Every single time you work on a machine learning project you start to develop assets. These are either going to be considered valuable or it's possible they will not yield any real depreciable asset. This topic is like a flashback to an article in the Harvard Business Review from 2019.