Thursday, February 19th 2026
NVIDIA GeForce RTX 5090 Melts 12V-2×6 Connector Despite -100 W Power Limit
The melting and burning 12V-2×6 connectors on high-end NVIDIA—and some AMD—GPUs is well documented at this point, and although there is mounting evidence to support the fact that the new connector standard itself is partially to blame, there is still the idea that high power draw is to blame for these meltdowns. This is not necessarily the case, according to one user of the Mobile01 forum. According to the forum post about the incident, the top row of pins in a Gigabyte Aorus Master Ice RTX 5090's GPU power connector has melted down, despite the user having implemented a steep undervolt and power limit.
When installed, the user had limited GPU power to 500 W and dropped the voltage limit down to 0.9 v—that's a reduction of 100 W and 0.205 v compared to stock settings. It's worth noting that the GPU in question was connected to the PSU using a 16-pin adaptor, and the connector melted at the GPU side, not the PSU side, and the first indication of the meltdown was frequent crashing during gaming sessions. Both connector ends at the GPU side seem to be affected, but it's entirely possible that the GPU would have suffered even more damage had the power limit been higher. Currently, independent research seems to indicate that the 12V-2×6 cable itself seems to suffer from an unstable or inadequate connection, which increases the current and temperature across individual pins. Unofficial adaptors also seem to exaggerate these contact issues.
Source:
Mobile01 forum
When installed, the user had limited GPU power to 500 W and dropped the voltage limit down to 0.9 v—that's a reduction of 100 W and 0.205 v compared to stock settings. It's worth noting that the GPU in question was connected to the PSU using a 16-pin adaptor, and the connector melted at the GPU side, not the PSU side, and the first indication of the meltdown was frequent crashing during gaming sessions. Both connector ends at the GPU side seem to be affected, but it's entirely possible that the GPU would have suffered even more damage had the power limit been higher. Currently, independent research seems to indicate that the 12V-2×6 cable itself seems to suffer from an unstable or inadequate connection, which increases the current and temperature across individual pins. Unofficial adaptors also seem to exaggerate these contact issues.



133 Comments on NVIDIA GeForce RTX 5090 Melts 12V-2×6 Connector Despite -100 W Power Limit
Right now there are 0.5% chance that either my 9950X3D get fried or my 5090's power connector burn, perfectly acceptable by my standard :roll:
We can all see that Nvidia cards with a lower wattage have a lower failure rate due to the connector. Nvidia is 94% of the market, pretty easy to conclude the sample size is much larger than the whole 2 9070 XT models that actually have the connector. There are potential aggravating circumstances like the one AMD card placing an extreme bend and covering it up, allowing heat build-up. That would heavily impact the effective safe load. I digress, there just aren't enough AMD cards with the connector to say that disproves the much larger sample size we have on the Nvidia side.Consider that it keeps getting discussed because people keep reporting their burning connectors on a frequent basis. That we need to discuss it this often disproves your earlier rejection that it isn't a frequent issue. It'll keep getting brought up until the issue is fixed because, as expected, failures have and will continue.So you recognize the standard lacks safety measures. 600w under ideal conditions, basically no one runs the connector under ideal conditions. The connectors aren't perfect out of the factory, let alone factoring in other variables.
PCI-SIG engineers didn't design the standard, Nvidia did. That's the requirement for sponsoring it. PCI-SIG is just a body of companies, it's members do the work. When a draft is introduced, the other members are only responsible for ensure interoperability.
"Right now, Nvidia are reportedly working on two different versions of their new 12V2x6 connector,"
overclock3d.net/news/power_supply/rip_12vhpwr_nvidia_and_pci_sig_have_a_replacement_in_the_works/
I do not understand why anyone who knows the connector has no safety measures would defend it. Connector issues have not stopped, if anything they've gotten from frequent. At least with GPU prices so high now, maybe we'll start seeing a slowdown in reports.
They submitted the standard to PCI-SIG and pilot push the standard since 3000 series.
Also it is not just the connector, but also the faulty electrical design behind it.
From 3090->4090->5090 they've trimmed the electrical controllers from 3->1 ->1
The consumer have to pay the price for Nvidia having fancy FE board designs with less and less PCB space.
It is just one giant 12+ve plane and one giant 12-ve plane behind it without any load balancing between the wires.
