This is worth a calm look rather than a panic. The reports are real and they span several GPU generations. They are also anecdotal, and nobody can say from them how common the problem is or what caused each case.
What the November reports actually show
VideoCardz built its roundup from the X account UNIKO's Hardware. That account regularly collects reports of damaged GPU power connectors from Reddit, Chinese forums and other communities. So this is a secondary compilation of user photos, not a survey or a lab investigation.
The reported damage varies:
- Connector location: some cases are on the GPU side, some are melted adapters, and in the Thermaltake Toughpower Grand case the damage was on the modular connector at the power supply end, not on the card's 16-pin socket.
- Severity: VideoCardz describes complete connector deformation in some cases, and discolored contacts or localized burn damage in others.
- Cards named: RTX 5090 D, RTX 4090 (several reports), RTX 5080, RTX 5070 Ti, RTX 5070, plus a few unspecified GeForce cards.
The roundup is open about its limits. In its words, more than ten reports surfaced, but the incidents do not establish how frequently such failures occur. It also says nobody knows how many similar incidents go unreported, how many involve installation problems, or how many come from defective components.
One more point on scope. A Taiwanese outlet, King of Computer Media, notes that a separate RTX 5060 Ti case involved the traditional 8-pin connector, not the 16-pin. That is a reminder that not every burnt cable is a 12V-2x6 story.
Why the RTX 5070 reports stand out
Nvidia rates the RTX 5070 at 250 W total graphics power, against 575 W for the RTX 5090. The 16-pin connector is rated for up to 600 W. On paper, a 250 W card has a huge margin.
Two RTX 5070 reports sit in the November list:
- One Reddit user found a melted adapter after black screens and unexpected reboots.
- A second owner of an ASUS Dual RTX 5070 described a similar incident and got a refund after the retailer could not provide a replacement.
A separate VideoCardz report from November 10 covered another RTX 5070 with a melted connector. In that case the owner admits that a friend who installed the card may not have plugged the cable in fully. The sources don't make clear whether that is the same case as one of the two above, so I'm not counting it as a third incident.
There is also older context. In April 2025, VideoCardz covered a ZOTAC RTX 5070 paired with a Seasonic Focus GX-750 power supply, where the power cable melted. Photos suggested that one of the connector pins might have been pushed back, although the exact cause was never confirmed. That is a different incident and shouldn't be merged with this month's.
Low rated wattage doesn't make a connector immune. Board power says nothing about whether current is shared evenly across the pins. In the April 2025 coverage, VideoCardz cited reviewer Der8auer's demonstration of one wire carrying about 22 amps, roughly 260 W, with temperatures up to 150°C. That is more than the whole RTX 5070's 250 W board power. It is one demonstration and not a proof of what happened in any given case. It does show that uneven current is the thing to watch, not the card's headline number.
A short history of the connector
The 16-pin plug arrived with the RTX 4090 in 2022. Nvidia's own guidance at the time described the PCIe Gen 5 power connector as a way to power the graphics card with a single cable. Nvidia's support page frames that as simpler cable routing and better airflow. Early melting reports followed. Nvidia has said it was aware of about 50 cases and pointed to connectors that weren't fully plugged in as a common finding. That statement comes from a 2022 forum post reproducing Nvidia's words, which I couldn't retrieve directly, so treat it as attributed rather than confirmed.
The revised 12V-2x6 connector followed. Per Wikipedia's summary, PCI-SIG published it in July 2023, and it shortened the four sense pins so the connector will not push any power if it has not been fully seated. Nvidia has also reportedly required 12V-2x6 on all RTX 50-series designs.
Is it better? Probably, in the specific case of a half-inserted plug. But reports still appear, including this month. A revision that reduces one failure mode doesn't remove the others, such as damaged pins, poor cable quality, bad bends or uneven current sharing.
The electrical mechanism
The plausible explanation is simple physics, and VideoCardz's roundup lists the same factors:
- The connector spreads current across several pairs of contacts.
- If one contact is loose, recessed or dirty, its resistance rises.
- Higher resistance means more heat at that contact.
- If the current then shifts onto fewer pins, those pins carry more than they should.
- Mechanical strain from a tight bend near the plug can make contact worse.
This is a plausible mechanism, not a diagnosis of any particular card. The reports don't prove a manufacturing defect or a single root cause. They also don't prove that every owner has the same risk.
Does Nvidia owe owners more?
There are two fair views here.
- The "install it properly" view: Nvidia and others have long said full insertion matters. Plenty of reported cases involve unclear installation, and one owner above admits the plug may not have been seated fully.
- The design-tolerance view: the How-To Geek writer argues that a slightly loose plug shouldn't be able to destroy a card worth hundreds or thousands of dollars after four years. That is an opinion, but it is a reasonable one about engineering margins.
Both can be true. Evidence from forum photos can't settle the argument either way. Shiftdelete.net notes that Nvidia has not commented on the November reports, and I found no Nvidia statement in the material I reviewed.
What owners can do
These steps follow from the reported failure mechanism and Nvidia's long-standing emphasis on full insertion:
- Check that the plug is fully seated. There should be no visible gap between the plug and the socket, and the latch should be engaged.
- Avoid sharp bends and sideways pull near the plug. Nvidia's older RTX 40 guidance advised planning at least 1.4 inches (36 mm) of clearance above the card for cable bend and airflow. That page is dated 2022 and covers RTX 40 adapters, so treat it as general guidance for that adapter.
- Use the right cable for your PSU. Modular cables aren't always interchangeable between models, even when the plug physically fits.
- Don't improvise adapters. VideoCardz covered an RTX 5070 system where one 8-pin PCIe connection with a passive splitter creating two 8-pin plugs fed the GPU's 16-pin adapter. Nvidia's installation guidance, as VideoCardz reports it, calls for at least two independent dedicated 8-pin cables, or a native 16-pin cable rated for at least 360 W for that card.
- Watch for warning signs. Black screens, unexpected reboots, a burning smell, discoloration or deformation are reasons to power down. If you see damage, stop using that connection and contact the GPU or PSU vendor about inspection and warranty. An RMA isn't guaranteed in every case.
I'd stop short of recommending that anyone replace an undamaged card, or treat any adapter, undervolt or cable brand as a proven universal fix. The sources don't support that.
Bottom line
The verified part: reports of damaged GeForce power connectors, adapters and cables keep surfacing across RTX 40 and 50 cards, and the November cluster includes the 250 W RTX 5070. The unverified part: how common this is, and what caused each case. For most owners, the practical answer is boring and cheap. Seat the cable fully, don't bend it hard, use the PSU's own cable, and look at the connector again after any case move or cable tidy-up.
References
- We don't talk about Nvidia's biggest failure nearly enough, and your GPU may be suffering from it How-To Geek · 2026-11-17T19:30:30+00:00
- RTX 5070 Melting: 250W Cards Join GeForce Connector Reports shiftdelete.net
- NVIDIA Home nvidia.com