Everything we (sort of) know about the Geforce RTX 5000 series

Another graphics card generation is coming to an end as 2024 very soon turns into 2025. However, the generation has been eventful with increasingly advanced upscaling technology, the move to Unreal Engine 5 and the explosion of generative AI. How these aspects shape the next generation, we will probably find out very soon as Nvidia’s Geforce RTX 5000 series is apparently on the way.

When will the Geforce RTX 5000 series actually appear?

Much points to an unveiling during CES 2025. At 03:30 Swedish time on the night of January 7, Nvidia’s CEO Jensen Huang steps onto the Mandalay Bay stage in Las Vegas. It is likely that there will be a lot of talk about AI and data center hardware, but it is equally likely that we will get the first glimpse of the Geforce RTX 5000 series.

Looking back at the Geforce RTX 4000 series, the launch began with the flagship Geforce RTX 4090 taking the stage, followed by the RTX 4080 and finally the RTX 4070 Ti. Looking back another generation, the green giant started by releasing the Geforce RTX 3080, then the RTX 3090 and finally the RTX 3070.

Based on previous generations, a potential Geforce RTX 5090 or RTX 5080 will be released first, followed by a rolling launch with more models afterwards. It is likely that we will see the first launches soon after CES in the first quarter of 2025.

How good will the RTX 5000 series be?

Of course, we don’t know 100 percent what kind of specifications will be offered before the new models are officially unveiled, but there are detailed information from historically reliable sources which indicates the following setup:

If we start with our eyes turned towards the top model Geforce RTX 5090, it looks like it will be a flagship that lives up to the true meaning of the word. The large GB202 chip appears to be equipped with a whopping 21,760 CUDA cores. However, we do not yet know how high clocked the graphics chip is, but it does appear to be paired with 32 GB of GDDR7 graphics memory.

The memory speed of 28 Gb/s over the 512-bit wide memory bus gives a total memory bandwidth of 1,792 GB/s – a marked increase of 77 percent over the slightly more than 1,008 GB/s of the previous worst Geforce RTX 4090.

However, it is not only the memory that gets a boost, but also the graphics card’s maximum power budget (TBP), where this increases by 150 W to a total of 600 W. However, this does not necessarily mean that the card will have a power output of 600 W when under load. Instead, we hope to see it behave like the RTX 4090, where the power draw at default settings is significantly lower than the card’s TBP value.

If we look one step down in the hierarchy, we find the Geforce RTX 5080 and RTX 5070 Ti, which appear to be based around the GB203 graphics chip. The RTX 5080 appears to have 86 SM clusters versus the RTX 5070 Ti’s 70, resulting in both having 10,752 and 8,960 CUDA cores respectively. This is a real difference compared to the large GB202 circuit in the RTX 5090, which seems to result in a fairly large performance difference.

The amount of memory for the Geforce RTX 5080 and RTX 5070 Ti looks to be the same 16 GB GGDR7 over a 256-bit bus. However, the memory speed of the RTX 5070 Ti is not easy to find right now, so we cannot say how the total memory bandwidth will differ between the two. In addition, they both get an increased power budget compared to their predecessors, where the cards are 400 and 300 W, respectively.

Although not much is known about the specifications of the Geforce RTX 5070, it is probably not too bold to assume that the RTX 5080 and RTX 5070 Ti will be closer to each other in performance than the RTX 5070 Ti and RTX 5070.

The card that stands out the most on paper, and which will rule the roost in terms of performance, is undoubtedly the Geforce RTX 5090. It’s a massive jump up in CUDA cores compared to the RTX 5080 and the card has an awfully large amount of graphics memory with sky-high bandwidth . Not entirely unexpectedly, the card will be aimed at those who, in addition to games, work on their machine with, for example, AI workloads or video production. The Geforce RTX 5080 and RTX 5070 Ti look to be graphics cards aimed at enthusiasts and those who game at higher resolutions.

What will the RTX 5000 series cost?

The price is of course one of the most interesting points. In short, we can currently say that we do not know much about what the plans are for the Geforce RTX 5000 series in terms of price levels. However, this does not stop us from speculating.

There is probably very little doubt that the Geforce RTX 5090 will be an expensive rascal, but exactly how expensive we at the editorial office have a few shared theories about. Taking a free guess, our numbers look like this:

The prices we are speculating about are then the Founders Edition prices, so simply a guide price. It is likely that there will be various cheap partner models which for the RTX 5090 will result in prices that really stand out.

We will see a relatively large price jump between the RTX 5090 and RTX 5080 as the circuit in the RTX 5090 is so much larger. In addition, the Geforce RTX 4080 was very highly priced at launch, which was also heavily criticized by enthusiasts and reviewers. At the same time, however, we believe that Nvidia will not make any major price cuts in the enthusiast segment, as there will likely be a lack of competition from AMD at the top.

When do you think the Geforce RTX 5000 series will appear and what do you think the graphics cards will cost? Discuss in the thread!