Just as all hardware fans were waiting, during this early morning NVIDIA has made its appearance in the CES of Las Vegas and, as it could not be otherwise, among many other novelties it has presented its new generation of graphics cards. In this article we are going to summarize everything they have said about them, including of course their price and specs techniques.
The RTX 50 are now official, although it must be said that NVIDIA has dedicated less time to it during its conference than we expected. In any case, we can finally stop rumors, assumptions and leaks and we can start handling official information.
The quartet of NVIDIA RTX 50 Series graphics
There are four graphics cards presented: RTX 5090, RTX 5080, RTX 5070 and RTX 5070 Ti. As you might already guess, we were all surprised to have an entry-level Ti model instead of introducing the RTX 5060, which will presumably arrive a little later. All graphics cards carry new generation Blackwell chips, have a PCIe 5.0 interface, GDDR7 memory and DisplayPort 2.1a video output, as well as a single 16-pin power connector.
Starting with the RTX 5090the top-of-the-range model, has a recommended price of $1,999 which will then translate with taxes and so on into around 2,400 euros surely. It mounts a GB202 chip with 21,760 CUDA cores, being the first to overcome the 20,000 core barrier, along with 32GB GDDR7 under a 512-bit bus. It has a TDP of 575W, basically confirming all the specifications that we already knew based on leaks.
In addition, the Founders Edition of the RTX 5090 has a new design, and despite consuming 125W more than the RTX 4090, it is “only” double slot, that is, it is a little smaller than the current models.
We continue with the second most powerful model, the RTX 5080which mounts a Blackwell GB203 chip with 10,752 CUDA cores, 16GB GDDR7 of dedicated graphics memory under a 256-bit bus, and contrary to what the rumors said, they have not said anything about a 24 GB variant. It has a TDP of 360W (40W more than its predecessor) and its price is 999 dollarsso we can expect prices in Spain of between 1,200 and 1,300 euros.
We continue with the RTX 5070 Tiwhich has 8,960 CUDA cores and a memory configuration of 16GB GDDR7 which, according to NVIDIA, is twice as fast as the memory included in the RTX 4070. Its recommended retail price is $749so we can expect prices of about 900 euros.
Finally we have the “cheap” model, the RTX 5070which will have 6,144 CUDA cores and will be sold for $549so we expect prices of 650-700 euros. The “king of the mid-range” model, the RTX 5060, is still missing, which should of course be sold with a lower price tag. Oh, and NVIDIA has said that the RTX 5070 Ti will not be sold with a Founders Edition model, so we will only see it directly from the assemblers.
As for the release dates, the RTX 5090 and RTX 5080 will arrive on January 30, 2025, while the RTX 5070 Ti and RTX 5070 will arrive a little later, in February they have said. Below we leave you a summary table of all the specifications.
RTX 5090 | RTX 5080 | RTX 5070 Ti | RTX 5070 | |
---|---|---|---|---|
GPU | GB202-300 | GB203-400 | GB203-300 | GB205-300 |
SMs | 170 | 84 | 70 | 48 |
CUDA cores | 21,760 | 10,752 | 8,960 | 6,144 |
Memory | 32GB GDDR7 | 16GB GDDR7 | 16GB GDDR7 | 12GB GDDR7 |
memory bus | 512 bit | 256 bit | 256 bit | 192 bit |
Memory speed | 28Gbps | 30Gbps | 28Gbps | 28Gbps |
Bandwidth | 1,792 GB/s | 960GB/s | 896GB/s | 672 GB/s |
Feeding | 1×16 pin | 1×16 pin | 1×16 pin | 1×16 pin |
TDP | 575W | 360W | 300W | 250W |
Interface | PCIe 5.0 x16 | PCIe 5.0 x16 | PCIe 5.0 x16 | PCIe 5.0 x16 |
video outputs | 3x DisplayPort 2.1a 1x HDMI 2.1 |
3x DisplayPort 2.1a 1x HDMI 2.1 |
3x DisplayPort 2.1a 1x HDMI 2.1 |
3x DisplayPort 2.1a 1x HDMI 2.1 |
Recommended price | $1,999 | 999 dollars | $749 | $549 |
Estimated price in euros | €2,300-2,400 | €1,200-1,300 | €900 | €650-700 |
Launch | January 30, 2025 | January 30, 2025 | February 2025 | February 2025 |