A PNY GeForce RTX 5090 died while sitting idle on a desk, its 16-pin power connector melted on both the card side and the power-supply side. When the owner filed a warranty claim, PNY denied it the same day, arguing that only its own bundled cable or adapter is covered under warranty, according to Tom’s Hardware. The catch: the customer was not using a sketchy third-party adapter. He was running the native 12V-2×6 cable that shipped with his own ASRock PG-1600G power supply, a configuration built to the same industry connector standard the card itself uses.
The dispute, first reported around September 24-25, 2026, has reignited a debate that has followed Nvidia’s 12VHPWR and 12V-2×6 power connectors since the RTX 4090 launched in 2022: who is responsible when a $2,000-plus graphics card’s power connector overheats, and can a manufacturer’s warranty terms quietly narrow after the sale. For a GPU market already strained by record RTX 5090 prices and tightening GDDR7 supply, a high-profile warranty refusal adds a new kind of risk for buyers to weigh.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Happened: PNY’s Warranty Denial, Explained
According to reporting from Tom’s Hardware, TechSpot, and Wccftech, a Reddit user posted that their PNY RTX 5090 stopped being detected by the system after the card had simply been idling, not under a heavy gaming load. Inspection reportedly showed melting at both ends of the 16-pin connector: the receptacle on the graphics card and the plug on the power-supply cable. The user filed a warranty claim with PNY and received a denial the same day it was submitted, per TechSpot’s account of the exchange.
PNY support told the customer: “All PNY GPUs that require a 16-pin (12VHPWR / 12V-2×6) input include a PNY-approved power cable or adapter in the box,” and added, “This cable is the only configuration covered under warranty,” according to the emailed response reproduced in Tom’s Hardware’s report. When the customer asked which specific clause of PNY’s published warranty policy supported that reading, a senior support technician reportedly said the decision was final and closed the ticket, a detail also noted by Wccftech and The FPS Review.
PNY sells graphics cards through a mix of direct retail and authorized partners, and its published support channels, including the PNY support portal, list standard RMA and warranty-claim procedures without a visible, itemized cable-compatibility clause of the kind support staff cited in this dispute. That gap between the self-service documentation a shopper can read before buying and the standard support staff applied after a failure is the detail most outlets have focused their scrutiny on.
The Cable in Question Was Not a Third-Party Adapter
The detail driving most of the pushback in enthusiast forums and tech coverage is what cable was actually in use. The RTX 5090 was connected using the native 12V-2×6 cable that came bundled with the owner’s own ASRock PG-1600G power supply rather than a splitter, a multi-8-pin adapter, or a cable from an unrelated third-party brand. That distinction matters because the 12V-2×6 connector is an industry standard managed by PCI-SIG, not a PNY-proprietary interface. A cable built to spec by a reputable PSU maker is, in principle, interchangeable with the cable PNY ships in the box.
PNY’s support reply, as quoted by Tom’s Hardware, framed it differently: “If a third-party power cable, adapter, or PSU cable is used and damage occurs – including melting, burning, overheating, or deformation of the connector – PNY cannot accept the product under warranty,” and further, “Damage caused by non-PNY cables falls under the responsibility of the manufacturer of that cable or power supply.” Under that reading, any cable that did not ship inside the PNY retail box, even a PSU manufacturer’s own 12V-2×6 cable built to the same PCI-SIG specification, would fall outside coverage.
The Fine Print Problem: What PNY’s Published Warranty Actually Says
The most damaging part of the story for PNY is not the denial itself but the apparent gap between that denial and PNY’s own public documentation. Multiple outlets, including Tom’s Hardware and Wccftech, report that PNY’s current published GeForce warranty terms do not contain explicit language restricting coverage to PNY-branded cables or adapters. If accurate, that means the customer was held to a policy that support staff described verbally and in email, but that does not appear in the written terms PNY makes available to buyers before purchase.
Adding to the tension, a separate user reportedly surfaced an exchange from roughly a year earlier in which PNY support told a different customer that a native PSU-supplied 12V-2×6 cable was an acceptable substitute for PNY’s own adapter. If that record is genuine, it suggests PNY’s own support guidance on this exact question has been inconsistent over time, not just unclear in the fine print.
