B shares were subsequently cancelled, and Class A shareholders converted 8,644,928 Class A shares into Class B shares to maintain the differential voting rights ratio[157] Events After Reporting Period Other than as disclosed in the announcement, there have been no other significant events affecting the Group since the end of the reporting period - Other than as disclosed in the announcement, there have been no other significant events affecting the Group since the end of the reporting period[159] Corporate Governance and Dividends The company is committed to achieving high standards of corporate governance and complies with the Corporate Governance Code set out in Appendix 14 of the Listing Rules; the Board recommends no final dividend for the year ended December 31, 2022; the Annual General Meeting will be held on June 23, 2023 - The company has complied with the applicable code provisions of the Corporate Governance Code set out in Appendix 14 of the Listing Rules, except for the Chairman and Chief Executive Officer roles being held by the same person[159][160] - The Board recommends no final dividend for the year ended December 31, 2022[161] - The Annual General Meeting will be held on June 23, 2023, and share transfer registration will be suspended from June 19 to June 23, 2023[161][162] Auditor and Audit Committee The Group's auditor, PricewaterhouseCoopers, has confirmed that the figures in the consolidated financial statements in the announcement agree with those in the audited consolidated financial statements; the Audit Committee has reviewed the Group's consolidated financial statements and discussed accounting policies and internal control matters with management and the auditor - The Group's auditor, PricewaterhouseCoopers, has confirmed that the figures in the consolidated financial statements in the announcement agree with those in the audited consolidated financial statements[164] - The Audit Committee has reviewed the Group's consolidated financial statements and discussed accounting policies and practices and internal control matters with senior management and the auditor[165] Publication of Annual Results Announcement and Annual Report This annual results announcement has been published on the Stock Exchange website and the company's website; the company's annual report for the year ended December 31, 2022, will be published and dispatched to shareholders in due course - This annual results announcement has been published on the Stock Exchange website www.hkexnews.hk and the company's website www.sensetime.com[166](index=166&type=chunk) - The company's annual report for the year ended December 31, 2022, will be published and dispatched to shareholders in due course[166] Consolidated Financial Statements Consolidated Statement of Profit or Loss Consolidated Statement of Comprehensive Loss Consolidated Statement of Financial Position Consolidated Statement of Changes in Equity Consolidated Statement of Cash Flows Notes to the Financial Statements General Information SenseTime Group Inc. was incorporated in the Cayman Islands on October 15, 2014, primarily engaged in selling advanced AI software, platforms, related services, integrated software-hardware products, and AIDC services in China, Northeast Asia, Southeast Asia, and other regions; the company is a leading AI software company, with Professor Tang Xiao'ou as the ultimate controlling shareholder, and was listed on the Main Board of the Hong Kong Stock Exchange on December 30, 2021 - The company was incorporated in the Cayman Islands on October 15, 2014, primarily engaged in selling advanced AI software, platforms, related services, integrated software-hardware products, and AIDC services[89] - The company is a leading AI software company, with Professor Tang Xiao'ou as the ultimate controlling shareholder[89] - The company was successfully listed on the Main Board of the Hong Kong Stock Exchange on December 30, 2021[89] Basis of Preparation The consolidated financial statements are prepared in accordance with International Financial Reporting Standards and the disclosure requirements of the Hong Kong Companies Ordinance, based on the historical cost convention, and modified for the revaluation of certain financial assets and financial liabilities measured at fair value - The consolidated financial statements are prepared in accordance with International Financial Reporting Standards and the disclosure requirements of the Hong Kong Companies Ordinance[90] - Prepared on a historical cost convention, and modified for the revaluation of certain financial assets and financial liabilities