Talking about the future GeForce RTX 50 may seem a bit hasty because, in reality, we are still waiting for the entry-level models of the current generation, the RTX 4060 (which will debut this month) and the RTX 4050, to arrive on the market. , perhaps in the absence of some improved revision of them, we can consider the family completed, which, let us remember , was initially presented about nine months ago , in September 2022.
However, a reference shown by Jensen Huang, CEO of NVIDIA , at the Computex 2023 kickoff event, has caused rumors and speculation to skyrocket. It was, specifically, a reference to the still unpublished Hopper Next architecture . Let’s remember, for some context, that Hopper is currently the company’s architecture for data centers and other advanced computing platforms.
However, if we go back in time, you will remember that when Ampere was already aiming to be the architecture of the (then future) RTX 30, the architecture called to succeed it was expected to be Hopper . However, some time later we learned that it would be Ada Lovelace who would succeed Ampere, since Hopper would be delayed with respect to the initial plans . Thus, there are many who think that Hopper will be the architecture chosen for future GeForce RTX 50.
As a general rule, the speeches given by CEOs when they get on stage have between nothing and absolutely no improvisation when it comes to the data they share. In other words, if Huang mentioned Hopper Next, which we can understand is nothing more than an evolution of the current Hopper, aimed at large systems, it is because this iteration of said technology will play a very important role in the medium term for NVIDIA.
Thus, various analysts and experts in the field have decided to join the dotted lines and the conclusion they have reached, and although we will still have to wait to confirm it seems plausible, it is that Next Hopper is nothing other than Blackwell, the code name from the upcoming line of desktop GPUs that have been rumored for some time now to jump to 3 nanometers and stick with a monolithic design. And, of course, we can take it for granted that a good part of its improvements with respect to the current generation will come not so much in terms of brute force (there will also be something, of course), but of even more specialization in artificial intelligence.
Thus, when the company itself already “advances” (albeit apocryphally) its plans with respect to Hopper Next, that is, Blackwell, we can conclude that NVIDIA has decided to repeat the life cycle of two years per generation. , which makes us expect that the future GeForce RTX 50 will see the light of day sometime in 2024 , probably during the fourth quarter of the year.