Ahead of the 2025 CES, a large number of Nvidia RTX 5090, 5080, and 5070 GPUs were leaked.

Ahead of the 2025 CES, a large number of Nvidia RTX 5090, 5080, and 5070 GPUs were leaked.

It seems certain at this point that Nvidia will unveil its RTX 50 series GPUs at the 2025 CES. No one can say for sure, as there are only a smattering of leaks and rumors, but thanks to a fortuitous revelation we have the clearest indication.

PC gaming company Zotac added (and quickly removed) the GeForce RTX 5090, 5080, and 5070 GPUs to its website, but not fast enough for Video Cardz to take screenshots. But it wasn't fast enough that Video Cardz didn't take a screenshot. In addition to the name, we also got a glimpse of some of the more important specs for high-fidelity gameplay.

In short, I'm excited about the pure horsepower of these, but my electric bill is going to make me sweat a little. But let me elaborate a bit more.

Through the website's online store filter and Google search results meta description, I was able to identify the following five GPUs:

Incidentally, the D variant of the RTX 5090 is a model that will only be available in Japan.

Yes, Palpatine may have exaggerated a bit here, but the main specification of the top-of-the-line RTX 50 series model is the use of GDDR7 memory instead of the GDDR6 seen in the 4090 and others.

The main reason this is important is that it could potentially provide up to 60% faster on-the-fly access to critical data, while consuming about 20% less power. Besides, other rumors point to a possible increase in the amount of CUDA cores by up to 33%

.

This would dramatically increase the number of parallel computations a GPU can handle. This means that your shiny new card will be able to balance all kinds of graphics processing, from maintaining fidelity at high resolutions while doing ray tracing and path tracing calculations. Faster and more power efficient GPUs are the life of the next generation of GPUs.

Returning to the Zotac leak, take the RTX 5090 as an example, it has 32 GB of GDDR7 memory. This is the part that makes me a little nervous because it nullifies the 20% reduction in power demand due to increased storage.

In other words, if the TGP of the RTX 4090 is 450W, the RTX 4090 would be 480W, and this is before considering the possibility of overclocking.

Considering average US electricity rates (16.83 cents per kilowatt-hour), the running cost could be 40 cents per day, which is significantly higher than any gaming console or integrated graphics option (before considering other power consumption of the system ) could really start to add up.

Don't get me wrong; I'm not saying that the power consumption of a dedicated GPU is the same as a dedicated GPU. But you should consider choosing between the pure horsepower of a dedicated GPU or the superior balance of an integrated option.

This is a good question. We reported the RTX 5060 leak yesterday, but it was not in the Zotac lineup. I am still quietly confident that we will see this in some of the announcements from Nvidia (Zotac may be avoiding this for the initial launch), but time will tell.

Categories