Federated Learning is a generic machine learning approach whose goal is to train high-quality models with data distributed over several independent providers in a privacy-preserving fashion. Instead of gathering the data on a single central server, the data remains locked on the data providers, while the algorithms and predictive models are exchanged between them. By overcoming the data sharing bottleneck, this novel paradigm can help machine learning reach its full potential notably in healthcare and drug development, where data is particularly sensitive. Adoption of federated learning is indeed expected to lead to models trained on datasets of unprecedented size, hence having a catalytic impact towards precision/personalized medicine. Pilot projects on FL include pharmaceutical companies that share their valuable data with competitors for the purpose of drug discovery, as well as Oncology Centers across Europe for applications in clinical research.