12VHPWR to 12V-2×6: A Short History of a Troubled Connector
This is not the connector’s first controversy. Nvidia introduced the 12VHPWR standard alongside the RTX 4090 in 2022 to replace multiple 8-pin PCIe power connectors with a single, denser 16-pin plug capable of delivering up to 600 watts. Within months, enthusiast and repair-focused outlets such as Gamers Nexus and Tom’s Hardware were documenting melted connectors on early RTX 4090 cards, with root causes debated across partial cable seating, tight bend radii near the plug, and manufacturing tolerances on the pins themselves. That current-concentration theory is still being tested four years later: TechPowerUp reported that a September 2026 Hardware Unboxed teardown measured roughly 22 amps running through a single connector pin on an RTX 5090, a load enthusiasts say leaves little margin for error even on a design that has already been revised once. The RTX 5090 generation has followed the same pattern since launch: DSO Gaming reported the first RTX 5090 melted-connector case in February 2025, TechSpot reported 2 additional incidents in April 2025, and it reported a third early case in May 2025.
PCI-SIG, the industry body that maintains the PCI Express and associated power-connector specifications, published a revised version of the connector known as 12V-2×6 in 2024, changing pin geometry to improve contact reliability and reduce the risk of partial insertion going unnoticed. The RTX 5090 at the center of this dispute uses that newer 12V-2×6 connector, not the original 12VHPWR design, according to the reporting reviewed by Tom’s Hardware. That the revised connector can still reportedly melt, on both the card and cable side, is part of why this case is drawing wider attention rather than being dismissed as an isolated, already-solved problem. The margins appear to be shrinking further as workloads change: TechPowerUp reported in July 2026 that an RTX 5090 connector reached 91.7°C with Nvidia’s DLSS 5 enabled, compared with 80.9°C with it off, suggesting the revised connector’s thermal headroom narrows measurably under newer rendering techniques, not just under raw horsepower. Tom’s Hardware reported a similar picture in September 2026: RTX 5090 Founders Edition power draw neared 650W under DLSS 5, and the card’s 16-pin connector temperature measured above 90°C.
Readers who want the mechanical details of installing and securing this connector correctly can check our 12VHPWR GPU cable setup guide, and anyone unsure whether their own power delivery is healthy may want to run through our power supply testing walkthrough before assuming a cable is safe simply because it looks fully seated.
Why This Case Is Different From Earlier Melting Reports
Earlier connector-melting cases in 2022 and 2023 mostly centered on whether the cable had been fully clicked into place, and manufacturers generally accepted warranty claims once physical inspection ruled out obvious user error like a bent pin or an unseated plug. What sets the PNY case apart is that the dispute is not really about whether the connector melted. It is about whether using an industry-standard cable that did not physically ship inside PNY’s box voids coverage regardless of whether the cable itself was defective, damaged, or improperly used. It is also increasingly clear this is not a single-brand problem: Hardware.com.br reported in October 2026 that an MSI RTX 5090 Suprim melted while running on a 1,300-watt ASRock Taichi power supply, a separate incident that suggests the underlying connector risk is spread across board partners rather than isolated to PNY’s cards or policies. GizmoChina had already reported in June 2026 that a Club386 RTX 5090 Founders Edition failed despite a native 12V-2×6 cable, and that the failed test used a 1,000W be quiet! Dark Power 13 power supply.
That framing effectively shifts the burden onto the buyer to prove they used the exact cable from the box, a bar that becomes harder to clear the longer a graphics card has been installed inside a case, cables tucked away and boxes long since discarded.
Market Impact: A Bad Look at a Bad Time for GPU Buyers
The timing compounds the reputational risk. RTX 5090 pricing has already climbed well past its original suggested price this year; our earlier coverage tracked the card hitting $4,329 as Nvidia pushed its next-generation RTX 60 series to 2028, and a separate report found resale and grey-market listings for the card touching as high as $9,500 in some regions as AMD gained ground in overall GPU sales share. Tight GDDR7 memory supply, detailed in our report on Micron’s decision to discontinue 2GB GDDR7 chips, is squeezing new-card availability even further.
When a card costs several thousand dollars and is hard to replace quickly at that price, a warranty denial framed around cable branding rather than a verified defect lands differently than it would on a budget part. It pushes buyers to scrutinize not just benchmarks and cooling designs before purchase, but the exact wording of a board partner’s warranty terms, something few shoppers currently do.
