Drawing from a comprehensive and unique dataset encompassing both quantitative and qualitative weather risk measures, the study finds that both numerical and textual repre- sentations of weather risk can predict future credit risk, expected stock returns, and firm fundamentals. To explore the textual dimension of weather risk, this paper utilizes ad- vanced natural language processing (NLP) techniques, including Term Frequency-Inverse Document Frequency (TF-IDF), Word2Vec, and leverages Large Language Model (LLM) such as BERT (Bidirectional Encoder Representations from Transformers). To conduct the empirical analysis, this study utilizes Artificial Intelligence (AI) using TensorFlow/Keras, Deep Learning (DL), and Machine Learning (ML).