This is an interesting story, apparently AMD secured the luxury of getting exclusive access to Hynix HBM2 graphics memory. The second iteration of High bandwith memory (stacked memory to be used on-d... Rumor: AMD gets priority advantage to Hynix HBM2
Lets be real here. HBM1 is in short supply and there is only Fiji using it. It has only 4 slices. So, everyone here expects HBM2 to be in plentiful supply and AMD to be bad guy preventing all those companies standing in queue from getting their hands on that? It will be in short supply again, so AMD wanting to have it on GPU and APU is enough to be limited by its availability. So let them share with everyone, let nVidia make 4 different GPUs, AMD 3 GPUs and 4 APUs. And then lets fight oner those 400 graphics cards nVidia gets to make per week. But if that is 100 cards per type per week, then those would have to be very expensive cards to guarantee return of investment into design. At this point AMD wanting it everywhere may be idea I like, but if supply is not high enough then they'll not going to get return of investment in design. I do not know how many Fury(x) are made per week, but according to shops, it is not enough and it seems to be HBM limitation. You have seen that AMD had working "nano" and Fury X2 back then. But you can't buy them because they can't supply them in volume.