This study discusses the credit risk problems faced by Internet financial enterprises, and conducts the research on credit risk assessment and early warning of supply chain finance based on the econometric model. With the development of financial technology and fierce competition in the industry, Internet financial enterprises are faced with credit, information disclosure, supervision, compliance and other risks. From the perspective of credit risk management, this article constructs a credit risk warning indicator system that integrates financial and non-financial indicators. Principal component analysis and grey correlation method are used to optimize the indicators, and a convolutional neural network model is used for credit risk warning. By using principal component analysis and grey correlation method to optimize indicators, an indicator system consisting of 7 financial main factors and 9 non-financial indicators was established as input for the CNN model. Through the simulation application of 5-year data from 81 listed companies, the results show that the CNN model has a warning accuracy of 96.30%, which is better than other models. The comprehensive research results provide practical models and methods for credit risk management of Internet financial enterprises, which has important theoretical and practical significance.