Another Nvidia RTX 5090 cable melts despite MSI's "foolproof" yellow-tipped GPU connector

midian182

Posts: 10,701   +142
Staff member
What just happened? News that another power cable connector melted while plugged into an RTX 5090 is becoming more common these days. What is unusual in this case, however, is that it's the second one involving MSI's supposedly foolproof solution to the problem.

Redditor VersauteGurke explains in a post that he'd been experiencing an issue in which his monitor suddenly turned off and showed a "DisplayPort not connected" message.

After attempting a series of potential fixes, including reinstalling the driver, VersauteGurke discovered that the cable connector had partly melted.

Thankfully for the user, the very expensive RTX 5090 graphics card was fine, with no visible damage to the port or pins. The power supply being used was a Corsair HX1500i 1500W, so a lack of power wasn't an issue.

An interesting extra detail of this incident is that the cable being used was one of MSI's yellow-tipped models. The company was one of the first to ship PSU/graphics card 12V-2x6 power leads with plastic housing molded in bright colors. The idea behind them is that users push the connectors far enough into the cards so that no color remains visible. This ensures the connectors are fully seated, reducing in danger of overheating.

This isn't the first instance of one of these MSI cables melting. There was an almost identical incident in April also involving an RTX 5090 and a Super Flower 1300W ATX 3.1-compliant PSU. The user in that case continued to experience blue screens of death before discovering the GPU-end of the cable's connector had been damaged.

There's plenty of debate on the post's thread over whether this latest incident was caused by user error. Several comments point out that the cable looks to have been bent within 3cm from the GPU end – outside of the guidelines – though this often cannot be avoided when using the massive RTX 5090 in PC cases that aren't very large.

According to this Reddit thread, there have been six confirmed and one unconfirmed RTX 5000 12VHPWR melting cases, though there are doubtlessly many more that go unreported. It's still a better rate than the RTX 4090 – last year, a repair shop said it was receiving 200 melted Lovelace flagships every month.

Permalink to story:

 
Boy, if only Nvidia was competent enough to use a proper connector this time around that they knew worked properly. Too bad it's only their retail customers that are having these "minor" issues...
 
That moment when your intelligence is below what MSI considers "fool", lmfao.
Clearly, they need to redesign the damn thing - the whole 12VHPWR. It's a fiasco.
Agreed. this connector clearly needs load balancing hardware on both ends as mandatory. USB-C does, which is how it can do 240w power delivery without melting.

You NEED to have overprovisioning on dumb connectors to prevent failures, which is why the old 150w 8pin actually topped out at like 425w.
 
It’s happened TWICE with these connectors - and neither has been verified as not being user error….

Not sure we should be so quick to call this a disaster or fiasco or whatever…
Funny, you dont hear about all the melted AMD cards.....

It's almost like Nvidia is pushing the 16 pin connector way too close to the edge, like people have been calling out for years.

Sometimes, when people notice something, they are correct.
 
I think Nvidia have a 12 year old in charge of the GeForce program now - they can't do anything right anymore. The drivers, connectors, and hardware itself are all just either treading water or actively getting worse.
All the people who made Nvidia the behemoth it now is have pivoted to producing vast amounts of AI hardware and software to sell into datacentres. These are being snapped up in truly environment-destroying quantities by various companies and countries that don't really know what to do with them but are sure they need to spend billions on facilities to do something.
 
Funny, you dont hear about all the melted AMD cards.....

It's almost like Nvidia is pushing the 16 pin connector way too close to the edge, like people have been calling out for years.

Sometimes, when people notice something, they are correct.
I’m not talking about the “general” melting connector issue but the specific one with MSI’s colored connectors. People love to make mountains out of molehills…
 
I’m not talking about the “general” melting connector issue but the specific one with MSI’s colored connectors. People love to make mountains out of molehills…
People have been desperate to label the 12v connector issues as a molehill since they first started melting.

Given this is a niche product in a niche market for cards you can barely find, losing 2 in a short time frame is a bit of an issue. This is ONLY MSI cables, and only those who are not using the cable that came with their PSU instead. There's not that many of them.
 
People have been desperate to label the 12v connector issues as a molehill since they first started melting.

Given this is a niche product in a niche market for cards you can barely find, losing 2 in a short time frame is a bit of an issue. This is ONLY MSI cables, and only those who are not using the cable that came with their PSU instead. There's not that many of them.
Who are these “people”?!?
While it’s a niche market, we’re still talking several thousand cards sold… and only 2 have melted - possibly neither due to the fault of the GPU, PSU, motherboard or connectors…
 
Seasonic's new mitigation psu concept isn't available for the market until q1/2026. So anything on the market including Seasonic's $560 Noctua PSU ( atx 3.1) still has a risk of burn in based on the new mitigation concept. The problem might be bigger than we are made to perceive.

PC enthusiast gaming 2025 bring your own thermal camera. 🤪1000023484.jpg1000023478.jpg
 
The yellow-tip idea from MSI is actually smart in theory, but it's clearly not a fix-all. Between tight cases, stiff cables, and GPUs the size of small countries, "fully seated" can mean different things depending on your setup. Maybe it's time for a complete connector rethink instead of color-coding band-aids.
 
