Performance fluctuates greatly, and cloud-based smart chips and computer clusters dominate. According to the 2023 Earnings Report data, the 2023 operating income was 709 million yuan, a year-on-year decrease of 2.70%. Net profit to mother was 836 million yuan, after deducting non-net profit of 1,047 million yuan. Classified by product, according to 2022 data, cloud smart chips and accelerator cards had a gross profit margin of 63.47%, accounting for 30.10%, or 173.52%; edge smart chips and accelerator cards had revenue of 38 million yuan, gross profit margin of 31.11%, accounting for 5.19%, a year-on-year decrease of 78.40%; terminal intelligent processor IP revenue of 1.14 million yuan, accounting for 0.16%, a year-on-year decrease of 83.44%; intelligent computing cluster system revenue was 459 million yuan, gross profit margin 70.42%, accounting for 70.42% 62.89%, or 0.64% year on year; other revenue was $0.12 billion, accounting for 1.66%.

The “four-dimensional linkage” of R&D, talent, market and ecology combines internal and external integration to build competitiveness.

In terms of R&D, the company maintains efficient R&D investment to enhance the core competitiveness of core processor chips in the field of artificial intelligence and build technical barriers. In the first half of 2023, the company invested 483 million yuan in R&D, accounting for 421.56% of revenue. As of the first half of 2023, the company had a R&D team of 980 people, accounting for 77.47% of the total number of employees. More than 77.04% of R&D personnel had master's degree or above. R&D investment is focused on two aspects: 1) In terms of intelligent processor microarchitecture and instruction set, the company has so far independently developed five generation intelligent processor microarchitectures and five generation commercial intelligent processor instruction sets. In the first half of 2023, the sixth-generation intelligent processor microarchitecture and instruction set are being developed. The next-generation intelligent processor microarchitecture and instruction set will focus on optimizing scenarios such as training and reasoning for large natural language processing models and recommended systems, and will enhance product competitiveness in terms of programming flexibility, energy efficiency, power consumption, and area. 2) In terms of the basic system software platform, in the first half of 2023, the company's basic system software platform was also optimized and iterated compared to the previous version. Among them, the inference software platform has achieved breakthroughs and progress in key areas such as model performance optimization, support for large models and AIGC inference services, and optimization of inference performance. The training software platform has been optimized in terms of versatility and performance, and improvements and iterations have been achieved in key areas such as large models, AIGC training, and recommendation systems. The continuous improvement and development of the basic system software platform will promote the better implementation of hardware products.

In terms of talent, the company attaches great importance to the improvement and development of the talent system. While actively introducing high-level talents, it also focuses on internal talent development. At present, the company has established a mature and stable R&D team, sales service system, management and support team. In order to improve the talent team building and enhance the cohesion and combat effectiveness of the core team, the company has a complete system of target management, remuneration incentives, training and promotion. At the same time, the company continues to optimize its organizational structure, refine job responsibilities, promote informatization construction, and lay a solid foundation for sustainable development.

In terms of ecology, the company is committed to ecological construction to enhance brand influence. First half of 2023. With the continuous increase in the company's user industry and number of users, and the continuous optimization of the basic system software platform, users can use the company's software and hardware platforms more flexibly for development. Second, with excellent product energy efficiency and good service awareness, the company's brand influence continues to grow. Furthermore, the company attaches importance to industry-academia cooperation. Currently, many colleges and universities have offered artificial intelligence courses based on Cambrian software and hardware platforms, and established a relatively complete college curriculum ecosystem. This move will further help the development of the company's ecological construction.

