Brokers Morning Meeting Highlights | The most certain areas to land by 2025 are AI cloud services and high-quality data.

Yesterday, the market fluctuated and rose slightly, with the three major indexes rising. The total turnover of the Shanghai and Shenzhen stock markets was 1.73 trillion yuan for the whole day, a decrease of 229.4 billion yuan from the previous trading day. In terms of market trends, the hotspots revolved around AI, with more stocks rising than falling. Over 4100 individual stocks in the market rose, with over 170 stocks hitting the daily limit or rising more than 10%. In terms of sectors, medical information technology, DeepSeek concept, computing power leasing, Sora concept and other sectors led the gains, while banking, photovoltaic equipment, and port sectors fell the most. As of the close of yesterday, the Shanghai Composite Index rose by 0.56%, the Shenzhen Component Index rose by 0.52%, and the Growth Enterprise Index rose by 0.44%. At today's securities morning meeting, Sinolink predicts that the strongest certainty in the 25-year deployment is AI cloud services and high-quality data; CICC points out that DeepSeek technology breakthroughs, cost reductions lead to widespread applications; and Open Source Securities indicates that AI applications may accelerate deployment and drive demand for inference computing power. Sinolink: Predicts that the strongest certainty in the 25-year deployment is AI cloud services and high-quality data Sinolink states that in terms of investment opportunities in the artificial intelligence industry chain, the logic of insufficient supply brought about by large investments in the upstream computing power side under the background of the FOM giants in the past 2 years is weakening, and the focus is shifting from training to inference computing power demand after application landing, with broader demand and more open competitive landscape. In addition to the AI investment by domestic giants driving the upstream, it has followed the path of the United States over the past two years, which is the most certain for AI applications. In terms of predicting more landing expectations in 25 years combined with DS's innovative effects, the most certain aspect is the AI cloud services and high-quality data sections. Regardless of which direction of application exploration succeeds, it will have a continuous demand-driving effect on these two elements. CICC: DeepSeek technology breakthroughs, cost reductions lead to widespread applications CICC states that DeepSeek went global at the end of January 2025, with a DAU of 22.15 million on the APP side, ranking second only to ChatGPT in the AI product daily active users ranking, and ranking first in the download rankings in the Apple App Store in 157 countries and regions. CICC believes that the technological innovation and outstanding engineering capabilities behind DeepSeek going global lead the global technology trend. Its cost reduction achievements have important implications for the deployment of end-cloud applications, and suggests focusing on investment opportunities in application layer under the background of domestic model performance optimization in 2025. Open Source Securities: AI applications may accelerate deployment and drive demand for inference computing power Open Source Securities points out that several companies, including CHINA LIT and Zhewen Interactive Group, have deployed DeepSeek-R1 locally. The DeepSeek-R1/V3 model, based on algorithm optimization, has achieved high performance and low cost advantages, and its open source nature has helped its user base grow rapidly in a short period of time, driving many enterprises to complete the localization deployment of models. This is expected to accelerate the landing of social, educational, e-commerce, and office applications and drive demand for inference computing power on the end side. In the short term, DeepSeek's continuous release of V3, R1, JanusPro, and acceleration of model competitions have also led to the release of high-performance large models by companies such as Ali Tongyi Qwen2.5, with the continuous upgrade of multi-modal capabilities of models driving the growth of demand for multi-modal language corpus data. It is recommended to continue to focus on AI for further layout. This article is a reprint from "Cailianshe", GMTEight editor: Xu Wenqiang.