Investment Rating - The report suggests a focus on the company with a dynamic PE of approximately 10 times for 2025, indicating a significant advantage compared to peers [3]. Core Insights - The integration of the DeepSeek model into the company's existing AI framework enhances performance across various applications, including natural language understanding, SQL generation, and chart interpretation [3]. - DeepSeek's low inference cost allows the company to reduce computational requirements during model training and inference, leading to cost savings and improved project profit margins [3]. - The transition from "AI integration" to "AI native" signifies the company's rapid internalization of large models into its product and technical capabilities, fostering a broader AI ecosystem and promoting collaboration and innovation among upstream and downstream enterprises [3]. Summary by Sections Event - The DeepSeek model was released on January 20, showcasing significant advancements in performance for complex tasks, while also reducing training and inference costs [1]. - The company's AI applications, such as ChatBI and intelligent document processing, can now seamlessly switch to the DeepSeek model, enhancing their capabilities [1]. Technical Performance - The integration of DeepSeek-R1 has led to a 20% improvement in content generation accuracy compared to other mainstream open-source models, with a result usability rate of 95% [1][3]. - The enhanced reasoning capabilities of DeepSeek provide deeper insights and more reliable decision support for users [1]. Cost Efficiency - The low inference cost of DeepSeek is expected to lower the computational demands of the company's models, directly benefiting profit margins [3].
浪潮数字企业：浪潮海岳接入DeepSeek，实现AI技术新高度