The growing global concerns regarding climate change, natural resource depletion, and environmental risks have made green investments one of the fundamental pillars of modern economic and financial policymaking. However, accurately evaluating the sustainability and economic effectiveness of green projects remains a significant challenge. Traditional investment evaluation methods mainly focus on short-term financial indicators and have limited capability in simultaneously analyzing economic, environmental, and social variables. This study aims to develop an AI-based sustainable finance framework for evaluating green investments. In this research, machine learning algorithms, multidimensional data analysis, and intelligent forecasting models were integrated to design a sustainable evaluation system. The dataset consists of information from 1,250 green investment projects in renewable energy, clean transportation, smart buildings, and resource management sectors during the period 2018–2025, collected from international financial and environmental databases. After preprocessing, feature extraction, and data normalization, the proposed model based on deep neural networks and XGBoost was developed and compared with traditional financial evaluation methods. The results indicate that the proposed framework outperforms conventional models in predicting financial sustainability, assessing environmental risks, and analyzing long-term investment returns. Furthermore, the AI model demonstrated a superior capability in identifying complex relationships among ESG indicators, economic returns, and investment risks. The findings suggest that AI-based frameworks can play a critical role in developing sustainable financial systems, improving investors’ decision-making processes, and directing financial resources toward environmentally sustainable projects.