In the Industry 4.0 scenario, early fault detection has gained a relevant importance. In this new digital context, Internet of Things (IoT) and machine learning have become fundamental. By collecting data from industrial processes and by analysing those, it’s possible to statistically detect, forecast and prevent faults. In this way, it’s possible to have a more secure working place and a more efficient working process in terms of times and costs. This approach has been widely applied to all kinds of processes, with attention to automatic ones. For this reason, a lot of approaches have been studied to early detect faults in the metal Additive Manufacturing (AM) technique.