measured at fair value[90] Changes in Accounting Policies and Disclosures The Group first adopted amendments to IAS 16, IFRS 3, IAS 37, and Annual Improvements to IFRS Standards 2018-2020 for the fiscal year beginning January 1, 2022, which had no significant impact on the consolidated financial statements, and new standards not yet adopted in the future are also not expected to have a significant impact - The Group first adopted several standard amendments for the fiscal year beginning January 1, 2022, which had no significant impact on the consolidated financial statements[91] - New standards and amendments not yet adopted in the future are not expected to have a significant impact on the Group's consolidated financial information[92] Segment Information The company develops software and hardware products for different vertical industries and use cases based on the same AI infrastructure platform and model training framework, without distinguishing operating segment financial information; in 2022, mainland China accounted for the highest proportion of revenue, and non-current assets were mainly concentrated in mainland China; Customer A was one of the major customers, and revenue recognition was primarily at a point in time - The company does not distinguish operating segment financial information, and executive directors review the Group's financial performance as a whole[94] Revenue by Geographical Location In 2022, the Group's revenue primarily came from mainland China, accounting for 83.2% of total revenue, with revenue from Northeast Asia increasing Consolidated Total Revenue by Customer Location (thousand RMB) | Region | 2022 Amount | 2021 Amount | | :------- | :--------- | :--------- | | Mainland China | 3,168,880 | 4,147,620 | | Northeast Asia | 521,866 | 365,759 | | Southeast Asia | 95,860 | 56,944 | | Other | 21,904 | 129,940 | | Total | 3,808,510| 4,700,263| Non-current Assets As of December 31, 2022, the Group's total non-current assets were RMB 7,646,991 thousand, primarily concentrated in mainland China, accounting for 98.1% Total Non-current Assets by Asset Location (thousand RMB) | Region | 2022 Amount | 2021 Amount | | :------- | :--------- | :--------- | | Mainland China | 7,498,318 | 3,310,770 | | Northeast Asia | 66,301 | 74,633 | | Southeast Asia | 27,054 | 62,392 | | Other | 55,318 | 118,218 | | Total | 7,646,991| 3,566,013| Information on Major Customers In 2022, revenue from Customer A accounted for 15.20% of the Group's total revenue, making it one of the Group's major customers Revenue Share from Major Customers | Customer | 2022 Share | 2021 Share | | :----- | :--------- | :--------- | | Customer A | 15.20% | 11.68% | | Customer B | * | 10.73% | Disaggregation of Revenue In 2022, the Group's revenue was primarily recognized at a point in time, accounting for 93.0% of total revenue, with revenue recognized over time accounting for 7.0% Revenue Breakdown by Recognition Timing (thousand RMB) | Revenue Recognition Method | 2022 Amount | 2021 Amount | | :----------- | :--------- | :--------- | | Recognized at a point in time | 3,543,301 | 4,364,889 | | Recognized over time | 265,209 | 335,374 | | Total | 3,808,510| 4,700,263| Expenses by Nature In 2022, total expenses were RMB 7,749.1 million, with employee benefit expenses accounting for the highest proportion at RMB 4,012.5 million; hardware costs and project subcontracting service fees, depreciation and amortization, and professional and other consulting fees were the main expense items Expenses by Nature Analysis (thousand RMB) | Expense Item | 2022 Amount | 2021 Amount | | :------------------- | :--------- | :--------- | | Employee Benefit Expenses | 4,012,452 | 4,577,655 | | Hardware Costs and Project Subcontracting Service Fees | 1,118,932 | 1,265,429 | | Depreciation and Amortization | 967,685 | 657,830 | | Professional and Other Consulting Fees | 800,666 | 659,612 | | Server Operations and Cloud Service Fees | 250,985 | 201,730 | | Marketing, Conference and Travel Expenses | 247,663 | 183,710 | | Utilities, Property Management and Administrative Expenses | 168,912 | 105,525 | | Data Labeling Fees | 59,895 | 53,362 | | R&D Tools and Consumables | 42,509 | 52,140 | | Auditor's Remuneration | 19,465 | 15,856 | | Listing Expenses | – | 134,579 | | Other Expenses | 59,942 | 109,283 | | Total | 7,749,106| 8,016,711| - Employee benefit expenses in 2022 included share-based compensation expenses of approximately RMB 464,473,000, a significant decrease from RMB 1,583,925,000 in 2021[101] Other Losses – Net In 2022, net other losses were RMB 1,367.3 million, a significant