Nvidia has removed the GeForce RTX 5090 from its official marketplace, and market rumors suggest the 16GB RTX 5080 is also set to be discontinued. In its place, according to the reports, would come an RTX 5080 with 24GB of video memory, which would become the most powerful card in the Blackwell generation until the Rubin series arrives, expected in 2028.
The 24GB 5080 would use the same graphics die as the current 5080. That chip is already fully utilized, so this would not be a more powerful Super variant or a noticeable performance jump. The only expected difference is memory, going from 16GB to 24GB of GDDR7 using 3GB modules. A slight clock increase might come too, but it would be minimal in real-world gaming.
The RTX 5090 had already disappeared from some retailers, including Micro Center, and its absence from the marketplace has fuelled the idea of a permanent exit. The picture is confused, however: in mid-October Nvidia denied stopping the 5090, with Jeff Fisher confirming the card is still being produced and shipped, while admitting limited production capacity. Earlier rumors had also suggested the GB202 chip would be reserved for professional cards.
A 24GB RTX 5080 is not a new idea: AlexTech had already covered it in 2024, before the Blackwell launch, and the same configuration was later linked to the RTX 5080 Super. In July 2026 the RTX 50 Super had reached at least one partner, but Nvidia had suspended the launch due to the cost of 3GB GDDR7 modules.
The window indicated by the sources is early 2027, with availability expected between December and January and a possible launch around CES. Nvidia has not commented, and it usually does not confirm rumors. For now, the only firm data concerns the specifications of the current RTX 5080, while the rest remains within the realm of supply-chain reports.
For buyers, the picture is clear: the 16GB 5080 seems headed out, while the 24GB version promises more memory for AI models and heavy workloads without changing computing power.
No comments yet. Be the first!