On the market side, based on the results of continuous expansion and deep cultivation in recent years, the company has carried out extensive business deployment and implementation in the Internet, operator, finance, power and energy industries and customers. For example, in the Internet industry, the company's chips and accelerator cards and several leading Internet companies have entered the mass sales process in scenarios such as vision, voice, graphic recognition, and natural language processing. The company has carried out in-depth technical cooperation and adaptation in the field of search and recommendation with leading Internet companies, and is actively promoting product implementation. In the financial industry, the company continues to expand the mature business applications of cloud products in many leading banks. At the same time, in line with the application needs of the financial industry, it actively promotes technical cooperation and in-depth algorithm adaptation with leading banks and leading ISVs in the field of large language models, laying a solid foundation for large-scale implementation of subsequent products. In the operator industry, relying on cloud-based product chips and accelerator cards, the company's sales and service personnel actively participate in centralized procurement of leading operators, actively promote adaptation to their intelligent computing platforms, and continue to discuss and further verify the application of large language models and large-scale cluster architecture designs. Furthermore, based on the advantages of cloud-based products, the company optimized the performance of the company's products in the field of AIGC and large language models for the recently emerging field of large models, and promoted technical and product cooperation with various industry customers and ISVs.

Cloud processors and intelligent computing clusters develop collaboratively, and software platforms strengthen ecological barriers.

In terms of cloud products, according to the IDC report quoted in the company's investor relations activity record sheet on April 3, 2023, China's cloud intelligent server market is expected to reach 10.86 billion US dollars by 2025. As the core components of cloud intelligent servers, smart chips and acceleration cards account for more than 2/3 of the total server value. It is estimated that the cloud smart chip and acceleration card market will exceed 7 billion US dollars in 2025. According to the investor activity records quoted above, in the cloud intelligent computing market around 2023, based on the advantages of the software ecosystem, Nvidia's GPU chips and accelerator card products occupied more than 90% of the market share, and the market share of the company and other chip manufacturers was temporarily small; the company's cloud products are mainly used in the Internet, finance and other fields, where Internet customers occupy most of the market space in the cloud intelligent computing market. Customer development in the Internet industry has made some progress. For example, close communication has been carried out with various business departments of leading Internet companies such as Ali and Baidu, and products have been introduced. The adaptability performance in visual and voice scenarios has exceeded customer expectations, and some scenarios have already generated a certain amount of revenue.

In the field of finance, the company introduced and adapted products with a number of leading banks. Among them, the measured performance of the company's cloud products exceeds that of competitors in various business scenarios of China Merchants Bank, which can greatly improve the execution efficiency of customer-related businesses. In addition, the company and server vendors were shortlisted for the leading operator's centralized procurement project for artificial intelligence general computing equipment from 2021 to 2022, taking the first step in expanding into the operator industry. In terms of customers in other industries, the company continues to expand, accelerate the implementation of scenarios, and enable AI in traditional industries. In addition, the company has developed close cooperation with leading customers in the Internet, operators, smart finance, smart cars, smart rail transit, smart farming and other industries, and has sold products to 100 customers.

In terms of intelligent computing clusters, with the planning and launch of intelligent computing infrastructure construction in various regions, the intelligent computing cluster system business has a broad market space and market demand. Combining new urban infrastructure and industry needs such as autonomous smart cars, the market space for intelligent computing centers is expected to reach tens of billions in the next 3-5 years. The company's intelligent computing cluster system business combines the company's self-developed accelerator card or training machine products with server equipment, network equipment and storage equipment provided by partners, and equipped with the company's cluster management software. The core computing power source is a cloud smart chip developed by the company. The intelligent computing cluster mainly focuses on the application of artificial intelligence technology in data centers. It provides customers with relatively weak technical ability to deploy artificial intelligence applications with overall software and hardware solutions to scientifically configure and manage the software and hardware of the cluster and improve operational efficiency. In the field of intelligent computing clusters, the company has strong competitiveness and a good brand effect accumulated in early implementation projects. Although the company's market share is lower than that of major competitors, the company's intelligent computing cluster system business is still in the first tier of the domestic market share.

In terms of software platforms, as a chip company, building a software platform ecosystem is essential. Since its establishment, the company has continuously invested in R&D in the software platform ecosystem, effectively supporting customer delivery and commercial application of products in the company's three “cloud edge” product lines. It has provided a software development platform with complete functions and good versatility for developers to develop intelligent applications on the company's chip products, and has formed a certain developer influence. The company also launched the “Developer Ecosystem” project to continuously improve the developer community and forum platform, and use the company's accumulated scientific research and technology advantages to cooperate with universities to cultivate the usage habits of future developers. As original algorithms and technologies in the field of artificial intelligence are still emerging, new application scenarios in the artificial intelligence industry are still being introduced, and the company's chip products and its architecture and instruction set are still being upgraded and expanded, and the company needs to carry out long-term continuous iterative development and optimization of a unified software development platform. In the future, the company will also continue to iteratively develop a unified software development platform for each product line, continue to expand its functionality, efficiency, ease of use, and good support for emerging artificial intelligence applications through continuous iterative upgrades, and promote the consolidation and increase of the company's chip product market share.

