CITIC Securities: Multiple financial institutions have completed the privatization deployment of DeepSeek, optimistic about the development and investment opportunities of financial IT vendors

According to the Zhitong Finance APP, CITIC Securities Company released a research report stating that finance is an important direction for the implementation of generative AI. Since 2023, with Bloomberg GPT leading the way, the global exploration of AI + finance has begun. Domestically, on one hand, financial institutions are bidding for large models for internal use, with several institutions recently announcing the completion of DeepSeek's privatization deployment, accelerating the popularization of AI applications. On the other hand, financial IT vendors are developing large models for external services, with investment research and advisory scenarios transitioning from Copilot to Agent forms. There are optimistic views on the development and investment opportunities for financial IT vendors empowered by AI.

CITIC Securities Company's main viewpoints are as follows:

AI + Finance: The curtain rises, and scenarios deepen

At the beginning of 2023, Bloomberg released BloombergGPT, which initially showcased the application opportunities of generative AI in financial scenarios and highlighted the necessity of developing vertical large models for finance. AI functions such as conference call summaries and news summaries are gradually reshaping the Bloomberg terminal.

Domestically, financial institutions are accelerating the implementation of large models. According to statistics from the WeChat public account "Intelligent Hyperparameters," there are 133 large model bidding projects initiated by financial institutions in 2024, primarily for internal applications. Baidu and iFLYTEK lead in the number of bids won. Recently, several financial institutions have stated that they have deployed the DeepSeek large model, marking an acceleration in the implementation of AI applications. On the other hand, financial IT vendors are developing financial large models based on corpus accumulation and scenario positioning, emphasizing external applications. The two core scenarios of investment research and advisory are gradually becoming clearer, with more financial vertical characteristics and imaginative application opportunities.

Investment Research Applications: Investment Agents are expected to initiate innovations in commercial forms

The investment research scenario targets institutional investors, with financial data terminals as the carrier. Taking Tonghuashun iFinD as an example, it combines its self-developed HithinkGPT large model to build the AiFinD module, providing AI capabilities such as dialogue, writing, drawing, minutes, search, and translation. Based on the transformation of interaction methods and enhanced product functions, combined with cost-performance advantages, it is expected to strengthen its market share in financial data terminals.

Regarding Agents, Tonghuashun will build the Tongchuang Intelligent Agent platform in 2024, becoming the first financial vertical Agent platform in China to support one million daily active users. It offers over 1,000 Agents, 300+ plugins, and 3,000+ workflows for selection, which can be deeply embedded in institutional investors' investment research workflows. Referring to overseas B-end Agent applications, we believe that in the future, there may be a charging model based on usage, enriching the commercialization forms of financial data terminals.

Advisory Applications: Investment Agents help enhance platform user stickiness

Investment Agents target individual investors, with financial information service products as the carrier. Financial information service vendors offer basic products for free, while charging for specialized data and decision support functions. Taking Tonghuashun as an example, it provides a series of AI value-added services. The Qancai system, combined with the Hithink GPT large model, has upgraded in areas such as dialogue and stock selection, serving 5 million investors daily and recently enhancing its deep thinking capabilities. Jiufang Zhituo, based on the Jiuzhang large model, has launched digital humans Since 2024, Tonghuashun and Jiufang Zhitu have innovated around Agents for AI products, providing a rich selection of intelligent agents and enhancing the experience of interacting with individual investors. In the future, investment Agents are expected to strengthen the stickiness of financial information service platforms and support long-term commercialization.

AI Foundation: Xinchuang provides a foundation for domestic AI + finance implementation

Financial Xinchuang aims to achieve independent control of basic software and hardware in the financial industry. Since 2020, financial Xinchuang has undergone three phases of pilot projects, penetrating from office systems to general business systems and core business systems.

In the process of implementing AI + finance applications, financial Xinchuang is expected to further promote the construction of a foundation for domestic AI applications and strengthen financial IT investment. The banking industry has taken the lead in proposing to eliminate "IOE," with major banks taking the lead in Xinchuang in recent years. The proportion of IT investment to revenue has increased during this process, and the migration of core systems to distributed architecture and single-track operation has been basically completed. Joint-stock banks and small and medium-sized banks are in the initial stages of Xinchuang, with further progress expected in the future. In the securities industry, the core trading system's Xinchuang is driving the entire industry's Xinchuang, with the technical route of distributed + in-memory computing gradually becoming clearer.

Investment Strategy:

Since the end of September 2024, with the reversal of capital market expectations, the trading activity of A-shares has significantly warmed up, with daily transaction amounts currently stabilizing at the trillion yuan level. Combining market β with the α brought by AI, we are optimistic about investment opportunities for financial IT vendors.

Risk Factors: Risks of intensified competition in the financial information service industry; risks of licensed operation of securities trading information; risks of technological updates; risks of AI applications not meeting expectations; risks of Xinchuang policy advancement not meeting expectations