It also lands at a moment when secondary-market demand for RTX 5090 cards is unusually high, driven by both gamers chasing the fastest available consumer GPU and AI hobbyists repurposing 32GB cards for local inference workloads. A card that fails outright, with no warranty fallback, is a bigger financial hit in that environment than it would have been at launch pricing back in early 2025, simply because replacing it today costs measurably more.
How AIB Warranty Policies Typically Handle Connector Damage
Board partners generally structure GPU warranties around manufacturing defects rather than damage tied to a specific accessory, and most publish terms that describe coverage in terms of parts and workmanship rather than naming an approved cable list. Reporting on this specific case has not surfaced directly comparable, currently quoted policy language from other RTX 5090 board partners such as Asus, MSI, Gigabyte, or Zotac addressing this exact native-cable scenario, so it would be inaccurate to claim their policies are equivalent to or different from PNY’s interpretation. What is verifiable is narrower: PNY’s support team denied this specific claim in writing, and the wording it used does not appear to match PNY’s own published terms, based on the outlets that reviewed both documents.
Enthusiasts who want a general primer on protecting a card physically, separate from the warranty question, can also see our guide on Cooler Master’s extended three-year GPU Shield warranty program, which reflects a different industry approach: some partners are lengthening coverage windows rather than narrowing what counts as a covered failure.
Key Facts in the PNY RTX 5090 Warranty Dispute
| Detail | Reported Fact | Source |
| Card involved | PNY GeForce RTX 5090 (exact SKU not specified in reporting) | Tom’s Hardware |
| Connector type | 16-pin 12V-2×6 (revised version of 12VHPWR) | Tom’s Hardware |
| Cable used | Native 12V-2×6 cable bundled with an ASRock PG-1600G PSU | Tom’s Hardware, TechSpot |
| What melted | Connector on both the card side and the cable/PSU side | Tom’s Hardware |
| When failure occurred | While the card was reportedly idle, not under gaming load | TechSpot |
| PNY’s stated reason for denial | Only PNY-branded cable or adapter is covered under warranty | Tom’s Hardware, Wccftech |
| Time to denial | Same day the claim was submitted | TweakTown |
| Written warranty terms | Reportedly do not explicitly state a PNY-cable-only clause | Tom’s Hardware, Wccftech |
| Prior PNY guidance | A year-old exchange reportedly showed PNY support approving native PSU cables | Wccftech |
12VHPWR / 12V-2×6 Connector Timeline
| Period | Development |
| 2022 | Nvidia launches the RTX 40 series with the original 12VHPWR connector, replacing multiple 8-pin PCIe connectors with a single 16-pin plug rated up to 600W |
| 2022-2023 | Early melting reports surface on RTX 4090 cards, with outlets including Tom’s Hardware and Gamers Nexus investigating partial seating and pin-contact issues |
| 2024 | PCI-SIG publishes the revised 12V-2×6 connector specification, changing pin geometry to improve contact reliability |
| 2025 | RTX 50-series cards, including the RTX 5090, ship with the newer 12V-2×6 connector as standard |
| September 2026 | A PNY RTX 5090 owner reports a connector melting on both card and cable ends; PNY denies the warranty claim over cable branding, not a confirmed defect finding |
Competitive Comparison: Warranty Terms as a Buying Factor
Graphics card shoppers have historically compared board partners on clock speeds, cooler size, and price premiums over Nvidia’s reference design. This case adds a factor that rarely shows up in buying guides: how a company defines “approved” accessories after the sale. A policy that is not written down clearly, and that support staff apply inconsistently over time, functions as a hidden variable in what a warranty is actually worth. That is a harder thing for a shopper to research than a boost clock, and it is exactly the kind of detail that tends to surface only after something goes wrong.
For a $1,999-plus flagship card, that gap between advertised warranty and applied warranty carries more weight than it would on a budget GPU, simply because the replacement cost of getting it wrong is so much higher.
What Comes Next: Predictions
- Expect PNY to face pressure to clarify or amend its written warranty documentation to either explicitly state a cable-brand restriction or drop the practice, given the gap reporters identified between policy and practice.
- Other RTX 5090 board partners are likely to face direct questions from reviewers and outlets about whether their own warranty terms would treat a native PSU 12V-2×6 cable the same way PNY reportedly did.
- Coverage of this dispute will likely push more buyers to photograph or retain the original bundled cable and box specifically as warranty evidence, an extra step most GPU owners currently skip.
