Performance summary: In the first half of 2024, the company achieved operating income of 0.065 billion yuan, a year-on-year decrease of 43.42%, achieved net profit to mother of -0.53 billion yuan, and losses narrowed 2.70% year on year. 2024Q2 achieved revenue of 0.039 billion yuan in a single quarter, a year-on-year decrease of 0.21%, and realized net profit to mother of 0.303 billion yuan. Losses narrowed 4.71% year-on-year.

The company's cash payments for the purchase and construction of fixed assets, intangible assets and other long-term assets in the first half of 2024 were 0.117 billion yuan, up 155.01% year on year, advance payment was 0.55 billion yuan, up 7167.02% year on year, and inventory was 0.235 billion yuan, up 7.39% year on year. Among R&D expenses, testing and testing processing costs were 0.102 billion yuan, an increase of 849% over the previous year.

The company continues to promote the optimization of basic software system platforms. The company's training software platform updated and iterated on PyTorch 2.1/2.3 support, implemented native MLU support from the Transformers and Accelerate communities, and added the Cambrian backend for Triton 3.0.x. At the same time, the company open-sourced the cross-platform AI compiler front-end Triton-Linalg, improving the development efficiency of product adaptation. In addition, the company's inference software platform also supports and optimizes literary and literary models such as the Llama series and the Qwen series, as well as multi-modal models such as StableDiffusion3.

During the reporting period, the company successfully verified the pre-training function of large language models on the cluster. The company's smart chip products have been shipped in batches in natural language application scenarios in the Internet field, and has carried out business and technical cooperation with leading domestic model algorithm companies.

We believe that as the company continues to iterate and optimize the basic software system platform for generative AI scenarios and continuously improve the usability and stability of the basic software system platform, it is expected to further enhance the competitive advantage of software and hardware collaboration and accelerate the implementation of next-generation products in the AI scenario.

Investment advice: We expect the company's revenue from 2024 to 2026 to be 1.344 billion yuan, 2.244 billion yuan, and 3.348 billion yuan, respectively, up 89.4%, 67.0%, and 49.2% year-on-year, maintaining the “gain” rating.

Risk warning: downstream demand falls short of expectations; implementation and commercialization of artificial intelligence technology falls short of expectations; industrial policy changes; macroeconomics fall short of expectations.