That 'Majority Buying Up' stat is probably the most concerning; it pretty much validates the generational price rises between tiers for NVIDIA (for the consumers willing to pay, that is).
Its also super vague. What does it really say? It is absolutely not true everyone's 'buying up' -
only 2% did that for Turing. The 1660ti is only just out, I doubt it contains any of those. This is a slide for shareholders more than anyone else, trying to create the impression its business as usual. People are
not buying up. They are simply not buying.
So let's get into this; 50% of installed base is Pascal; 2% upgraded to Turing. That is 1% of the total installed base with a Turing RTX card, and that was over the last six months, or more if you count pre order time. For a new gen that's out over 2 years past the previous one, that is extremely worrying. I'm sure everyone's buying up. The everyone is however just a handful of people
And those 1080ti owners? The ones that did 'buy up' had to get sent two or three replacements to kill the space invaders... The rest shrugs and moves on.
What's really happening is very simple, Pascal high end users have zero incentive to buy into Turing, apart from promises and blurred DLSS 4K plus some god rays in one game and pools of water in another. You could see this coming from miles away, I saw it when Huang shouted 'ten gigarays' with nothing to show for it, trying his best Steve Jobs impersonation; everything was amazing and awesome. It was vague then and it still is today, there wasn't much content then and there isn't today. By the time this gen is somewhat adopted, 7nm is there waiting and Nvidia is there
very eager to move high margin units again to regain share value. They already gave up with Turing's high end, really, they probably will focus on volume in the lower half of the stack;
without RT. Say byebye to those adoption rates, and now you know why Nvidia tells us Turing sans RT can still do things 'a bit faster' than Pascal. They're
already selling it without the dedicated hardware
What that means is that 7nm for Nvidia is the actual start of RTX. Not Turing. They basically wasted a full gen for us and the contingency plan is in place.
That first pic, I'm curious what math is behind that because it totally doesn't mix with the other ones. OTOH early Pascal also had some supply issues and only launched with GP104, which, from a 980ti wasn't really worthwhile, and it launched with FE's that were objectively not interesting.
