Incident: On August 30, 2024, Cambrian released its 2024 semi-annual results report. 24H1 achieved operating income of 0.065 billion yuan, yoy -43.4%, and net profit to mother of -0.53 billion yuan, yoy +2.7%.

The negative impact on the supply chain is significant, and 24H1 performance is under pressure. 24H1 achieved operating income of 0.065 billion yuan, yoy -43%, mainly due to factors such as the “entity list”, which adversely affected the company's supply chain; achieved a gross profit margin of 62.72%, a year-on-year decline of 8.14pct; realized net profit to mother of -0.53 billion yuan, yoy +2.7%, which was narrower than the loss of 14.72 million yuan in the same period last year. The loss was mainly due to the company's continued vigorous investment in R&D.

Looking at a single quarter, the company achieved operating income of 0.039 billion yuan, qoq +52.34%; realized net profit to mother of 0.303 billion yuan, a decrease of 0.076 billion yuan over the previous quarter; and achieved a gross profit margin of 66.08%, an increase of 8.47 pcts over the previous month.

The sharp increase in prepaid accounts highlights expectations for business growth. The company's prepaid accounts rose sharply from 0.148 billion yuan in 2023H2 to 0.55 billion yuan in 2024H1. The prepaid account will eventually be carried over to operating costs. We believe this reflects the company's active future production plans and shows good growth expectations.

From a business perspective, the company continued in-depth cooperation with downstream customers in the first half of this year, and some progress was made in some fields: 1) Cooperation with customers in this field was continuously deepened in the Internet field, batch shipments were achieved in natural language application scenarios, and product measurement capabilities and iterative expectations all met customer needs. 2) In the field of large models, the company continues to develop business and technical cooperation with leading domestic algorithm companies. 3) In the field of large visual models, we have obtained multiple product compatibility certifications from customers. 4) In the financial field, the company and many financial institutions have tested large language models and explored industry applications, and actively promoted the implementation of applications in the financial field.

Judging from the company's financial situation over the years, results are usually released in the second half of the year. With the gradual preparation of the company's products, results can be expected in the second half of the year.

Actively invest in research and development, and continue to strengthen products and ecosystems. The company has a R&D team of 727 people, of which 78.82% of R&D personnel have master's degree or above, which strongly supports the company's technological innovation and product development. In 24H1, the company continued to invest heavily in R&D. The R&D expenses in the first half of the year were 0.45 billion yuan. Judging from R&D progress:

In terms of hardware, the company is developing a next-generation intelligent processor microarchitecture and instruction set, focusing on optimizing scenarios such as large models of natural language processing, large models of video image generation, and training and reasoning of large models of recommended systems; in terms of software: On the one hand, the company promotes the versatility of training platforms and shortens the company's product adaptation cycle. On the other hand, it is also actively building an ecosystem. For example, the company's training software platform implemented native support for MLU by the Transformers and Accelerate community, added the Triton3.0.x Cambrian backend, and open-sourced the cross-platform AI compiler front-end Triton-Linalg, improving the development efficiency of adapting the industry's artificial intelligence applications to the company's products. Furthermore, the company successfully verified the pre-training function of large language models on the cluster, and achieved industry-leading continuous and stable operation of a single training task over a long period of time, leading the industry in computational efficiency and energy efficiency.

Investment advice: The company's product strength is constantly being strengthened, ecological construction is gradually improved, and positive progress is continuously being made at the business level. We expect the company to achieve operating income of 1.679/2.271/3.085 billion yuan in 2024-2026, and net profit to mother of 0.446/-0.208/0.148 billion yuan, respectively. Maintain a “buy” rating.

Risk warning: supply chain policy fluctuations, R&D progress falling short of expectations, increased industry competition