Seriously, why does this thread exist when there's so much material to read on the internet already... It's like people just don't understand how electronics and IC's work. Extra heat build up will kill components (obviously), but the tolerance will range slightly on the quality of the GPU die itself (as far as I know, the GPU die isn't 1:1 perfect copy and can range in quality) and the quality of the components used by AIB partners for the PCB, by now there just hasn't been any of the AIB partners that skimp out on the PCB and it's components, IMO it's a thing of the past.
The cooling solution is the icing on the cake. Both companies have cards that can fail equally if you put them under more stress than what they're supposed to run at, end of story, mileage will always vary.
Though I will mention that sometimes AIB partners can **** up. Like when I had a Gigabyte R9 380 turn off it's fan until it hits 50c ON IDLE. That's just too high, why the hell! Eventually AMD fixed that with a driver, after 6 months... A bit late, but better then never at all. Gigabyte did have BIOS updates, but most of the new ones wouldn't even flash for some reason, and people complaining in the forums didn't get any answers. ON THEIR OWN FORUMS, WHAT!? Lol, Indian tech support is providing better service than them. I'm joking, but still, I haven't forgotten that they messed up that time. It's like they're becoming the next ASUS, where you won't get an answer until you post on their facebook posts, smh.
The only card I've ever had that was dying was a GTX 550Ti, after about 3-4 years it started showing pink dots on the desktop, but was fine when a 3D load was put on. It wasn't overclocked at all, it just had crappy cooling because it was a Palit model. The non existent airflow in the case at the time was killing the card faster, because the single fan on the graphics card didn't have any cool air to take in. I was surprised that the GPU die was showing symptoms first rather than the other components that ran at higher temperatures, even though they are rated to be operational at 100c and above.
A genuine old complaint from HD 7000 and RX 200 series mid to high range cards is that AMD AIB partners rarely accommodated for overclocking, some of the cards already ran with high temperatures as it is, and people overclocked on top of that, some guys just didn't have common sense maybe, I don't know.
I have never bought a reference design board with blower fans because they just wouldn't work as well as the open fan design (or whatever it's called) in my situation. I hav
Bottom line, OP's topic discussion shouldn't have been brought up... yet. Graphics cards come and go, it's rare for any gamer nowadays to be using the same card for more than 10 years, hell I wouldn't use the same card for more than 5 years in my case (they just outlive themselves based on the games I need to play). And the cards in my possession usually never die even after lots of torture.
I recently ghetto modded two low-end cards, a HD 3450 and a GT 8500, so they have more heat capacity )because the AIB partners just went nope! And had small heatsinks instead, and thought it was okay for the card to die after the warranty was up). I did this for fun mostly, but in the end I wanted the GPU's to survive longer, I will use them for an old Pentium 4 PC with Windows XP so I can run older games perfectly, since some games don't work right on newer OS versions. That's the only reason why I'd want the GPU to last this long, is for them to be able to play older titles with compatibility issues.
Sorry for the long winded post, I think I've already mentioned that it's impossible for me to make short posts since I want to just speak my mind. At least this time I didn't press the backspace by accident and lost all the text and over an hour of time typing it, because the auto-save system on this site hates me. Now if I can be excused, I'll go play some Morrowind on an overclocked integrated GPU.
I'm terribly attached to my Zotac GTX 580. 
LoL
