This work presents a crosstalk-aware timing estimation method using a two-step machine learning approach to improve accuracy and efficiency in predicting crosstalk-induced delay during routing.