go watch the video posted above to understand what's going on..it's all about cheaping out on components , poorly designed boards + OC done by third party manufacturers, the nvidia's own FE is having ZERO issues reported.
You answered your own question, go watch the video yourself, you need to do this.
Looks like I do have to explain for you.
Ok, so, when Nvidia send out the PCBS they have to, well NEED to send out a minimum requirements list to meet the demand of the GPU chip, clealrly nvidias minimum is not enough, which has allowed all these AIBS to run these chip parts, now if NVIDIA, yes that's right NVIDIA I said, had tested properly and extensively they would know the minimum requirement is TOO LOW, as we can see from just the general public, so im sure if the general public are getting problems why the hell could Nvidia not.
Now it's in nvidias best interest to get the "MINIMUM REQUIRED SPEC" right for their own good, afterall NVIDIAs name is on ALL the boxes, and it's their GPU chip.
Now if Nvidia had sent out a revised more moderate minimum spec, say at least 2 arrays of MLCC, or minimum of 1, rather than NONE (which in this case is SHOCKING as we can see) this would not be a problem right now.
So in answer to your original statement of it's not nvidias fault. IT IS VERY MUCH NVIDIAS Fault (AIBS never had enough time to test themselves in all this rush for stable flow, so it's down to nvidia to get the minimum requirements right) All this is explained also in Jays latest video.