- Given that this is not the first connector-melting story tied to 12VHPWR and 12V-2×6 hardware, expect renewed calls for PCI-SIG or Nvidia to issue clearer guidance on which third-party cables are considered spec-compliant and safe to use.
- If the customer’s account holds up under further scrutiny, this case is likely to become a reference point cited in future warranty disputes involving high-end GPUs, similar to how early RTX 4090 melting reports became a recurring citation in connector-safety discussions.
What Buyers Can Do Right Now
Until board partners clarify their positions, the safest practical step for current RTX 5090 owners is to keep using the exact cable and adapter that shipped in the box, even if a PSU’s own native 12V-2×6 cable is technically spec-compliant. It is also worth photographing the connector’s seating periodically and, where possible, checking its temperature under load, since partial insertion is difficult to spot with a glance once the card is mounted vertically or behind tempered glass; TechPowerUp’s coverage of Hardware Unboxed’s September 2026 testing recorded an RTX 5090 cable running at roughly 175°C, a reading well above what most owners would catch without a thermal camera or infrared thermometer. Buyers evaluating a new card purchase should ask a retailer or manufacturer directly, in writing, whether a PSU’s native cable is covered under warranty before relying on one, rather than assuming a shared industry standard guarantees shared coverage.
It is also worth checking whether a power supply’s included cable is explicitly certified for 12V-2×6 use rather than an older 12VHPWR cable being reused or relabeled, since the two connectors are not universally interchangeable across every PSU generation. A mismatched or worn cable, even one that appears to click into place, can run hotter than a fresh, correctly rated cable over months of use, which is part of why some reviewers recommend periodically reseating and inspecting the connector on any card pulling significant power through a single 16-pin plug.
The Broader Stakes for Nvidia’s GPU Ecosystem
Nvidia does not sell every RTX 5090 directly. Most units reach customers through board partners like PNY, Asus, MSI, Gigabyte, and Zotac, each setting its own warranty terms within Nvidia’s reference design constraints. That structure means a connector controversy tied to one partner’s support policy does not automatically implicate other partners’ terms, although the Founders Edition failure reported in June 2026 shows the connector risk is not limited to board-partner cards, and it does reflect on the health of the 12V-2×6 standard as a whole. If board partners cannot agree amongst themselves, or even internally over time, on which cables are safe and covered, that is arguably a bigger long-term problem for Nvidia’s ecosystem than any single denied claim.
Frequently Asked Questions
What GPU was involved in the PNY warranty dispute?
A PNY GeForce RTX 5090, whose exact retail SKU has not been specified in the reporting reviewed by Tom’s Hardware and other outlets.
Was the customer using a third-party or knockoff cable?
No. The reporting indicates the customer used the native 12V-2×6 cable bundled with their own ASRock PG-1600G power supply, not a splitter, generic adapter, or unrelated third-party brand.
Why did PNY deny the claim?
PNY support told the customer that only the PNY-branded cable or adapter included in the box is covered under warranty, and that damage from any other cable, even a PSU manufacturer’s own native cable, falls outside coverage, according to the response quoted by Tom’s Hardware.
Does PNY’s written warranty actually say this?
Multiple outlets, including Tom’s Hardware and Wccftech, report that PNY’s currently published warranty terms do not contain explicit language restricting coverage to PNY-branded cables, which is the central point of controversy in the story.
Is this the old 12VHPWR connector or the newer 12V-2×6 version?
The reporting identifies the connector involved as the newer 12V-2×6 design, the revised specification PCI-SIG published in 2024 to address issues with the original 12VHPWR connector introduced alongside the RTX 4090 in 2022.
Have other RTX 5090 board partners commented on this specific case?
No verified public statement from Asus, MSI, Gigabyte, Zotac, or Nvidia directly addressing this individual PNY claim has surfaced in the coverage reviewed so far.
How can I protect my own graphics card’s warranty in the meantime?
Keep using the exact cable and adapter that shipped with the card, retain the original packaging where possible, and ask the manufacturer in writing whether a PSU’s native cable is covered before relying on one.
Does a melted connector always mean the cable was defective?
Not necessarily. Prior connector-melting cases have been linked to a mix of causes including partial seating, tight cable bends near the plug, and manufacturing tolerances, and the specific cause in this case has not been independently confirmed in the available reporting.