The burnt connectors are just the symptom.
With only one 12+ve plane and one 12-ve plane,
the simplest and safest solution should always be '2 thick wires with large pins'
was the current guaranteed the same?
note: information and claims and statements from the news piece.
edit:
better:
the user claims to have limited GPU power by dropping the voltage limit down to 0.9 V DC with unknown software which may have used less Power in total. The current in question per in is unknown. That should be a drop of ~ -0.20 V DC compared to stock settings.
watt = voltage multiplied with current in the direct current world
-- i read quite often that less voltage causes more frames and most likely more current draw. i never verified as i do not usually buy nvidia desktop cards and do heavy paid games and heavy paid overclocker benchmark software. specialists should answer if the current stays the same with dropped voltage, higher fps on nvidia 4090, 5090. - or if the current increases with those higher fps
valid measurement per pin with true rms current clamp meter. not valid software hoax software. valid measurement could be elmor labs calibrated put in between measurement board for all board connectionsi wonder how many years i will read about it?
2026 done
2027, 2028 most likely
2030 hopefully not
the hate is real for nvidia here. from the hardware perspective. driver i accepted the garbage after 20 years with gnu gentoo linux. from technical perspective i would have expected them to phase out that connector and use something else. not a revisioned crap of existing 12v high power something connector. it is a 12 volt dc low power connector to power a 10 watt device .Pi think the clean look esthetics
should be a white build, pure white. capacitors in red black etc. user tend to paint non white parts and connectors
glass to show parts and cables look ugly. the hole thing about sleeves to put an extra layer of regular pattern around each cable. than use cable combs.
it is not lego. electronics have wires. or ribbon cables which are wires in tape form. or slot in card connectors like m2 2280 nvme.
Nvidia is just in the business of selling GPUs, and making sure they do. The usage is secondary. They keep releasing products, so if something breaks, just get a new one and further Huang's jacket collection instead. We saw the same thing with their RT push. Turing had barely functional RT but got released anyway. Even today you could wonder what the fuck that was ever good for, upscaling similarly is a bandaid for overall performance degradation, it is not an improvement. I can still look at a massive pile of games released under Pascal and running 120+ FPS on Pascal that look better than the shit getting produced today on GPUs twice as fast.
It was no surprise really. Since Turing, Nvidia has been on a sharp trajectory downwards: less VRAM/$, perf/$ gen to gen nuked into the ground, Ampere on a horrible node, the list goes on and on. This isn't just AI. This is what monopolists do and 12vhpwr should have been the moment we collectively said FU instead of downplaying what is essentially a fire hazard you paid for.False equivalency is a popular thing these days, to hide obvious fact.
And yes, 8-Pin connectots have molten in the past, but the frequency throughout the years is not even close to what 12V-2x6 has within 3 years already.
This 12V-2x6 connector is basically two 6-Pins (even less due to pin thickness reduction) pushing 300W each, while those are rated for only what? 75W-100W? On top of that there is no load-balancing, so god knows how much more each pin recieves due to poor connections of surrounding pins.
This connector sucks, and there is nothing good about it which is worth defending.
There is no doubt in my mind that even my 4070 Ti connector will eventually melt, despite being a sub 300W card.
This cable will burn, and it's just a matter of time when, after the "next bad plug" or after 0.5 - 3 years.
The heat over pins are too much and the "quality" just give some more time.
2 days ago
wccftech.com/user-reports-ryzen-7-9800x3d-and-9850x3d-failures-within-a-few-weeks-first-9850x3d-failure-case/
My 12VHPWR cable is already over 3.5 year old too :roll: , I reused the cable LOL, no sign of anything yet.
But don't worry, physics always works, it's just slower sometimes.
12VHWPR is perfectly fine IF you manage to plug it in correctly. Literally hundreds of millions of 12VHPWR powered cards are currently working flawlessly in hundreds of millions PCs worldwide.
I'm glad some people are put off by the 12VHPWR connector though, way too many folks are buying Nvidia these days :roll:
Translating the OP, he indeed used the OEM part provided by the manufacturer, so that should be a valid and reliable option:What we don't know is OP had "an improper bend" in the cable, but that just further demonstrates how bad a design this is. It shouldn't require this level of scrutiny for a $2-3k GPU.