Nevermind energy consumption, floor space, maintenance, inconsistency in quality, loading-unloading times, postprocessing / cleaning, etc.
3D printing is not really scalable.
Folks that suggest we can just buy 10,000 3D printers to replace 8 injection molding machines haven't really stepped into a high volume factory, much less operate one.
Energy consumption of other manufacturing methods is also quite high, for instance, injection molding is super energy intensive, as are most forms of casting and machining.
It's clear that 3D printing is no substitute for injection molding, stamping or or other such mass fabrication methods. But, like with every other manufacturing method: it is complementary, there are applications for all of these.
And yes, I've stepped into factories, and worked with/for them, beside that I had a metal workshop and was partner in an early CAD/CAM company. So FWIW I think I understand the limitations of the tech quite well, and I think that if 3D printing could be sped up by a factor of 100 or so (which is definitely in the realm of the possible) that the other disadvantages would matter a lot less than they do today.
Injection molding can't work with many geometries; for example, complex enough boat has a lot of places where the material has to go along a non-convex paths. 3D printers can print anything, say DMLS, including closed volumes (how to remove material from those volumes is another question). Having many heads the process can be sped up - but with DMLS it's mostly moving mirrors which needs to be parallelized, as laser beams have little inertia.
Comparing with injection molding it seems IM is a good shortcut for many - not all - cases, while 3D printing is a fundamentally generic technology.
