This study investigates the factors influencing Mainland Chinese students' satisfaction with AI-based chatbots and their academic performance in Malaysian universities. By integrating the Technology Acceptance Model (TAM), Social Cognitive Theory (SCT), and Expectancy-Value Theory (EVT), the research examines the roles of perceived risk, perceived enjoyment, trust, emotional value, internet addiction, reuse intention, satisfaction, and AI self-efficacy. A cross-sectional survey was conducted among 400 Mainland Chinese students using stratified random sampling. Data analysis using Partial Least Squares Structural Equation Modeling (PLS-SEM) reveals that perceived risk negatively influences satisfaction, while perceived enjoyment and trust positively affect reuse intention. Emotional value indirectly enhances academic performance through reuse intention, and AI self-efficacy moderates the relationships between reuse intention, internet addiction, and academic performance. The findings contribute to theoretical frameworks by expanding TAM to include emotional and trust-related factors, while also offering practical implications for improving AI-based educational tools in higher education settings. Future research should explore additional mediators and moderators to deepen the understanding of AI chatbot adoption and its impact on academic outcomes.