EVGA confirmed yesterday the cap issue is real, it caused them to change their production run.
However others are saying the crashes still happen even with FE cards and Evga cards. So while the cap issue may be real, it's an issue that may not be the full picture.
Igor's lab was wondering before if the PSU load may be to much for some power supplies as the huge transient spikes may be tripping protection features in some overly sensitive psu's and causing the crashes. LTT was saying the 850w platinum seasonic they were using in their test bench kept having the OCP trip when running the card and had to switch it out.
There may be other design choices nvidia/aib makers choose on the pcb that are also causing issues, or the gpu dies could have problems, or the super hot memory. who knows.