Shangtang (00020) Mid-year 2024 Financial Report Interpretation: Generative AI Business Soars, Operational Total Power Exceeds 20,000 PetaFLOPS.

On August 27th, the leading AI company in the industry, SenseTime Group (00020), released its mid-year performance report for the first half of the year ending June 30, 2024. The data shows that SenseTime Group achieved operating income of 1.74 billion yuan in the first half of the year, a year-on-year increase of 21%, demonstrating strong growth momentum. Of particular note is the significant growth of 256% in revenue from generative AI business, reaching nearly 1.1 billion yuan, accounting for a high proportion of 60% of total revenue, becoming the core engine driving the company's growth. This performance not only highlights SenseTime's deep accumulation in the field of AI but also anticipates significant results from its strategic transformation. SenseTime Group's Chairman and CEO Xu Li stated, "This mid-year performance reflects that SenseTime's strategic goal of comprehensive transformation to generative AI has been partially achieved." In Xu Li's view, the core advantage of the growth of SenseTime's generative AI business lies in the deep synergy of "large devices + large models," possessing the ability to create world-class large models, and building unique advantages in native multimodal, natural interaction with video streams, and low-cost model inference architectures. In terms of profitability, SenseTime achieved nearly 800 million yuan in gross profit in the first half of the year, an 18% year-on-year increase, with a gross profit margin maintained at a solid level of 44%. The net loss narrowed by 21% year-on-year, while EBITDA loss narrowed by 26.5%, sending a strong positive signal to the market - SenseTime is gradually emerging from the shadow of losses and steadily moving towards profitability. Technology Leading the Future: Core Reasons for the Rapid Growth of Generative AI Business The growth of SenseTime's performance in the first half of the year mainly came from the explosive growth of its generative AI business. The rapid rise of generative AI business is not accidental but a result of long-term technological accumulation and forward-looking layout by SenseTime. Looking back at SenseTime's development history, it is found that its persistence and investment in the strategy of "large devices + large models" are the key to maintaining a leading position in the field of AI. Through deep collaboration between large-scale computing infrastructure (large devices) and advanced large model products (SenseNova), SenseTime has bridged the gap from laboratory technology to large-scale commercial applications. Specifically, SenseTime's large device SenseCore provides powerful computing support, enabling the company to support trillion-parameter multimodal model training and large-scale AI inference. In addition, SenseTime's large model system "SenseNova" is continuously upgraded, significantly enhancing its comprehensive capabilities. During the reporting period, SenseTime's large device intelligence computing service achieved significant expansion, with the total operational computing power scale exceeding 20,000 PetaFLOPS, up by approximately 70% compared to March 2024; the number of GPUs online reached 54,000, leading the industry. The computing power managed by SenseTime's large device has achieved unified scheduling nationwide, expanding new computing nodes in cities such as Shanghai, Shenzhen, Guangzhou, Fuzhou, Jinan, and Chongqing. Shanghai Lingang HoldingsAIDC has reached 14,000 petaFLOPS, far exceeding the planned total computing power of 3,740 petaFLOPS. While supporting SenseTime's own large model research and development, powerful computing power can also support the training and deployment of large models for external customers. The strong computing power can support the simultaneous training of more than 20 trillion large models and support the full life cycle generation of trillion-parameter large models. The company has the capability to design, construct, and operate from a thousand-card cluster to a ten-thousand-card cluster, allowing efficient support for training and inference of multimodal models with up to 2 trillion parameters. In addition, SenseTime has realized elastic and on-demand scaling of inference services through technical innovation, significantly reducing the overall cost of large-scale AI inference and improving the overall operational return on investment of the intelligence computing center. Empowered by the large device, SenseTime's progress in generative AI technology is unstoppable. The "SenseNova" large model system was iterated to version 5.0 within a year of its release in April 2023, surpassing GPT-4 Turbo in the SuperCLUE test, demonstrating outstanding performance in traditional scenarios such as code writing, data analysis, and medical question answering, and gradually expanding application boundaries to emerging fields such as electric vehicles, finance, and Siasun Robot & Automation. These leading technological advantages in technology and implementation have not only earned SenseTime market favor but also laid a solid foundation for its future development. Differentiated Advantages: Leading Technology, Facilitating Customer Easy Implementation In the intensifying competition of large models today, differentiation is becoming a key factor for AI companies to stand out. SenseTime is well aware of this, and therefore, continues to strengthen its inference and multimodal capabilities based on large models and large devices. In the inference scenario, SenseTime's large device has achieved a 4-fold increase in the number of concurrent inferences (QPS) under the same computing power and electricity costs through innovative