Haitong: AI applications entering the acceleration period from 1 to 100, paying attention to cloud service providers, etc.

Haitong released a research report stating that looking back at the mobile internet boom in the A-share market from 2010 to 2015, there were opportunities for hardware equipment investments in the early stages, and during the acceleration period, there was a proliferation of content and applications. By the end of 2022, the emergence of ChatGPT may mark the first year of AI applications, and from 2023 to 2024, domestic and international giants will increase their investment in AI algorithm development. The computing power hardware in the A-share market has seen a significant increase. The DeepSeek trend represents the entry of AI applications into the 1-100 acceleration phase, and domestically, there may be an advantage in the application stage. It is advisable to focus on cloud service providers and AI applications, as there is still certainty in the medium-term computing power. Looking back at 2010-2015: After the supply chain comes the broad rise in application scenarios. During the mobile internet boom from 2010 to 2015, the improvement of communication infrastructure and the emergence of popular products initiated the trend of smartphones replacing feature phones. In the stock market, micro financial data confirmed the industry trend, and the market trend followed along the Apple industrial chain, with investment opportunities based on various hardware devices appearing. However, Chinese companies were positioned in the lower end of the smartphone value chain. In the acceleration period, content and application sides in China flourished. From 2013 to 2015, the mobile internet boom entered a period of outbreak, with the commercialization of 4G networks and the trend of affordable smartphones driving the rapid increase in smartphone penetration. China's large user base and diverse application scenarios led to a flourishing landscape in the content and application areas of mobile internet. In the stock market, domestic companies had a more evident advantage in the content and application fields, with these sectors showing greater elasticity compared to the hardware sector in the 2010-2012 period. From 2023 to 2024: In the 0-1 stage of the AI boom, computing power has seen a significant increase. The emergence of the phenomenon-level product ChatGPT at the end of 2022 marked a breakthrough or important turning point in the application of artificial intelligence technology from specialized fields to general scenarios, likened to the iPhone moment in the AI field. Computing power is crucial for AI development, and to build large-scale AI computing clusters, global tech giants have increased their capital expenditure on AI computing infrastructure for training and deploying advanced artificial intelligence. The AI wave in the US stock market has been reflected in the A-share market. Similar to the early stages of the mobile internet boom from 2010 to 2015, in the 0-1 stage of the AI application wave from 2023 to 2024, hardware performance has higher certainty, and stock price gains are more apparent. The surge in overseas computing power leaders has led to an increase in sub-segments such as optical modules in the A-share market. Although the capital expenditure expectations of tech giants still support the performance of AI computing power, the price gains in related areas have been very noticeable, and the path to expanding AI applications into super applications is still long. Since October 2024, the stock price volatility of the computing power sector, represented by NVIDIA, has further increased. Outlook: AI applications represent the vast ocean from 1 to 100. At the beginning of 2025, the domestically developed DeepSeek large model has attracted widespread attention at home and abroad. The low-cost, high-performance features of DeepSeek accelerate the democratization of AI and may accelerate AI applications. Drawing insights from 2010-2015, Haitong believes that cost reduction after technological breakthroughs is the inevitable path for the large-scale commercialization of new technologies and products, and in the application field, China's vast market and user base provide fertile soil for application innovation, with domestic companies having great potential. The low-cost, high-performance, and open-source characteristics of DeepSeek are expected to drive the acceleration and iteration of AI large models, speeding up the commercialization of AI applications. Looking ahead, cloud service providers will benefit from the growth in demand for computing power and cloud services brought by open-source models, followed by consumer electronics, humanoid robotics (Siasun Robot & Automation), AI healthcare, AI entertainment, and other fields that will benefit from the large-scale commercialization of AI applications. Furthermore, in the medium term, as AI applications are commercially deployed on a large scale, the demand for computing power remains relatively high.