The primary object of Aristotle’s metaphysics is the so-called ‘first immovable mover’. The hypothesis of its existence is the conclusion of a long argument, set out in Physics 8, that is based on physical data and principles. This raises the following question: how can one argue for something non-physical starting from physical premises? I suggest that the main reason why this argument leads outside the domain of physics is that it involves inferences from limited to unlimited changes (which comprise, in turn, an extreme case analysis and a concept stretching procedure) and an essentialist notion of physical entities.