A 24GB Nvidia GeForce RTX 5080 will follow the 16GB graphics card, says leak

There will reportedly be two versions of the Nvidia GeForce RTX 5080, an initial one with 16GB of GDDR7 of VRAM, and a second one with 24GB of memory that will come later. If it’s true, this rumor has echoes of the GeForce RTX 4080 GPU launch, with that chip originally set to come in 12GB and 16GB graphics cards. However, the former was later “unlaunched” due to the negative reaction of PC gamers to its underwhelming specs.

The new Blackwell-based GPU is expected to launch alongside the GeForce RTX 5090 at the start of 2025, with the latter expected to be the best graphics card in the Nvidia gaming stable. However, the most-recently leaked GeForce RTX 5080 specs don’t paint a flattering picture of Nvidia’s new second-tier GPU, with it looking pretty weedy compared to the RTX 5090, thanks to half the VRAM and less than half the number of CUDA cores.

It’s the former that could potentially be resolved if this latest leak is true. This particular rumor comes from Chiphellforum user PolyMorph, who stated that the “5080 will be released in 24G later” (Google Translate from Chinese) in a post on the forum. Of course, none of this has been officially confirmed by Nvidia, so don’t take this as fact, but PolyMorph has previously turned up some reliable Nvidia leaks, including sharing a photo of the GeForce RTX 3090 GPU with the correct “GA102-300-A1” codename, before it was released.

As Wccftech notes, Nvidia could potentially make a 24GB version of the RTX 5080 without changing the rest of the GPU’s spec too, as GDDR7 can be supplied in both 2GB and 3GB chips, meaning the new memory could be a drop-in replacement without having to change the width of the memory bus, or indeed the number of memory chips surrounding the GPU.

That’s a much better situation than we had with the RTX 4080 12GB and 16GB, where the two graphics cards didn’t just have a different amount of VRAM, but also a lesser spec, with a narrower memory interface and fewer CUDA cores on the 12GB card (which was later released as the GeForce RTX 4070 Ti).

This time, once 3GB GDDR7 chips are available in decent yields, Nvidia could upgrade the RTX 5080 from 16GB to 24GB without changing the core spec elsewhere, and without adding the huge extra cost of making a 32GB card. As such, we could expect the Nvidia GeForce RTX 5080 16GB to launch first, and a second RTX 5080 24GB to arrive later, perhaps with an overclock on the GPU as well.

If you’re interested in what’s in store with Nvidia’s forthcoming new GPUs, check out our full Nvidia GeForce RTX 5090 guide, where we run you through everything we know so far about the new flagship graphics card.