Imagine we need to develop a recommendation system for Netflix. Initially, we can create a model utilizing a dataset consisting of thousands of movies and users. How would you go about expanding this model to handle the entire Netflix dataset, which encompasses millions of movies and users?