The yellow-tip idea from MSI is actually smart in theory, but it's clearly not a fix-all. Between tight cases, stiff cables, and GPUs the size of small countries, "fully seated" can mean different things depending on your setup. Maybe it's time for a complete connector rethink instead of color-coding band-aids.
Until they figure it out, the beta testing continues and you have to go on to the next beta test phase with new PSU/ per pin voltage control ( in concept until q1 2026 ; with fail safe feature ) or Asus's strix per pin control/ monitoring from the gpu side. So this would be the 3rd beta test starting from atx 3.0 and initial 12VHPWR design. the second beta test is with atx 3.1 and the industries hope/ that the burn in issues will slow down. Another day another burn in cable story.
Did anyone know that Seasonic's 12V-2x6 cable is blue which doesn't mitigate risk if you plug it in correctly. The elephant in the room is that it's not that MSI might have wonder things to take advantage of the situation it that Nvidia blamed this initially on plugging it in shy of 100% perfection to mitigate burn in risk.
On this is for sure Nvidia loves one time use graphics cards!🤡🤡🤡1000023832.jpg
 
The entire point of a connector is to prevent user error. If user error caused this failure then the connector failed one of its most basic functions
We don’t know what the problem was.. and it only prevents the user-error of not seating the cable connection properly… if the cable was crimped due to poor cable management/small case, that’s nothing the connector could have done to fix.
 
We don’t know what the problem was.. and it only prevents the user-error of not seating the cable connection properly… if the cable was crimped due to poor cable management/small case, that’s nothing the connector could have done to fix.
The point of a connector is to eliminate user error. however, we now have multiple cables from multiple manufacturers all with the same issue.

At somepoint we might have admit that the design is fundamentally flawed. that, or sticking 600 watts into a 12 pin connector is just a bad idea. And since it's DC, I think its only 6 wires that are providing power.
 
The point of a connector is to eliminate user error. however, we now have multiple cables from multiple manufacturers all with the same issue.

At somepoint we might have admit that the design is fundamentally flawed. that, or sticking 600 watts into a 12 pin connector is just a bad idea. And since it's DC, I think its only 6 wires that are providing power.
No… we have TWO reported melted connectors with this specific model… none of the other incidents were with this connector.

Nothing can protect against every user error. These special connectors only protect from a user not seating the cables properly… any other issue will not be resolved… much like me wearing a seatbelt won’t save me from some one shooting me with a pistol…
 
No… we have TWO reported melted connectors with this specific model… none of the other incidents were with this connector.

Nothing can protect against every user error. These special connectors only protect from a user not seating the cables properly… any other issue will not be resolved… much like me wearing a seatbelt won’t save me from some one shooting me with a pistol…
Well the issue on the 5090 is that all of the positive 12V connectors coming from the cable are all bridged and I have 1 protection circuit between the bridge and the power delivery to the card. So, yeah, very bad design. Second, sending 600 watts into 6 wires is a recipe for disaster. I'm not against a 12pin connector, I'm against a 600w 12 pin connector. The 8pin was severely over engineered and under spec'd, the 12 pin was under-engineered and over spec'd. Then, nVidia put LESS protection on the card side on the 50 series than it did on the 40 series.

I argue that:
1)the 12pin is fundamentally flawed
2)nVidia is removing protection circuits from their cards
3)600 watts DC is just too much for a 16 gauge wire where only 6 wires are providing power
4)user error isn't a valid excuse as connectors were inherently invented as a safety feature to prevent user error.
 
Well the issue on the 5090 is that all of the positive 12V connectors coming from the cable are all bridged and I have 1 protection circuit between the bridge and the power delivery to the card. So, yeah, very bad design. Second, sending 600 watts into 6 wires is a recipe for disaster. I'm not against a 12pin connector, I'm against a 600w 12 pin connector. The 8pin was severely over engineered and under spec'd, the 12 pin was under-engineered and over spec'd. Then, nVidia put LESS protection on the card side on the 50 series than it did on the 40 series.
All well and good but not proven - there were more incidents with the 4090 than the 5000series - most of which can be fixed with an anti-sag bracket…
I argue that:
1)the 12pin is fundamentally flawed
Possibly, but we have no proof
2)nVidia is removing protection circuits from their cards
No proof that this is true - or that it matters
3)600 watts DC is just too much for a 16 gauge wire where only 6 wires are providing power
Shouldn’t be…
4)user error isn't a valid excuse as connectors were inherently invented as a safety feature to prevent user error.
NOTHING prevents all types of user error… and 2 faults out of thousands of cards is pretty good…
 
Possibly, but we have no proof
It's melting on luxury products and therefore not providing a luxury experience. We have plenty of proof, whether or not you choose to ignore it is up to you. We do live in a world where ignoring evidence is the new standard so I guess going for par is fine
 
It's melting on luxury products and therefore not providing a luxury experience. We have plenty of proof, whether or not you choose to ignore it is up to you. We do live in a world where ignoring evidence is the new standard so I guess going for par is fine
Evidence is not the same as proof. We have circumstantial evidence but 0 proof…. Very important to understand the difference between the two.
 
Every one of the 12V pins has cooked! That's exactly the same failure mode as the previously reported yellow tip failure.
 
Funny, you dont hear about all the melted AMD cards.....

It's almost like Nvidia is pushing the 16 pin connector way too close to the edge, like people have been calling out for years.

Sometimes, when people notice something, they are correct.

AMD doesn't compete at top tier.
 
Back