increase from RMB 399.9 million in 2021, primarily composed of fair value losses on financial assets measured at fair value through profit or loss and net foreign exchange losses Other Losses – Net Details (thousand RMB) | Item | 2022 Amount | 2021 Amount | | :--------------------------------- | :--------- | :--------- | | Fair Value Loss on Financial Assets at FVTPL | (885,078) | (514,661) | | Change in Net Assets of Investment Funds Attributable to Limited Partners | (17,188) | – | | Net Foreign Exchange (Loss) / Gain | (453,432) | 132,435 | | Total | (1,367,311)| (399,958)| Income Tax (Credit) / Expense In 2022, income tax credit was RMB 239.8 million, compared to an income tax expense of RMB 35.5 million in 2021, primarily due to an increase in deductible tax losses; the Group applies different tax rates in various regions, with high-tech enterprises in mainland China enjoying a 15% preferential tax rate; as of December 31, 2022, the Group's unrecognized deferred income tax assets were RMB 1,357.6 million Income Tax (Credit) / Expense Details (thousand RMB) | Item | 2022 Amount | 2021 Amount | | :----------- | :--------- | :--------- | | Current Income Tax | 22,714 | 7,269 | | Deferred Income Tax | (262,536) | 28,237 | | Total | (239,822)| 35,506 | - In 2022, income tax credit was RMB 239.8 million, primarily due to an increase in deductible tax losses[55][112] Tax Rates by Region The Group is exempt from tax in the Cayman Islands and British Virgin Islands, with a Hong Kong profits tax rate of 16.5%, Singapore income tax rate of 17% (with a 5% preferential rate for certain activities), Japan corporate tax rate of 23.2%, Malaysia income tax rate of 24% (with a 17% preferential rate for small businesses), and Saudi Arabia income tax rate of 20% - Cayman Islands and British Virgin Islands are tax-exempt[103][104] - Hong Kong profits tax rate is 16.5%, Singapore income tax rate is 17% (with a 5% preferential rate for certain activities)[105][106] - Japan corporate tax rate is 23.2%, Malaysia income tax rate is 24% (with a 17% preferential rate for small businesses), and Saudi Arabia income tax rate is 20%[107][108][109] PRC Corporate Income Tax and Withholding Tax The statutory corporate income tax rate for PRC businesses is 25%, but high-tech enterprises such as Beijing SenseTime, Shenzhen SenseTime, Shanghai SenseTime, and Shanghai Lingang enjoy a 15% preferential income tax rate; dividends distributed to overseas investors are generally subject to a 10% withholding tax, which can be reduced to 5% for eligible Hong Kong investors - The statutory corporate income tax rate for PRC businesses is 25%, with high-tech enterprises enjoying a 15% preferential income tax rate[110] - Dividends distributed to overseas investors are generally subject to a 10% withholding tax, which can be reduced to 5% for eligible Hong Kong investors[111] Tax Losses and Deferred Income Tax Assets In 2022, the super deduction policy for R&D expenses was adjusted from 175% to 200%; as of December 31, 2022, the Group's unrecognized deferred income tax assets were RMB 1,357.6 million, and total unutilized tax losses were RMB 7,704.3 million - In 2022, the super deduction policy for R&D expenses was adjusted from 175% to 200%[114] - As of December 31, 2022, the Group's unrecognized deferred income tax assets were RMB 1,357,648,000[116] Unutilized Tax Losses Expiry Dates (thousand RMB) | Expiry Date | 2022 Amount | 2021 Amount | | :----- | :--------- | :--------- | | 2023 | 8,162 | 8,162 | | 2024 | 87,797 | 87,797 | | 2025 | 339,949 | 177,775 | | 2026 | 919,258 | 614,259 | | 2027 | 1,940,533 | 49,395 | | 2028 | 416,029 | 418,653 | | 2029 | 963,974 | 963,974 | | 2030 | 688,746 | 688,746 | | 2031 | 587,216 | 571,523 | | 2032 | 841,191 | – | | Indefinite | 911,467 | 668,920 | | Total | 7,704,322| 4,249,204| Loss Per Share In 2022, basic loss per share was RMB (0.19), a significant narrowing from RMB (1.74) in 2021; diluted loss per share was the same as basic loss per share due to the anti-dilutive effect of potential ordinary shares when the Group incurred a loss Loss Per Share (RMB) | Indicator | 2022 | 2021 | | :------------- | :----- | :----- | | Basic Loss Per Share | (0.19) | (1.74) | | Diluted Loss Per Share | (0.19) | (1.74) | - Due to the Group incurring a loss, potential ordinary shares have an anti-dilutive effect, thus diluted loss per share is the same as basic loss per share[120] Dividends For the year ended December 31, 2022, the company did not declare or pay any dividends - For the year ended December 31, 2022, the company did not declare or pay any dividends[121] [Prope