Direct successors to the RTX 50 series are already in the spotlight. The NVIDIA RTX 50 SUPER series is in advanced development. and could see the light of day sooner than many expected. Although it was originally rumored to debut in January 2026 during CES, reliable sources now point to a significant advance.
According to media such as TweakTown, NVIDIA plans to launch its new SUPER graphics cards in the fourth quarter of 2025, just in time to take advantage of the holiday sales. If confirmed, it would be an unusual strategy for the company, which doesn't usually update its models so soon after the standard version.
An early release with minor changes
What we know so far indicates that The differences between the RTX 50 series and the SUPER versions will be limited.. A significant jump in gross performance is not expected, since The models mostly maintain the same number of CUDA cores than the base version.
The main improvement will focus on memory, which will become GDDR7 in 3 GB modules per chip, allowing for wider configurations without changing the bus width. This will facilitate board design and could reduce manufacturing costs.
Along these lines, SK Hynix has publicly confirmed that already working on 7Gb GDDR24 chips (3 GB) that will power this new generation. This is not only relevant for gaming, but also for professional applications and artificial intelligence, areas in which NVIDIA has been gaining ground.
Therefore, the strategy seems clear: more memory and minor efficiency improvements to cover certain specific needs, especially in tasks that do consume large amounts of VRAM.
Expected models and leaked specifications
Currently, there are three models that are strongly rumored to lead this update: RTX 5080 SUPER, RTX 5070 Ti SUPER and RTX 5070 SUPERThere is no concrete data on a possible RTX 5090 SUPER or 5060 SUPER at the moment.
La RTX 5080 SUPER would maintain the 10.752 CUDA cores of its predecessor, but would go from 16 GB of VRAM to 24GB of GDDR7An increase in power consumption is also expected, reaching 415W compared to the standard model's 360W. This increase in memory would also slightly increase bandwidth from 960GB/s to 1024GB/s.
La RTX 5070 Ti SUPER would retain its 8.960 CUDA cores, but would add an additional 8 GB of VRAM, going from 16 to 24 GB GDDR7Its consumption would also rise to 350W.
Por último, la RTX 5070 SUPER would be the only one with real improvement in cores: would increase to 6.400 from the 6.144 of its standard version. It would also gain in memory capacity, going from 12 GB to 18GB of GDDR7, using six 3 GB modules.
These updates, while modest, could have an appreciable impact on memory-intensive applications, although In traditional games the difference could be almost non-existent except in very demanding or poorly optimized titles.
A step forward in memory, rather than performance
The fact that the SUPER models barely change their processing capabilities makes the focus of this update clear: Improve the working environment for VRAM-intensive tasksThis could include everything from high-resolution video games with complex textures to advanced editing or productivity applications.
Furthermore, NVIDIA seems to be taking advantage of the moment to adapt its product line to the new higher-density GDDR7 memories, available from manufacturers such as Samsung and SK Hynix. This will allow for configurations such as 24 GB on 256-bit buses or 18 GB on 192-bit buses, previously unpublished in cards of its range.
Although, despite the memory increase and some minor improvements, many analysts agree that The RTX 50 SUPER does not represent a revolution compared to their standard counterparts. For those who already purchased the RTX 50 at launch, this upgrade might not justify a further outlay.
Those who have not yet upgraded their GPU will have to consider whether The extra VRAM really makes up for a potential price increase, especially considering that in past launches, initial market prices were quite far from the official figures proposed by NVIDIA.
The move to move up the launch reinforces the idea that NVIDIA is looking to give an extra boost to sales before the end of the year, perhaps as a reaction to lower than expected demand for the standard models or in preparation for the arrival of the RTX 60 already on the horizon.
With several sources pointing to an arrival between October and December 2025, everything indicates that We will see these cards before the expected CES 2026Until then, the available data is the result of leaks, although they come from sources with fairly reliable backgrounds.
Updates are intended for those who need more memory without having to upgrade to the highest rangeThis makes them a middle-of-the-road option for content creators, demanding gamers, or developers looking to get the most out of their rigs without changing architecture.
