As an important area for large model applications, the financial industry is creating significant value through large model technology in actual business operations. Financial industry scenarios are complex and data-intensive, with each business link relying on in-depth mining and processing of information, especially in decision support. Financial large models leverage their natural advantages in handling massive data and complex calculations to play a key role in core financial business scenarios such as risk assessment, investment decisions, customer service, and marketing. They help financial institutions reduce costs and increase efficiency, improve operational efficiency and decision accuracy.
Based on research and analysis of China's financial big models in the first half of 2024, Frost & Sullivan (Frost & Sullivan, abbreviated as "Frost & Sullivan") jointly with LeadLeo Research Institute released the "China Financial Big Model Market Tracking Report, 2024H1". This report conducts an in-depth analysis of the market share and competitive landscape of China's financial big models, elaborates on the main participants under various business models and product delivery forms, and also delves into the core considerations for financial institutions deploying financial big models.
01
In 2023, the scale of the Chinese financial large model market was 1.593 billion yuan, and it is expected to grow to 13.179 billion yuan by 2028. In the first half of 2024, the market scale had reached 1.6 billion yuan, indicating that the financial large model market is rapidly expanding.
With the continuous progress of large model technology and the accelerated digital transformation of the financial industry, the Chinese financial large model market is experiencing rapid growth. It is expected that the market size will leap from 1.593 billion yuan in 2023 to 13.179 billion yuan in 2028. In the first half of 2024 alone, the market size of financial large models reached 1.6 billion yuan, and the annual scale is expected to achieve a significant growth of 140%. This growth is mainly due to the strong promotion of large financial institutions, especially banks, insurance companies, and securities firms. These institutions are gradually viewing financial large models as the core tool for digital and intelligent transformation, using AI technology to optimize businesses such as risk management, customer service, and intelligent investment advising. At the same time, the lightweight deployment characteristics of financial large models enable small and medium-sized fintech institutions to widely adopt them, further driving market expansion.
02
In the first half of 2024, in the Chinese financial big model market, MaaS deployment accounted for 52% of the market share, leading the large-scale application of small and medium-sized institutions. Private deployment accounted for 48%, becoming the preferred choice for large financial institutions.
In the first half of 2024, among the deployment models of financial big models in China, the Maas model has significantly reduced the technical threshold and initial investment pressure for small and medium-sized internet financial institutions due to its 'out-of-the-box and pay-as-you-go' features. At the same time, private deployment has become the preferred choice for large financial institutions due to its advantages in data security, compliance, and deep customization. In the coming years, MaaS will continue to dominate, continuously meeting the needs of financial enterprises for cost reduction and efficiency improvement.
03
In the implementation and application of financial large models, standardized products account for about 60% of the market share. In the next few years, the market share of standardized products will continue to grow, and it is expected that by 2026, it will exceed 70%.
The deployment cycle of customized solutions typically ranges from 3 to 6 months, with investment amounts ranging from several million yuan to tens of millions yuan; whereas standardized products can be quickly deployed within 4 to 6 weeks by integrating modular architecture with pre-trained parameter solutions. These products rely on mature large model platforms, providing strong computing power, resource scheduling, model training, and inference capabilities, ensuring efficient execution efficiency. In the next few years, standard products will leverage their ability to promote the efficient integration of cloud technology and private environments, helping customers achieve overall architecture upgrades and iterations, as well as providing flexible support capabilities to cope with future technological changes and changing business needs, thereby driving continuous growth in their market share.
04
Financial large models have significantly empowered front- and middle-office scenarios such as customer service and data analysis. However, further deepening is still needed in complex financial decision-making areas.
Currently, financial large models mainly optimize core business scenarios such as marketing, credit, and customer service by deeply integrating with manufacturers' existing databases and big data platforms, thereby significantly enhancing the customer experience. At the same time, in the middle and back-office, large models effectively improve business process efficiency and decision response speed through automated data processing and analysis capabilities, further promoting operational efficiency. However, for highly complex financial decision-making scenarios such as portfolio optimization and derivative pricing, the intelligence level of financial large models still needs to be further deepened and enhanced. In the future, more financial large models will rely on native data middle platforms and intelligent application entities to provide more precise support in highly complex business decision-making scenarios, thereby further driving the intelligent upgrade of the financial industry.
05
After the battle of hundreds of models in the era of negative gross profit margins, in the next three years, China will form a pattern of "3-5 closed-source giants + 1-2 open-source platforms". At the same time, leading enterprises will implement a dual-track strategy of open-source and closed-source.
In the next three years, the open-source model market is expected to see only 1-2 leading enterprises that can support long-term investments with cash flow from "non-large model businesses." These companies will strategically choose open source to unlock technological dividends and cultivate a developer ecosystem; at the same time, they will use closed-source models to protect core technologies and achieve high-value monetization. Due to the high maintenance costs of open-source models, which require tens of thousands of computing cores and continuous iteration, these leading enterprises will introduce a "limited open source" model (such as open source of some modules + core closed-source) or adopt the "open source as a service" model (open-source models must be deployed on their own cloud platforms), to achieve a two-way drive between business and ecosystem, thereby blurring the boundaries between closed-source and open-source.
In the wave of digital and intelligent transformation of financial institutions, enterprises such as Alibaba Cloud, Baidu Smart Cloud, and Huawei Cloud have dominated the financial large model market with their profound technical accumulation, precise grasp of the industry, and rich project implementation experience. These leading financial large model enterprises not only delve deeply into their respective technical fields but also drive their business with strong technological advantages to precisely meet market demands, injecting powerful momentum into the intelligentization and digitization process of the entire industry.
