Quite apologetic
iPhone users do not exist if there is no iPhone apps, so they are
economically the most important objective, but
strategically the developers are the most important user base.
Machine Learning is quite CPU/GPU intensive, the iMac Pro its not enough, serious ML developers are running rigs from 4 GPU to 16 GPUs, and not just because they want to look Macho with as many gpu, in ML training is a huge cpu time consuming part of the development effort, GPUs do it the best (until the new Google Tensor training processor gets out Google's servers rooms), but a single Vega 64 is not rival to Quad-GTX1080ti (the most popular ML setup out there, ask
@AidenShaw about he knows a thing or two about).
the point is for as long as 2 years the Mac Pro needed an update that wasn't even on the cards, WHY THIS HAPPENED?
Ok the TC-MP was a Mistake, why they didn't notice that soon enough (because they where on other duties not caring the Pro needs)