In this case, a Reddit user posted an image of damage around a graphics card’s power-connector area after noticing a smell while using a system they associated with PNY and DigitalStorm. The owner said the card was underclocked and framed the incident humorously around playing Shattered Pixel Dungeon and then Crypt of the NecroDancer. They also alleged that PNY declined assistance because the cord was not supplied by PNY, while DigitalStorm offered to help facilitate contact.
Those details make for an alarming story, particularly given the RTX 5090’s price and power demands. They do not, however, establish that either game, a supposedly low GPU load, the card itself, the cable, the power supply, or the system builder caused the damage. Nor is there an independently verified PNY warranty decision in the available record. The important takeaway is not a verdict on this one PC, but a clearer view of what is known, what remains unproven, and what RTX 5090 owners should do now.
What the owner reported—and what the report cannot prove
The firsthand post is evidence that its author reported an apparent connector failure and shared an image of the affected card-side area. The owner said they noticed a smell after an earlier hour of gaming, followed by another game, and later stated that the GPU had been underclocked.
That is meaningful evidence of a reported incident, but it is not a forensic diagnosis. There are no contemporaneous measurements of GPU board power, voltage, per-wire current, connector temperature, frame rate, or power-supply behavior. The available material also does not independently establish the exact PNY card model, the system’s full configuration, the PSU model, the cable’s manufacturer and revision, or whether adapters, extensions, or particular routing choices were involved.
Most importantly, game names are not power telemetry. A visually simple game can still coexist with elevated system power draw under certain conditions, and an underclock does not by itself tell us the actual power limit, transient behavior, connector contact quality, or current distribution. Conversely, nothing in the report proves that high power was involved. Without logs and physical examination, the incident cannot responsibly be labeled a “low-load failure,” a GPU defect, cable failure, or installation error.
The uncertainty extends to the location where the problem began. Connector damage may involve the card receptacle, the cable plug, the conductors, the PSU-side connection, or several elements after heat builds up. A photo taken after the event cannot by itself identify that sequence.
The PNY coverage claim needs careful wording
The strongest version of the circulating claim—“PNY will not cover it”—goes further than the evidence currently supports.
What is available is the owner’s allegation that PNY’s response amounted to saying the cord was not PNY-supplied and that the customer had therefore caused the problem. The owner also said DigitalStorm offered support in communicating with PNY. No public response from PNY or DigitalStorm was retrieved to confirm the exchange, explain a denial, identify the applicable warranty party, or disclose a final claim determination.
PNY’s published policy says GeForce RTX 50-series graphics cards carry a three-year limited warranty and describes PNY’s obligation as repairing or replacing defective product parts. That establishes the existence of a stated limited warranty, not eligibility for this particular system. A prebuilt PC can add important complications: the retailer or system integrator may be the primary point of service, the cable may have been supplied with the power supply rather than the GPU, and the terms of the system purchase may determine the proper route for a complete-machine claim.
For readers, the practical distinction matters. It is fair to say that an owner alleges a cable-based PNY rejection. It is not fair, on the current record, to say PNY has a verified blanket policy of refusing RTX 5090 connector claims involving non-PNY cables—or that this owner’s claim is conclusively closed.
Anyone facing a similar dispute should preserve the entire system in its post-incident state where safely possible. Photograph both ends of the cable, the GPU receptacle, the PSU-side socket if accessible, labels on the GPU and PSU, and any visible discoloration. Keep invoices, serial numbers, the original cable and packing materials, service-ticket records, and written correspondence. Do not discard the cable or attempt to clean, reshape, or reuse a heat-damaged connector before the warranty provider has documented it.
The connector is 12V-2x6, not simply “the 16-pin plug”
“16-pin connector” is understandable shorthand, but it obscures an important technical detail. PCI-SIG says 12V-2x6, specified in CEM 5.1, replaced the earlier 12VHPWR connector. Nvidia’s RTX 5090 quick-start documentation identifies the GPU connection as a PCIe Gen5 16-pin power connector.
The practical guidance in Nvidia’s installation instructions is direct: the plug should be fully inserted and seated flush against the power receptacle. That is the clearest confirmed action owners can take, but it should not be turned into a universal explanation for every failure. A fully inserted plug is an installation requirement; it is not proof that every damaged connection was improperly seated, and an image after a heat event ordinarily cannot reconstruct the original seating state.
Likewise, cable brand alone does not settle liability or root cause. The reported PNY/DigitalStorm system allegedly involved a cord that was not PNY-supplied. Yet another independently reported RTX 5090 incident involved an MSI Gaming Trio OC and a PSU-supplied Corsair 600 W 12VHPWR cable, with reported melting on both sides of the connection. That does not absolve any particular cable, card, PSU, or assembly method in any individual case. It simply means that “third-party cable” versus “official cable” is too crude a test for assigning blame.
