YouTube channel 修电脑的张哥 (roughly “Zhang the Computer Repairman”) posted a video this morning citing the sheer volume of Nvidia GPUs it has received for repair in just the past two days. In every case the damage stems from burnt interfaces between the 12+4-pin power cable and the card itself.
Seven cards have come in for repair (via Uniko’s Hardware), mostly RTX 5090 units, along with Asus and MSI RTX 4090 models. They arrive from a range of manufacturers: alongside Asus and MSI there is also a Gainward card and a Founders Edition RTX 5090.
In the video, either the socket interface is burned—sometimes blackening the PCB—or the 12+4-pin power connector itself has melted into the socket, forcing users to cut the cable off before they can send the card in for testing.
Latest Videos From
PC Gamer
Zhang carries out the necessary repairs on each card—removing the interface, cleaning the contacts, and soldering on a new 12+4-pin socket—then tests every one. With the new sockets fitted, the repaired cards are back in action, and while the GPUs run through some 3DMark benchmarks he checks the temperatures of both the cards and the cables.
Some still run a little hot, with the cables at the PSU also running pretty toasty, but nothing particularly worrying at this point.
Still, the fact remains that this is a very high volume of cards arriving at a single repair shop, all showing the same symptoms and all from different manufacturers. He also notes the users are not necessarily running weak PSUs, and some cards even come from lab computers (though those may be pushed harder than a card used for gaming at home).
It may be that this is a very popular repair channel on BiliBili and YouTube, so he might receive more cards than other shops, but it’s obviously still an issue. It feels like, over a long enough timeline, a whole lot of RTX graphics cards using a 12+4-pin connector are going to suffer from this problem.
(Image credit: YouTube channel 修电脑的张哥)
One thing Zhang notes is that during use, if you touch the back of your finger against the power cable—maybe an inch or so above the GPU connector—you should feel it being a little warm. But if it’s notably too hot to touch, then there’s likely an issue.
He likewise suggests keeping an eye on power draw via GPU-Z, and particularly the 12+4-pin voltage readout. With failing cards, he notes that if the 12+4-pin voltage drops well below 12 V—to 11.3 or 11.4—then there’s likely an issue and you should check the temperature by hand at that point.
What’s clear, though, is that this issue isn’t going away. It isn’t purely down to dodgy vendor adaptors or old PSUs, or even misaligned cables. The amount of power being drawn by these beefy cards is melting cables and connectors, and if you have one of these GPUs it’s worth monitoring it regularly.
That said, if you want a little more peace of mind, there are new gadgets that can offer some. Not complete protection, but some forewarning of an issue before it becomes a melty mess—such as this new OCP dongle from Super Flower.
Best graphics card 2026
1. Best overall: AMD Radeon RX 9070
2. Best value: AMD Radeon RX 9060 XT 16 GB
3. Best budget: Nvidia RTX 5050
4. Best mid-range: AMD Radeon RX 9070 XT
5. Best high-end: Nvidia GeForce RTX 5090
Have thoughts, questions, or benchmarks on this topic? Join the discussion on our forums.
Join Discussion ➔
Únete a la conversación