TL;DR

For playersOffer from Amazon

Play games on Amazon Luna for your next adventure, included with Prime

• A rotating selection of games, no download needed
• Play on TV, laptop or phone
• Fast, free delivery for your gear, too

Start playing with Prime →Free trial · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

An October report says Nvidia may launch a 24GB RTX 5080 Super, but the card remains unconfirmed. If it becomes the top GeForce card gamers can realistically buy, Nvidia’s lineup could be led by a GPU with 21,760 CUDA cores fewer than the RTX 5090, which has 21,760 in total. Nvidia executive Jeff Fisher has said the 5090 is not being discontinued, but that does not clarify how many cards are available or where its GB202 chips are allocated.

Nvidia has not announced a 24GB RTX 5080 Super, but an October report says a version of the card may arrive and become the GeForce GPU most gamers can actually buy. That would leave Nvidia’s practical gaming flagship with 10,752 CUDA cores, about half the RTX 5090’s 21,760, even though Nvidia executive Jeff Fisher has said the 5090 is not going away.

The reported 5080 refresh is expected to add 8GB of memory to the existing card’s 16GB, while keeping its GB203 silicon. Leaks cited in the report also point to a 415-watt power rating, up from 360 watts on the standard RTX 5080. Those details have not been confirmed by Nvidia, and the card’s name, price, specifications and launch date remain unsettled.

The RTX 5090 uses Nvidia’s larger GB202 Blackwell chip and has 21,760 CUDA cores. The RTX 5080 has 10,752, so the 5090 carries slightly more than twice as many. If the rumored refresh retains the 5080’s chip configuration, additional memory would not bring it close to the 5090 in core count or make it a new version of that silicon.

The report says the RTX 5090’s scarcity has already shaped the market. It cites a price of about $8,000 for the cheapest 5090 and about $1,700 for the cheapest 5080 at the time of writing, compared with a $1,999 launch MSRP for the 5090. These are reported prices at a particular time, not Nvidia’s official pricing or a guarantee of what buyers will find elsewhere.

Memory Could Not Replace Flagship Silicon

A 24GB 5080 could be a more practical high-end option for people whose priority is 4K gaming, especially in demanding games where 16GB of video memory can be a constraint. The report points to Indiana Jones and the Great Circle, where 4K path tracing can strain a 16GB card. More memory may help with workloads that need it, but capacity alone does not establish how the rumored model would perform across games.

The larger issue is the gap between a product’s place in Nvidia’s lineup and its availability. If shoppers cannot find a 5090 at anything close to its MSRP, a 5080 refresh may function as the top consumer choice regardless of the 5090’s specifications. That would make real-world availability, not just the model number, central to which card counts as Nvidia’s gaming flagship.

For buyers, the distinction matters when weighing an upgrade. A larger memory buffer could be useful for certain games and settings, but it would not turn a 5080 into a 5090. Buyers would need independent reviews and confirmed pricing to judge whether the capacity and power changes justify a purchase over the existing 5080 or a competing option.

The 80-Class Gap Has Grown

The reported shift fits a longer change in Nvidia’s product tiers. Two generations ago, the RTX 3080 had 8,704 CUDA cores, against 10,496 in the RTX 3090, or about 83% as many. Both used the GA102 die, with the 3090 using a fuller version of the same chip.

With the RTX 40 generation, the 4080 moved to the smaller AD103 die while the 4090 used AD102. The 4080’s 9,728 cores were about 59% of the 4090’s 16,384. In the current generation, the 5080’s 10,752 cores are just under half the 5090’s 21,760. These core counts show the widening hardware gap, though they do not by themselves predict gaming performance.

The 5090 and Nvidia’s RTX Pro 6000 Blackwell workstation card both use GB202 silicon. XDA Developers’ report argues that Nvidia may have a stronger commercial incentive to allocate some of those chips to higher-priced professional products. That is an explanation offered in the report, not a confirmed account of Nvidia’s production decisions.

Availability and Specifications Remain Unconfirmed

Nvidia has not confirmed the 24GB RTX 5080 Super, its reported specifications, or its release timing. It is also unclear whether the 5090’s scarcity reflects chip allocation, memory supply, production limits, demand, or a combination of factors. The report raises the possibility that Nvidia is struggling to source enough GDDR7 for a 32GB consumer card, but does not establish that explanation.

Rumors of an “SE” RTX 5090 using a cut-down GB202 are also unsubstantiated. Fisher’s reported comment does not rule that possibility in or out, nor does it tell buyers whether 5090 stock or prices will change. The cited store prices may also vary by seller and over time.

CES 2027 Could Clarify GeForce Plans

CES 2027 is identified in the report as the next likely checkpoint for Nvidia’s GeForce lineup. An announcement there could confirm whether a 24GB 5080 is coming, whether it carries Super branding, and how Nvidia positions it relative to the RTX 5090. Until then, the proposed refresh and the rumored 5090 SE should be treated as unconfirmed.

If Nvidia does announce the card, buyers will need to compare its price, availability and independent performance tests with the existing 5080 and 5090. Those details will show whether added memory changes the card’s value for games that exceed 16GB, or whether the upgrade mainly changes the lineup’s advertised capacity.

Key Questions

Has Nvidia confirmed a 24GB RTX 5080 Super?

No. The card and its reported specifications remain leaks; Nvidia has not announced them.

How many CUDA cores does the RTX 5090 have compared with the RTX 5080?

The RTX 5090 has 21,760 CUDA cores, while the RTX 5080 has 10,752. That puts the 5080 just under half the 5090’s core count.

Would 24GB of memory make the rumored 5080 as powerful as a 5090?

No such performance conclusion can be drawn from the reported memory change. More VRAM could help in games that need more than 16GB, but the rumored card is expected to retain the 5080’s GB203 chip, rather than use the 5090’s GB202.

Is Nvidia discontinuing the RTX 5090?

Jeff Fisher has said the 5090 is not going away, according to XDA Developers. That statement does not resolve how many cards will be available or at what prices.

When might Nvidia provide more information?

CES 2027 is the next expected checkpoint mentioned in the report, but Nvidia has not confirmed an announcement date or a product launch.

Source: XDA Developers