We are a leading AI software company serving a broad range of industries with a focus on computer vision software. We are the largest AI software company in Asia in terms of revenue in 2020, and the largest computer vision software provider in China in terms of revenue in 2020 with a market share of 11%, according to Frost & Sullivan. We have built market leadership by helping customers drive productivity, creativity and efficiency with our AI software platforms. As of June 30, 2021, our software platforms had been used by a total of over 2,400 customers, including over 250 Fortune 500 and other publicly-listed companies, 119 cities and over 30 automobile companies, while empowering over 450 million mobile phones and over 200 mobile apps.
Our business is underpinned by our original and cutting-edge research, recognized by over 70 first-prize awards in global academic competitions, more than 600 top-tier academic paper publications and over 8,000 AI patents and patent applications. We have built a first-of-its-kind universal AI infrastructure to achieve mass production of a diverse and growing portfolio of AI models with rich functionality and superior accuracy. We develop scalable AI software platforms to facilitate the rapid deployment of AI models and applications in numerous scenarios. Fueled by technological excellence and scale effects, we have achieved leading market positions in smart business, smart city, smart life and smart auto, serving a wide spectrum of industries across commercial space management, residential property management, urban management, manufacturing, infrastructure, transportation, mobile devices and applications, healthcare and automobiles, according to Frost & Sullivan.
Our culture is deep-rooted in academic excellence. In 2014, Professor Tang Xiao’ou, who founded the CUHK Multimedia Lab in Hong Kong in 2001, and other core members of the lab jointly started our company. As of June 30, 2021, 40 professors led our research efforts, and approximately two-thirds of our over 5,000 employees were scientists and engineers. This vast intellectual capital lays the foundation for a comprehensive and integrated innovation system from AI research to production and allows us to offer and continuously improve industry-leading, full-stack AI capabilities, cementing our commercialization success.
AI Models are the Building Blocks for AI Software, Underpinning Digital Transformation
According to Frost & Sullivan, AI software is expected to be one of the fastest-growing business areas in this decade. The global AI software market size is expected to reach USD121.8 billion in 2025, growing at a CAGR of 31.9% from 2020. The computer vision software market in China is expected to grow at a CAGR of 43.5% from RMB16.7 billion in 2020 to RMB101.7 billion in 2025. As the core building blocks of AI software, AI models are algorithms which can take unstructured data as an input and transform it into informative output. As digital transformation accelerates, AI models are being widely deployed at customer locations, over the cloud and on an increasingly large number of devices. The ability to produce high-performance AI models at scale and in a cost-effective way is crucial and represents a major technology entry barrier in the AI industry. Based on our innovations and technology breakthroughs, we have built a first-of- its-kind AI infrastructure capable of mass production of high-performance AI models. We have also enabled rapid and code-free deployment of AI models and applications in numerous scenarios through our software platforms. As our AI models become more sophisticated and accurate in processing real-world data and driving various AI applications, they become the building blocks to advance digital transformation across industries.
Centralized Mass Production of AI Models with our Proprietary AI Infrastructure — SenseCore
Our proprietary AI infrastructure, SenseCore, makes industrial-grade AI model production feasible. SenseCore is built on three pillars: (i) large-scale supercomputing power; (ii) massive data processing and desensitization; and (iii) shared platforms and production tools for developers. It distinguishes us through the following features:
‧ Supporting the training of state-of-the-art large AI models for high performance and accuracy
‧ Low-cost production of scenario-specific AI models, achieving economies of scale for model improvement
‧ Industry-leading automatic machine learning for efficiency and easy use
‧ Industry-leading privacy computing and data desensitization
‧ Cross-chip, cross-device and cross-cloud-platform adaptability
‧ Comprehensive AI functions across industry verticals
By using SenseCore to power all AI workloads, our researchers and engineers can develop AI models in hours rather than in weeks. As we continuously enhance SenseCore’s capabilities and capacity, our R&D staff developed an annual average of 0.44, 3.45 and 5.24 AI models per person in 2019, 2020 and the first half of 2021, respectively. As of June 30, 2021, we had developed over 22,000 commercialized AI models to power varied applications across industry verticals.
Rapid Deployment and Commercialization of AI Models through Software Platforms
We develop and offer standard software platforms with modular flexibility to customers empowered by SenseCore. Our software platforms can be seamlessly integrated with customers’ devices or IT infrastructure, and our AI models can be deployed both on edge devices and on the cloud through our software platforms.
‧ SenseFoundry-Enterprise for Smart Business: SenseFoundry-Enterprise is used to create interconnected, efficient and scalable operations that drive better business outcomes for customers. It is a one-stop software platform embedded with more than 9,300 AI models and enables various applications which meet various industry needs for real-world data perception and process automation across industries. SenseFoundry-Enterprise enables customers to build AI-integrated workflows and operations in a code-free, modular, flexible and scalable manner. SenseFoundry-Enterprise has been widely adopted in many industrial verticals, including commercial space management, residential property management, manufacturing, infrastructure, transportation and financial services. In the first half of 2021, we served 635 Smart Business customers.
