The NVIDIA GeForce RTX 5090 has a 14-layer PCB with 16+6+7 phase power delivery.
According to new information, NVIDIA’s next top “Blackwell” graphics card, the GeForce RTX 5090, will have different board architecture and power delivery than its predecessors. Benchlife claims that the next Blackwell-based GPU will have a novel 16+6+7 power stage design, which will be different from the 20+3 phase setup of the RTX 4090. The report’s indication of a 600 watt TGP supports past conjecture regarding the card’s power requirements. Although the actual TDP of the GB202 chip may be lower, this specification refers to the whole power allocation for the graphics subsystem. Because GDDR7 memory modules and power delivery are more complicated, the RTX 5090 will come with 32 GB of next-generation GDDR7 memory and use a 14-layer PCB. GPUs typically have a maximum of 12 layers for sophisticated overclocking schemes.
With PCI Expres

s 5.0 x16 interface compatibility and a 12V-2×6 power connector design, the next GPU will fully embrace contemporary connectivity standards. An early PNY RTX 5090 model with 40 capacitors but an ambiguous power delivery configuration caught our attention. NVIDIA is pushing the limits of power delivery and signal integrity for its next-generation flagship with more PCB layers and additional power phases. There are still concerns regarding the larger RTX 50 series lineup, even though these specs depict a potent gaming and professional graphics solution. We must wait till the CES-rumored introduction to find out how the 12V-2×6 connector will be implemented across various models, especially those under 200 W.