Foundation models are artificial intelligence (AI) systems based on a general class of models that are also termed as pre-trained models. These models are large neural network architectures and trained on a huge amount of textual data to learn the patterns and structures of language. These models construct the base for several natural language processing (NLP) tasks and applications. Foundation models can be fine-tuned on a broad range of downstream tasks such as GPT-3, CLIP and BERT. After training of foundation models, they can be fine-tuned on particular tasks such as language generation, text classification, translation, question answering and summarization etc. Although foundation models are not new but still need to explore with their new opportunities and challenges, the base of these models are self-supervised learning and deep neural networks. This chapter will explore the task and applications along with the role of foundation models in different areas.