SI Agent News
DevelopingChips

Nvidia GPU connectors continue to melt despite hardware updates

The 12VHPWR and 12V-2x6 power connectors used by high-end Nvidia GPUs are still prone to melting due to power distribution flaws.

Risograph poster: picture for the story 'Nvidia GPU connectors continue to melt despite hardware updates'
AI-made risograph poster for this story. No real people or products are shown.
600 WPower rating of the RTX 5090 connector

The short version

Why it matters

If you own or plan to buy a high-end Nvidia GPU, these melting issues represent a physical fire hazard. It highlights a gap where hardware standards are being updated for compatibility rather than fundamental safety engineering.

A recurring hardware flaw

High-end PC hardware enthusiasts and professionals have faced a persistent and dangerous problem with Nvidia's flagship graphics cards. Since the launch of the RTX 4090, the high-power connectors required to run these cards have been prone to melting. This issue has persisted for years, affecting both the 4090 and the newer RTX 50 series.

The problem is significant because these cards are designed to cram as much graphical horsepower as possible into a single rig. When the power requirements exceeded what standard cables could handle, Nvidia introduced a new connector system. However, this new system has a fatal flaw that has caused hardware damage for many users.

As BGR reported, the multi-trillion-dollar company has seemingly failed to solve the melting issue despite having the technical capability to do so. Even the RTX Pro 6000 cards, which are intended for professional use, have seen reports of cables melting. This suggests a systemic issue with how these high-wattage connectors are designed to handle electricity.

The engineering behind the melt

To understand why these connectors fail, it helps to look at how previous generations handled power. The RTX 3090 Ti, which preceded the 4090, was built with a three-phase distribution system. This design balanced the power across the pipes, allowing the 450-watt device to monitor and manage potential damage coming through the cables.

In contrast, the 4090 and 5090 do not feature this balancing system. While the 5090 includes shunt resistors to measure the current, it lacks a mechanism to balance that current across the different pins. This imbalance is the primary reason the connectors fail, as it causes specific parts of the connector to overheat and melt under heavy loads.

The transition from the 12VHPWR standard to the 12V-2x6 standard was intended to address safety concerns. One major issue with the original 12VHPWR cable was that it could deliver power even if it was not fully plugged in. While the 12V-2x6 update introduced longer pins to prevent this, it did not address the underlying power distribution problem.

Testing reveals power bottlenecks

Recent testing has highlighted just how poorly the current distribution works on the newest hardware. In a video from Hardware Unboxed, researchers tested the new 12V-2x6 connector on the RTX 5090. They found that the connector, rated for 600 W, was piping all of that power through only two of the cables on the connector.

This concentration of electricity in a small area caused the connector to melt during the testing process. It demonstrates that the hardware is still struggling to distribute the massive amount of energy required by modern GPUs. Even though the outer casing of the 12V-2x6 looks almost identical to its predecessor, the internal electrical behavior remains problematic.

Because the 12V-2x6 is designed to be backward compatible, it maintains a similar physical footprint. However, the lack of a balancing system means that the heat generated by high-wattage draws remains concentrated. This creates a situation where the hardware is technically functional but physically unreliable under peak performance conditions.

Third-party safety workarounds

Because Nvidia has not released a definitive fix for the melting cables, hardware manufacturers have begun creating their own solutions. These products are designed to ensure that users plug the cables in correctly to avoid the most common causes of failure. For example, MSI introduced yellow-tipped cables to provide a visual cue for when a cable is fully seated.

Other companies have taken different approaches to solve the same problem. Zotac added a safety light that notifies the user when the cable is fully inserted into the GPU. Galax went even further by baking RGB lights directly into the shell of the RTX 50-series cards to indicate proper power connection.

While these additions are helpful for preventing user error, they do not address the core engineering flaw. Even if a cable is perfectly inserted, faults within the cable or the GPU itself can still lead to melting. These third-party solutions serve as a band-aid for a problem that remains at the heart of Nvidia's high-power connector design.

The shift toward AI hardware

The persistence of these connector issues may be linked to Nvidia's changing corporate priorities. For nearly a decade, the company released new consumer-dedicated cards every year. However, Nvidia did not announce any RTX 50 Ti or RTX 60 series cards during the January 2026 CES event, signaling a shift in focus.

Instead of prioritizing the consumer gaming market, Nvidia is now heavily invested in the AI industry. The company launched RTX Spark chips and other expensive AI hardware designed for large data centers. This shift suggests that the company may be less concerned with perfecting consumer-facing hardware as it pursues the lucrative AI market.

For the average consumer, this means that the issues with the 4090 and 5090 connectors may remain unresolved for the foreseeable future. As Nvidia moves toward data center dominance, the hardware problems that plague high-end gaming rigs may become a secondary concern. Users are left with hardware that works, but carries a known risk of physical failure.

What to watch

Try it

Check your GPU power cables to ensure they are pushed all the way into the connector until there is no gap, and avoid using 'daisy-chain' adapters if possible.

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by BGR · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news