Nvidia’s RTX 5090 GPU is one of the most searched-for GPUs that is not publicly announced by the company but still comes with its fair share of leaks. Some of these are simply due to the sheer scale of the product lineup it is a part of in addition to the number of AIB partners involved and the general excitement around the product.
Nvidia’s RTX 5090 is expected to be announced during Nvidia’s keynote event during CES 2025 next month and is purported to be based on its new Blackwell architecture. It is expected to come with 32GB of next-generation GDDR7 memory, support for PCI-E 5.0, and deliver performance that dwarfs that ofthe incumbent RTX 4090thanks to a mix of hardware and software enhancements across the board.
Nvidia’s RTX 5090 PCB Leaks
Given the sheer amount of excitement surrounding the RTX 5090, it is not particularly surprising that much like its predecessor, gamers and industry professionals are being treated to plenty of leaks thanks to the number of players involved in bringing what is Nvidia’s flagship GPU to the market.
The most recent leak might be the most consequential of them all, confirming the PCB layout for what seems to be a 3rd party RTX 5090 motherboard, showcasing space for a large die while also confirming both, native PCI-E 5.0 support and the use of 16 GDDR7 memory modules, presumably in a 2GB x 16 configuration.
Pictures of the alleged RTX 5090 motherboard (Source: Chiphell.com)
The pictures, which were first visible on the Chinese forum Chiphell.com confirm much of what rumors already indicated: The RTX 5090 is a behemoth of a GPU, and the leaked design, which seems to be from a PNY variant, not Nvidia’s Founder’s Edition, indicates that a single 16-pin connector is in play for its power needs.
The Nvidia GeForce RTX 5090: Plenty Of Questions Before Release
The RTX 5090 has plenty of questions surrounding its release thanks to it essentially being considered the herald of the next generation of desktop-based GPUs with Nvidia historically pushing its flagship GPUs first.
It is rumored to cost anywhere between $2000 and 2500 (The RTX 4090 cost $1600 at launch for the elusive Founders Edition) and comes in 2 configurations (16GB and 32GB). There is also credible news about a China-specific RTX 5090D which is expected to be a cut-down version of the RTX 5090 that still offers comparable gaming performance but might adversely cap its power for AI-centric workloads to comply with the US administration’s sanctions on China and other countries.
It is expected to come with Nvidia’s next iteration of its DLSS technology with most dubbing it DLSS4 to match Nvidia’s previous naming norms. It is expected to be Nvidia’s de facto flagship GPU for high-end consumers who do not currently need a server-grade Blackwell-based solution from Team Green.
The Nvidia RTX 5090 is expected to be unveiled on the 6th of Jan 2025 as part of Nvidia’s keynote where CES Jensen Huang is expected to take the stage. For users who can’t wait for CESconsider checking out our guidecomparing Nvidia’s current flagship with its lower-end sibling, the RTX 4080 SUPER instead.
