RTX 5090 Connector Melting: New Reports and How to Reduce Risk

RTX 5090 power-connector melting is still happening. In the past week, owners of MSI GeForce RTX 5090 cards have reported fresh cases of scorched or melted 12V-2x6 connectors and cables, despite MSI's "GPU Safeguard+" per-pin current monitoring on its newest MPG Ai1600TS and Ai1300TS power supplies. Thermal-imaging work shared by independent hardware testers continues to point to uneven current distribution across the connector's twelve pins as a likely contributor, though other investigators looking at damaged units argue that incomplete seating and the contact resistance it creates is just as plausible a cause. Neither NVIDIA nor MSI has published a root-cause finding tied to these newest cases. Here's what's actually confirmed, what's still a theory, and what RTX 5090 owners can do today to cut their risk.

What's new in the past week

The latest reports describe MSI GeForce RTX 5090 cards whose 12V-2x6 connectors showed melting or scorch marks after months of normal use, not immediately after installation. In one widely discussed case, the card's own bundled cable melted at the GPU-side connector while running from a 1,000 W power supply, with no third-party adapter in the chain. In another, a card that had reportedly been running without issue for months developed a burned connector during a routine session. Both cases have reportedly gone to authorized service centers for inspection, and as of this writing neither MSI nor the card owners have published a confirmed cause or warranty outcome.

What makes this round of reports notable is that it keeps happening well into the RTX 5090's life cycle and on a card line that ships with the revised 12V-2x6 connector specifically designed to reduce these failures. That timing is why the root-cause question below matters more than any single incident.

This isn't a new problem for high-wattage GPU connectors

Melting power connectors have dogged NVIDIA's high-end cards since the original 16-pin 12VHPWR connector debuted with the RTX 40 series in 2022. Early investigations tied those failures mainly to connectors that weren't fully seated, which left a gap that raised resistance and heat at individual pins. PCI-SIG and the ATX 3.1 specification later introduced the 12V-2x6 revision, which recesses the sense pins so they only make contact once the power pins are properly seated — the goal being to stop a partially plugged-in card from drawing full power in the first place. The 12VHPWR and 12V-2x6 connector history on Wikipedia lays out that timeline, including the fact that melting reports continued even after RTX 50-series cards adopted the newer connector. The RTX 5090's recommended system power figures and connector requirements are listed on NVIDIA's own GeForce RTX 5090 product page.

What MSI's GPU Safeguard+ actually does

MSI's answer to the recurring melting reports is a feature it calls GPU Safeguard+, built into its MPG Ai1600TS and Ai1300TS power supplies. According to MSI's own explanation of GPU Safeguard, the PSU includes dedicated circuitry that can monitor current on each individual pin of the 12V-2x6 cable in real time, rather than only watching total GPU power draw. If it detects current piling up unevenly on a subset of pins, the PSU beeps and MSI Center displays a warning; if the imbalance persists, the power supply can shut itself off to protect the graphics card before damage spreads.

MSI paired that hardware-side monitoring with its Afterburner software. Per MSI's announcement of the Ai1600TS and Ai1300TS PSUs, when GPU Safeguard+ flags a problem it can notify Afterburner directly, which then drops the GPU's power limit automatically to reduce load on the connector while the user investigates. Setting this up isn't automatic, though: owners have to load the PSU as a data source inside Afterburner and select the correct 12VHPWR alarm channel for the connector actually in use, since the Ai1600TS has two separate 12V-2x6 outputs.

That matters for context on the latest reports. GPU Safeguard+ is a PSU feature, not something built into the RTX 5090 itself, and it only exists on two specific MSI power supplies. A melting incident on any other PSU, or on an Ai1600TS/Ai1300TS where the monitoring wasn't configured, would get no benefit from it at all.

The root-cause debate: uneven pins vs. poor seating

There are two leading explanations in circulation for why 12V-2x6 connectors keep overheating on high-power cards like the RTX 5090, and reporters and reviewers have not settled on one.

The first points to the card's internal wiring. Independent thermal-imaging testing of RTX 5090 cards under sustained load has reportedly found that current isn't split evenly across all twelve power pins — instead, a small number of pins carry a disproportionate share of the load, running hot enough to soften cable insulation even when the connector was confirmed to be properly seated. The suggestion from that line of testing is that because the RTX 5090 draws so much power through a single connector, any small difference in contact quality between pins gets amplified into a large difference in current and heat.