Why uneven current distribution remains a serious lead
Independent RTX 5090 reporting has offered a technically plausible mechanism for concern: uneven sharing of current among the connection’s 12 V conductors.
In one February 2025 investigation, testing of an RTX 5090 Founders Edition reportedly found one of six 12 V wires drawing more than 22 A at 575 W, with temperatures above 150 degrees Celsius. Concentrating too much current in one conductor or contact can create substantial localized heating, even where the overall connection is designed for high power delivery.
But the evidence is not uniform. ComputerBase reported lower imbalance in a separate configuration, measuring current from 5.3 A to 10.6 A across conductors. That difference is precisely why the current-distribution hypothesis should be treated as a potential mechanism rather than a settled explanation for all affected cards.
Variables can include the GPU, cable construction, PSU behavior, connector contact condition, system configuration, and test method. The available evidence supports concern that some RTX 5090 setups can show problematic current imbalance. It does not demonstrate a single universal RTX 5090 hardware defect, and it does not establish the cause of the reported PNY/DigitalStorm incident.
There is an additional cautionary example. An early report attributed a melted cable to RTX 5090D and RTX 5080 testing, but later examination found burn marks associated with previously tested RTX 4090 Founders Edition hardware. The subsequent conclusion was that no RTX 5090 had been harmed in that event. That correction is a useful reminder that viral images and initial attributions are not a substitute for component-level investigation.
What RTX 5090 owners should do now
This incident does not establish a failure rate. There is no reliable denominator for installed RTX 5090 cards, no public database of all connector failures, and no basis for calculating the likelihood that an individual owner will experience one. Still, the consequences of a damaged high-power connection are serious enough to justify a methodical inspection.
Owners building or upgrading a Windows gaming PC should take these steps:
- Power down fully before inspecting. Shut down Windows, turn off the PSU, and disconnect mains power. If there is any burning smell, visible melting, smoke, or discoloration, stop using the system rather than repeatedly testing it.
- Check seating at the GPU end. Follow the card maker’s documentation: ensure the 16-pin plug is fully inserted and flush. Do not rely only on whether the connector looks generally close to the socket.
- Avoid unnecessary strain at the plug. Route the cable so the side panel or cable bundle does not place immediate lateral force on the connector. Do not force a sharp bend directly beside the plug.
- Use the intended power path. Prefer the PSU maker’s approved native cable where applicable, and avoid unneeded adapters or extensions. This is good risk management, not proof that every other cable is unsafe.
- Inspect both ends when possible. Look for discoloration, deformation, loose fit, softened plastic, or heat marks at the GPU end and, if the cable design permits, the PSU end.
- Document before changing parts. Clear photos and a written timeline can matter if a system builder, GPU vendor, PSU vendor, or retailer needs to assess responsibility.
- Do not reuse damaged hardware. A visibly compromised plug, cable, or socket is not something to reseat and continue using while waiting for a warranty outcome.
Prebuilt owners should normally start with the system seller or integrator, because that party can validate the as-shipped configuration and coordinate component warranty handling. Do not assume the graphics-card manufacturer is necessarily the only—or even the first—appropriate contact.
The financial stakes are unusually high
The connector discussion comes at an awkward time for RTX 5090 owners because replacing the card is exceptionally expensive. An August 2026 U.S. market survey placed the RTX 5090 around $4,900, about 145% above its $1,999 MSRP. A September 14 report said first-party online inventory had nearly disappeared and cited third-party listings roughly from $6,500 to $9,500 or higher.
Those figures are market snapshots, not a guarantee of a future increase or a valuation for any used or damaged card. They do explain why clear warranty handling and careful evidence preservation matter more than usual. A connector dispute can involve a GPU, PSU, cable, and possibly a prebuilt-system warranty—at a time when an out-of-pocket replacement may be difficult to source and dramatically above list price.
Concern is warranted; conclusions should wait
The reported PNY-associated RTX 5090 damage deserves attention because it adds another public account to an already sensitive issue around high-power GPU connections. Other RTX 5090 connector incidents have been reported, and testing has shown that uneven current distribution can be a credible concern in some configurations.
But this particular report does not yet identify a root cause, demonstrate that lightweight games or low load triggered the event, prove a card-wide defect, or independently verify that PNY finally denied warranty service because of the cable. The most responsible response for owners is practical rather than speculative: inspect the connection safely, ensure it is fully seated, avoid cable stress, retain documentation, and escalate through the appropriate system seller and component vendors if any sign of overheating appears.
For a flagship GPU carrying both high electrical demand and unusually high replacement cost, that caution is not alarmism. It is basic protection against an unresolved problem becoming a much more expensive one.