Diffusion models have revolutionized artificial intelligence with their ability to generate unprecedentedly realistic imagery. This chapter explores the application of diffusion models for generating histopathological images, and their potential to address issues with limited high-quality and annotated pathology datasets. It highlights how diffusion models, with their iterative denoising process, create realistic and diverse synthetic images that can be conditioned on a semantic mask of nuclei locations. These models improve the diversity of the data set, support robust training for diagnostic algorithms, and mitigate the need for extensive annotated medical data. By examining foundational principles and recent advances, this chapter demonstrates the potential of diffusion models to improve diagnostic accuracy, assist pathologists, and transform the field of histopathology. Additionally, the chapter introduces a first-of-its-kind nuclei-aware semantic tissue generation framework (NASDM) which can synthesize realistic tissue samples given a semantic instance mask of up to six different nuclei types, enabling pixel-perfect nuclei localization in generated samples.