Manual evaluation of investment risk make results and solutions are not timely. The objective of the study is to explore intelligent risk data collecting and risk early warning of international rail construction. First, this study has identified risk variables by content mining. Second, risk thresholds are calculated by the quantile method based on data from 2010 to A.D. 2019. Third, this study has developed risk early warning system by the gray system theory model, the matter-element extension method and the entropy weight method. Fourth, the risk early warning system is verified using Nigeria coastal railway project in Abuja. This study found that: (1) the framework of the developed risk warning system contains a software and hardware infrastructure layer, a data collection layer, an application support layer, and an application layer. (2) 37 investment risk variables are recognized; (3) 12 risk variables thresholds intervals are not equally divided between 0 and 1, the others are evenly distributed; (4) based on the application of Nigeria coastal railway project in Abuja, the system verification results are consistent with real situations, which is shown that risk early warning system is reasonable and feasible. These findings offer a good reference for intelligent risk management.