The urgent need for the prediction of electricity demand triggered by the deep adoption of electric vehicles (EVs) contributes to the reliable electricity supply of the grid. However, an underestimated grid load would occur for peak and off-peak differences in the overwhelming EV charging context. From a behavior perspective, we aim to explore the long-term impact of EV charging on the underestimated load effect with charging behavior modification under climate goals, since the adopter's charging time and EV initial state of charge (SOC) are vital to shaping grid peak profile. Here, we propose a combined top-down and bottom-up integrated assessment model to predict the implication of EV charge on grid underestimated power supply gap under climate goals. Notes that the forward-looking behaviors are constructed through a data-driven mixture probability framework developed by EV minutely charging and driving data in Beijing. Our results indicate that under China's carbon-neutrality goals, the underestimated load increases by 15.8% relative to no climate target limitations, while guiding EV adopters to charge their vehicles in the off-peak could reduce 41 GW peak electricity demand, and thus curtail the grid load fluctuation 90 GW, 18% more than the overall coal-fired capacity increased forthcoming in the European Union and the United Kingdom combined. Coupled charging behavior guidance and Vehicle-to-grid integration are more productive. Our findings open avenues for enhancing future electricity supply security concerning deep vehicle electrification by charging behavior guidance using behaviorally realistic data.