With the continuous progress of technology and the continuous deepening of application, the intelligent financial sharing platform will surely create more outstanding value for the company and lead the enterprise to move in a more intelligent, efficient and sustainable direction. Based on the federated incremental learning method of financial information data sharing, this paper opens up a new way for the secure sharing of financial information. Moreover, this paper carefully constructs an efficient federated learning framework, which makes use of the characteristics of incremental learning, so that all participants can collaboratively update and optimize the financial information classification model without directly sharing the original data. In addition, through the federated averaging algorithm, the localized model parameters of each node are converged to the central server after secure transmission, and the server is responsible for integrating and processing these parameters to build a unified global model. Finally, this paper verifies the effectiveness of the system with experimental analysis, which can provide theoretical reference for the subsequent construction of enterprise financial sharing system.