Implementation, commercialization or acceleration of high-performance computing products

On March 12, 2024, the company issued the “Voluntary Disclosure Notice on the Company's New Product Development Progress” (hereinafter referred to as the “Notice”). The company successfully developed Jinghong series of high-performance intelligent computing modules and complete machine products for application fields such as AI training, AI reasoning, and scientific computing. We believe that the release of the Jinghong series of high-performance computing products has expanded the company's product layout in the field of AI computing, and commercialization of computing power products is expected to accelerate. Considering the delay in the procurement pace of downstream customers and the reduction in profit forecasts, the company's net profit for 23-25 is estimated to be 0.56, 2.22, and 323 million yuan (previous values of 256, 3.57, and 472 million yuan). Using the segmented valuation method, the GPU business was given 25E 32.5x PS, corresponding to a valuation of 33.807 billion yuan; 25E 19.3x PE for other businesses, corresponding to a valuation of 10.04 billion yuan. In summary, the company was given a target market value of 43.85 billion yuan, corresponding to a target price of 95.86 yuan (previous value of 94.27 yuan), and a “purchase”.

Jinghong series chips support AI computing and are expected to expand the company's reach to market space. According to the announcement, the Jinghong series supports mixed precision computation such as INT8, FP16, FP32, FP64, etc., supports new multi-card interconnection technology for computing power expansion, adapts to mainstream domestic and foreign CPUs, operating systems and server vendors, and can support the current mainstream computing ecosystem, deep learning frameworks and algorithm model libraries, greatly shortening the user adaptation verification cycle. We believe that the launch of the Jinghong series of products can enrich the company's computing power product line, and in particular fill gaps in high-performance intelligent computing products. It is expected to help the company expand its market business in application fields such as AI reasoning, AI training, and scientific computing, and enhance the accessible market space.

Commercialization capabilities have been verified once again, and the pace of commercialization is expected to accelerate the early company's products, mainly focusing on graphics display GPUs. The JM7200 was released in 2018, the first chip in the JM9 series was released in 2021, and the second chip in the JM9 series was released in 2022. Since then, the company began to gradually expand into the field of general computing. On July 24, 2023, the company issued the “Prospectus for Issuing A Shares to Specific Targets in 2023". It plans to raise no more than 3,974 billion yuan for high-performance general-purpose GPU chip R&D and industrialization projects and general-purpose GPU advanced architecture R&D center construction projects, which are still in progress. We believe that the company has been deeply involved in the GPU field for many years, and its commercialization capabilities have been gradually verified. With the successful development of the Jinghong series, the pace of commercialization of computing power products is expected to accelerate.

Domestic AI intelligent computing demand is expected to drive the company's revenue growth

We believe that as large models such as Sora and Claude 3 continue to be iterated and multi-model and text-based capabilities continue to be upgraded, AI computing power demand is expected to accelerate growth. According to AMD, the global data center AIGPU market is 45 billion US dollars in 2023, and the market size is expected to reach 400 billion US dollars in 2027, and the CAGR will reach 72.7% in 2024-2027. In this context, demand for AI computing power chips is expected to be released at an accelerated pace. Furthermore, considering that domestic access to overseas high-performance computing chips is limited, the localization of AI GPUs is expected to accelerate. Based on this, we believe that the release of the Jinghong series of high-performance computing products is expected to fully benefit from the AI wave and drive the company's steady revenue growth in 24-25.

Risk warning: Chip production costs are rising, and downstream demand falls short of expectations.