The second explanation is more conventional: incomplete seating and the contact resistance it creates. A connector that looks clicked-in but isn't fully flush against the socket leaves a small gap at one or more pins; that gap raises electrical resistance at exactly that contact point, and resistance under high current means heat. This was the dominant explanation for the original 12VHPWR failures on RTX 40-series cards, and it remains a live explanation for RTX 5090 cases too, particularly when a damaged connector shows localized melting at one or two pins rather than even wear.

Both explanations can be true for different incidents, and nothing here has been independently confirmed as the single cause for the newest MSI reports specifically. NVIDIA's own installation guidance is relevant to the seating theory regardless of which explanation turns out to dominate: the GeForce RTX 5090 Founders Edition user guide (PDF) instructs installers to plug the power adapter into the graphics card first "to ensure it's firmly and evenly plugged in" before mounting the card, and to confirm the plug is "fully inserted and seated flush against the power receptacle." Cable and connector maker Corsair gives similar advice in its own guidance on the 12VHPWR/12V-2x6 connector, which notes that the revised connector's sense pins are designed to prevent the card from drawing full power unless the main power pins are properly engaged first.

Where NVIDIA stands

As of this writing, NVIDIA has not issued a public statement addressing the specific MSI RTX 5090 melting reports from the past week. That silence doesn't mean NVIDIA has done nothing about the connector generally — its own product documentation covers recommended power-supply wattage and connector configurations for the RTX 5090, and its user guide includes the seating instructions quoted above — but there is no confirmed advisory, recall, or root-cause statement tied to these newest incidents. Readers should treat any claim about what NVIDIA has privately concluded, or any specific internal failure-rate figure, as unconfirmed unless it comes from NVIDIA directly.

Practical steps to reduce your risk

None of the following eliminates the possibility of a connector failure, since the root cause isn't settled, but each addresses a factor that's been repeatedly implicated in past incidents.

1. Reseat the connector deliberately. Unplug the 12V-2x6 cable or adapter, inspect for any visible gap, and push it back in until it sits flush against the card's receptacle — don't rely on the latch click alone. NVIDIA's own user guide recommends plugging the adapter into the card before the card goes into the motherboard, specifically so you can see and feel that it's seated evenly.
2. Avoid sharp bends right at the connector. Give the cable room to curve gradually rather than forcing an angle immediately behind the plug, which can stress individual pins unevenly.
3. Stick to cables and adapters rated for your card's full power draw. Avoid daisy-chaining adapters or using extensions that aren't explicitly rated for the RTX 5090's power delivery.
4. If you own an MSI MPG Ai1600TS or Ai1300TS, turn on GPU Safeguard+ monitoring. That means installing the PSU data provider in MSI Afterburner and selecting the correct 12VHPWR alarm channel for the connector you're actually using, as described in MSI's own documentation linked above.
5. Watch for warning signs during and after long sessions. A faint burning-plastic smell, discoloration on the cable or connector housing, or any softening of the insulation near the plug are reasons to power down and inspect immediately rather than keep running the card.
6. If you see damage, stop using the card and contact support. Don't attempt to keep gaming on a connector that shows melting or discoloration; contact NVIDIA, your card maker, or your retailer for an RMA rather than continuing to run it.

Related RTX 5090 coverage

This connector-safety issue is separate from other recent RTX 5090 storylines on this desk, including the ongoing questions around RTX 5090 supply and discontinuation rumors and how far RTX 5090 resale prices have climbed above its original MSRP. Neither of those pieces is about the melting-connector issue, and this article isn't about pricing or availability — just the safety question.

What to watch next

A few things would move this story from "debated" to "resolved." A formal root-cause statement from NVIDIA or MSI tied specifically to these newest cases would settle whether uneven pin loading, incomplete seating, or something else is to blame. A hardware or firmware revision to GPU Safeguard+, or its expansion to more PSU models and more vendors, would suggest manufacturers see per-pin monitoring as a necessary stopgap rather than a nice-to-have. And any statement from PCI-SIG or the ATX standards body about further revising the 12V-2x6 spec would indicate the industry views the current design as still unfinished business. Until any of that happens, the safest approach for RTX 5090 owners is the practical checklist above: reseat carefully, avoid stressing the cable, use per-pin monitoring if your PSU supports it, and watch for warning signs.

Written byMara Lindqvist

Mara Lindqvist edits the hardware desk. She covers graphics cards, processors, memory and storage, the foundries and chip designers behind them, and what the numbers on a spec sheet mean for people choosing a PC. Specifications in her stories come from manufacturer spec pages and datasheets.

More from Mara LindqvistAll Hardware coverage