technical architecture, and achieved elastic and on-demand scaling of inference services, significantly optimizing the overall cost of large-scale AI inference. In addition, based on original technologies such as integrated training and inference and collaborative computing, SenseTime continues to increase the GPU resource utilization rate of ten-thousand-card clusters to over 80%, enhancing the overall operational return on investment of the intelligence computing center. In terms of multimodal capabilities, SenseTime SenseNova achieves seamless integration and interaction of various data types such as text, images, and speech. After upgrading to version 5.5 in July 2024, SenseNova's comprehensive capabilities have increased by about 30% compared to version 5.0, significantly strengthening its multimodal capabilities. It has achieved real-time interactive experience benchmarked against GPT-4 in terms of performance and is the first multimodal real-time interactive large model in China. Multimodal capabilities not only greatly expand the application scenarios of large models but also allow AI systems to better understand human needs and intentions. For example, in the medical field, SenseTime's large models can analyze patient medical records and examination reports to provide more precise diagnostic recommendations to doctors; in the financial field, it can provide investors with more scientific investment advice by deep mining and analyzing market data. In the intelligent automotive field, the "Jueying" intelligent automotive business achieved revenue of 170 million yuan, an increase of 100% year-on-year, and added 15 new fixed-point models.In 2020, the number of new designated vehicles reached approximately 6 million. With the continuous expansion of the smart car market, SenseTime's smart car business is expected to continue to maintain rapid growth.In the field of intelligent driving, Jiuying is the earliest Chinese enterprise to follow a pure visual large model route. In April of this year at the Beijing Auto Show, they completed a fully automatic driving process without high-precision maps, relying solely on visual conditions. In the field of intelligent cabins, relying on the Dayunxin large model 5.5 and multi-modal capabilities, they have completed cooperation with several well-known enterprise brands in the field of car modules, such as intelligent voice interaction, instant drawing with characters, and smart medical AI diagnosis, creating a more comfortable and safe driving experience. So far, more than 3,000 top companies in various industries have used Sensetime's large models and intelligent computing services, covering industries such as the Internet, smart hardware, new energy vehicles, Siasun Robot & Automation, healthcare, and finance. Sensetime has also established deep cooperation with well-known domestic and foreign car companies such as Honda, BYD Company Limited, and Great Wall, promoting the widespread application of AI large models in the field of intelligent cabins and autonomous driving. The undervalued value: High potential behind low valuation Compared to foreign AI innovation companies, Sensetime has demonstrated considerable competitiveness and growth potential in multiple areas. In terms of product technology, Sensetime's anthropomorphic interaction business compared to Character.ai, enterprise-level services compared to Cohere, instant drawing compared to Midjourney, and multi-modal large models compared to ChatGPT, all show differentiated value and advantages. In terms of market share and customer resources, Sensetime has not only made breakthroughs but also has extremely strong growth potential. However, despite Sensetime's significant achievements in technology and the market, its market value is undervalued. This may be due to factors such as market sentiment and industry perception biases. But with the continuous innovation of the company's technology and the continuous expansion of the market, Sensetime's market value is expected to be reassessed. The market's feedback on Sensetime Technology has been positively received. Many brokerage institutions, including Zhongjin, Galaxy, Zhongxin, Haitong, have given positive evaluations of Sensetime Technology's future development prospects and have set target prices. For example, Zhongjin has set a target price of 2 Hong Kong dollars, corresponding to forecasted market-to-sales ratios of 14 times and 10 times for the next two years. Next, with breakthroughs in technological innovation, mature product expansion, and important customer cooperation, investor attention and recognition are expected to increase. After all, as an excellent investment target, Sensetime still has a huge growth space for its future market value. Conclusion According to IDC reports, the Chinese intelligent computing service market will continue to grow at a high rate in the next five years, with an expected annual compound growth rate of 57.3%. By 2028, the overall market size is expected to reach nearly 200 billion RMB. Sensetime, as a leading industry enterprise with a forward-looking layout, will benefit from the rapid growth of the market. With the continuous maturity of generative AI technology and the sustained growth of market demand, Sensetime is expected to accelerate its profit expansion in the next few years. In order to achieve this goal, Sensetime will continue to increase its investment in technology research and development and market expansion, consolidating and expanding its leading position in the field of large models and large devices. At the same time, Sensetime will actively explore new application scenarios and business models to promote the popularization and application of AI technology, further expanding the boundaries of its business and industry.