Customers care more and more about their privacy. People start to worry about their data and how big companies are using it. Decentralized Machine Learning helps conciliating the need for company to use customer data to stay competitive and the privacy awareness that their customers might have. This new paradigm uses data directly on the device and doesn’t need any data extraction. We will be reviewing how we can train accurate model without never having to get the data from outside the smartphone of a user and how it can power companies decisions and machine learning product. We will go through different concepts like Federated Learning or Secure Aggregation protocol. And we will talk about the pros and cons of using this method against having the data directly in a company server.