The company's share repurchase was completed, and 241,400 shares had been repurchased as of February 28, 2024. As of February 28, 2024, the implementation of the company's share repurchase has been completed and the repurchase period has expired. 241,400 shares have been repurchased through centralized bidding transactions through the Shanghai Stock Exchange trading system, accounting for 0.06% of the company's total share capital.

The highest price for the repurchase transaction was RMB 157.01 per share, the lowest price was RMB 100.81 per share, and the total amount of capital paid was RMB 30.201 million (excluding transaction fees such as stamp duty and transaction fees). All repurchases of shares will be used in employee stock ownership plans or equity incentives at an appropriate time in the future.

On April 15, 2023, the company issued the “2022 Listing Notice for Issuing A Shares to Specific Targets”. The number of shares issued in this fixed increase was 13.806 million shares, the issue price was 121.10 yuan/share, the total capital raised was 1,672 billion yuan, and the net capital raised was 1,649 billion yuan. A total of 18 companies participated in this distribution, all subscribed in cash. The actual capital raised will be used for advanced process platform chip projects, stable process platform chip projects, general intelligent processor technology research and development projects for emerging application scenarios. The details are as follows: 1) Advanced process platform chip projects, with a total investment of 950 million yuan, to be raised 718 million yuan; 2) stabilized process platform chip projects, with a total investment of 1,493 million yuan, to be raised 700 million yuan; 3) General intelligent processor technology research and development projects for emerging application scenarios, with a total investment of 234 million yuan to be raised 219 million yuan; 4) To supplement working capital, the total investment is 213 million yuan, and the capital raised is to be used to raise 36 million yuan.

Profit forecasts and investment recommendations:

On November 17, 2023, the company issued the “2023 Restricted Stock Incentive Plan (Draft)”. The first vesting period is 2024, with operating income of not less than 800 million yuan, reaching the trigger value, 80% company-level ownership, with operating income not less than 1.1 billion yuan, reaching 100% corporate level ownership; the second vesting period is 2025, and the cumulative revenue value for 2024-2025 is not less than 2.08 billion yuan, reaching the trigger value, 80% company-level ownership, with cumulative operating income of 2024-2025 The revenue value was not less than 2.6 billion yuan, reaching the target value, and 100% corporate ownership. The third ownership period is 2026. The cumulative revenue value for 2024-2026 was not less than 3.68 billion yuan, reaching the trigger value. 80% company-level ownership, and the cumulative revenue value for 2024-2026 was not less than 4.6 billion yuan, reaching the target value, and 100% corporate level ownership.

We believe that the company is a scarce AI chip manufacturer in China. It can provide AI chip products and platform basic software with cloud-edge integration, software-hardware collaboration, training and reasoning integration, and a complete ecosystem. We are optimistic about the demand for AI chips and opportunities for domestic substitution, and believe that the company will benefit from this.

We expect that in 23-25, the company's highest share of product intelligent computing clusters will grow at -10%, 100%, and 50%, respectively; the company's revenue growth rates will be -2.70%, 96.06%, and 48.88%, respectively. The gross margin of the company's intelligent computing cluster continued to increase, reaching 70%, 71%, and 72% respectively in 23-25; the company's overall gross margin was 65.80%, 68.21%, and 69.92%.

We expect the company's revenue for 2023-2025 to be 709/13.91/2,071 billion yuan, net profit to mother of -8.36/ -5.49/ -266 million yuan, respectively, and corresponding EPS of -2.01/-1.32/-0.64 yuan, respectively. Due to external restrictions on the AI chip production capacity of the company's products, future operations may be uncertain, and for the first time, coverage was given an “no rating.”

Risk warning: Product iteration falls short of expectations; market competition intensifies; trade frictions intensify.