It is not uncommon that industrial use of simulation lags years behind academic development of a new simulation capability. In the field of metal additive manufacturing, however, industrial adoption of certain simulation tools was exceptionally rapid. This rapid adoption provided a market for a considerable number of new software developments and –startups. AM simulation has certain characteristics that make it eminently suited for highly automated simulation workflows (e.g. fully automated finite element meshing by voxel approximation can be sufficient for AM distortion prediction) and many of the new software tools focused on user-friendliness and automation of the simulation workflow. It seems, however, that we have reached the limits of this: highly automated approaches entail the risk of not predicting key AM defects (e.g. buckling) and runtimes are sometimes unacceptably long for industrial use (e.g. microstructure and property prediction for an industrial part)