For now it is too early to say what the source of the problem is, but we know that overheating has occurred on Asus and Gigabyte cards. Like the rest of the RTX 4000 lineup, these GPUs use the new 16-pin 12VHPWR connector for ATX 3.0 sources; a choice that has come under some criticism for allowing alignment and/or contact problems if the cables are bent closer than 35mm from the connector (CableMod dixit). And given the size of modern graphics cards, it’s sometimes hard to keep cables from being overly bent and/or cramped.
Added to this is the fact that some implementations of the 12VHPWR connector may exhibit “thermal divergences” that would result in “security problems under certain conditions”, as stated by the PCI-SIG itself, the body responsible for its standardization. Burnt cables and connectors like those exhibited by Reddit users would be one of the consequences.
For its part, the popular youtuber specialized in hardware JayzTwoCents, which has been warning of deficiencies in the 12VHPWR standard for some time, affirms that security problems are intrinsic in its design, and ensures that these failures could also occur in future AMD Radeon cards (not without wondering with some sarcasm why why Intel, which is part of PCI-SIG, chose not to use such a connector when designing the ARC family).
Be that as it may, Nvidia claims to be investigating the facts. As he has been able to find out TheVerge Through a spokesperson, the company is contacting affected users to carry out the necessary investigations. Considering that the entire RTX 4000 family uses the same power connector, we will have to be very attentive to the results of the investigation.