The GeForce RTX 4090/4080 launch was marred by faulty 12VHPWR power connectors, resulting in burnt cables and melting PCBs. After months of speculation and third-party investigations, the issue was chalked down to poor manufacturing design and user neglect. Many AIBs reacted to this situation by updating the 12VHPWR adapter design to prevent half-inserted connections. Unfortunately, isolated cases of 16-pin connector-related burning/melting are still being reported.
The GeForce RTX 4090 Founders Edition and AIBs are reported as the primary victims of the 12VHPWR adapter burnings. In all the latest reporting, the graphics card had been in use for several months at the time of the incident. All three users claim that the adapter was wedged firmly in place, as recommended by NVIDIA a few months back.
Interestingly, one of the posters used the RTX 4090 (ASUS TUF Gaming) for neural net training, drawing 350W to 500W of power (TBP: 350W). They noted that their second RTX 4090 (a Gigabyte Gaming OC) worked perfectly alongside a third-party 12VHPWR adapter.
The second report comprises a midrange PC paired with a particularly high-end GPU. We’re looking at a Ryzen 5 7600 (~$200) paired with a GeForce RTX 4090 FE ($1,599) alongside a 750W PSU (Corsair SF750 80 Plus Platinum SFX). While a lower-wattage power supply can lead to crashes in particularly intensive titles, damage to the GPU is unlikely.
Either way, the Redditor had limited the power limit of their RTX 4090 to 315W (70%), making the 750W power supply a non-issue (850W is the min recommended). Furthermore, it wasn’t just the GPUs-side 16-pin power connector that was damaged, but the PSU-side dual 8-pin connectors were also burnt. The system was functional for over six months before the failure.
The third one is another GeForce RTX 4090 Founders Edition which was still functional at the time of the discovery. The Redditor noticed the damage to the adapter while between cases. Even though the power connector and the cable were burnt, the GPU was still working after the melting had begun.
At the end of the day, the number of GPUs affected by the 12VHPWR issue is less than 0.1%, making you wonder whether the connector’s design made it prone to being half-inserted. Or it could be a bad batch with debris lodged into the plug, resulting in the melting down the line. Either way, we’ll keep seeing such reports until the first batch of cards get past the warranty curve.