By the way, any card running at a high temperature, 24/7, is bound to be subjected to more stress. This matters because high heat (when combined with high current draw) is known to be a cause of silicon degradation through a process called electromigration.
The GPU Core, uncore, VRAM modules, driver ICs, MOSFETs, and various controller ICs are all made of silicon and thus could be subjected to degradation severe enough to cause failures.
GPUs also contain electrolytic capacitors, and those tend to be rated for X number of hours at 50C, with X decreasing as the temperature surpasses 50C.
Not all mining GPUs are bad though, since it depends on the seller/user who handled it, and also on the time of usage of the GPU in mining, under a certain temperature and climate.