Geforce RTX 5090 and 5080: New leak reveals all the important details

The Nvidia flagships in the form of the Geforce RTX 5090 and RTX 5080 are scheduled to be released this year. However, the company has not yet provided any official information about the graphics cards – but the reliable leaker “kopite7kimi” on X (formerly Twitter) published all the relevant specs of the RTX 5090 and 5080.

As a result, the RTX 5090 should, as expected, rely on the GB202 chip, which, however, is not used in full configuration. Has 21,760 CUDA cores instead of the Nvidia flagship. On the memory side, the leaker mentions 32 GB of GDDR7 VRAM, which is connected via a 512-bit interface. The power consumption should be a whopping 600 watts – and yet only a 2-slot cooler should be sufficient for heat dissipation.

The Geforce RTX 5080 should be a little more discreet. With 10,752 CUDA cores, around half of the shader units are present in the GB203 chip used. The memory is also exactly halved: there should be 16 GB of GDDR7 memory for a 256-bit interface. This comment is also the biggest point of criticism in the comments under the leak, since it has the same amount of memory on board compared to the RTX 4080 Super and the jump in performance will probably be limited. The TDP of 400 watts is also viewed critically, which is dangerously close to the previous top model RTX 4090 (here: 450 watts).

While the technical data of the Geforce RTX 5090 and RTX 5080 are stated more specifically for the first time, two pieces of information are still missing: Neither a release date nor a possible price were mentioned in the leak from kopite7kimi. There is speculation that the introductory price will be slightly higher than that of its predecessors, which came onto the market at an RRP of 1,949 euros (RTX 4090) and 1,469 euros (RTX 4080).