‧ SenseFoundry for Smart City: SenseFoundry software platform embedded with more than 14,000 AI models is transcribing raw and real-time city visual data into insights, alerts and actions. SenseFoundry is used to monitor the conditions of public facilities, detect incidents and track the impact of natural disasters. SenseFoundry is also equipped with an online incremental training engine derived from SenseCore, through which it provides AI- as-a-Service to cities. SenseFoundry facilitates the transformation of urban administration from human-intensive to human-computer interactive, from empirical judgment-based to data-driven, and from passive response to early detection. SenseFoundry has become the operating system for digital city operations and improved the safety, efficiency, convenience and environmental quality of cities. As of June 30, 2021, it had been deployed in 119 cities in China and overseas, mainly serving end users from the public sector, including municipal governments and their departments.
‧ SenseME, SenseMARS and SenseCare for Smart Life: We have built a multi-layer infrastructure to empower IoT devices and the Metaverse enriching user experiences. With a full stack of offerings including SDKs, AI sensors and ISP chips, our SenseME software platform, powered by over 3,500 AI models, enables a broad range of IoT devices to facilitate perception intelligence and content enhancement. Our SenseMARS software platform, powered by over 3,500 AI models, supports the development of Metaverse to create exciting new life experiences. SenseME and SenseMARS create the interface connecting the physical and digital worlds by empowering more than 200 types of mobile phones, AR and VR glasses, smart screens and consumer drones. As of June 30, 2021, SenseME and SenseMARS had empowered over 450 million mobile phones and over 200 mobile apps. Our AI software platform for smart healthcare, SenseCare, provides AI tools in diagnosis, treatment planning and rehabilitation. We have obtained three NMPA certifications and two CE marks for five SenseCare modules.
‧ SenseAuto for Smart Auto: Our SenseAuto software platform, powered by around 1,400 AI models, provides automobile companies with ADAS systems, smart cabin systems and AI-as-a-Service which enables them to develop and enhance their autonomous driving capabilities. Since 2017, we have been a strategic partner with Honda to provide it with our autonomous driving-related AI technologies. In addition, we have launched SenseAuto Robobus for L4 autonomous shuttle services for bus operating companies. We also developed SenseAuto Connect, our V2X product, that enables smart interactions among vehicles and their surroundings such as roads, traffic lights and roadside units. As of June 30, 2021, we had collaborated with over 30 automobile companies and been selected as the supplier for more than 20 million automobiles across over 50 vehicle models in the next several years. We were recognized by CB Insights Research as The Most Valuable Private Auto Tech Company in 2021.
With our software platforms, we have achieved economies of scale and shortened time-to- market of our AI model deployment and commercialization. In addition, we also provide SenseCore’s capabilities as a versatile AI-as-a-Service offering to customers through our software platforms, enabling them to produce AI models tailored to their business needs with minimal effort, expertise and investment. To further enhance SenseCore’s production capabilities and expand our AI-as-a-Service offering to more industry verticals, we are constructing a large-scale AIDC with a designed computing capacity of 3.74 exaFLOPS in Shanghai, which will bring our total computing capacity to 4.91 exaFLOPS. This AIDC is expected to be launched in early 2022, and become one of the largest supercomputers in Asia. We believe this AIDC and others we plan to build will not only accelerate our innovations and enhance our competitiveness, but also cultivate an open and rapidly growing ecosystem that further strengthens the connections with our customers.
We have achieved strong growth since our inception in 2014. Our revenues grew from RMB1,853.4 million in 2018 to RMB3,026.6 million in 2019 and further to RMB3,446.2 million in 2020, and from RMB861.2 million in the first half of 2020 to RMB1,651.8 million in the first half of 2021. Our gross profit margin grew from 56.5% in 2018 to 56.8% in 2019 and further to 70.6% in 2020, and from 72.1% in the first half of 2020 to 73.0% in the first half of 2021. Our research and development expenses in 2018, 2019, 2020 and the first half of 2021 were RMB848.7 million, RMB1,916.0 million, RMB2,453.9 million and RMB1,771.7 million, respectively. Our net losses in 2018, 2019, 2020 and the first half of 2021 were RMB3,432.7 million, RMB4,967.7 million, RMB12,158.3 million and RMB3,712.9 million, respectively. Our historical net losses were also largely attributable to the fair value losses of our preferred shares.
Eliminating the impact of items that our management does not consider to be indicative of our operating performance, we had adjusted net losses (non-IFRS measure) of RMB150.0 million, RMB1,155.2 million, RMB707.7 million and RMB578.3 million in 2018, 2019, 2020 and the first half of 2021, respectively.
Source: SenseTime-W (00020) Prospectus (IPO Date : 2021/12/07) |