TL;DR
Play games on Amazon Luna for your game nights, included with Prime
- A rotating selection of games, no download needed
- Play on TV, laptop or phone
- Fast, free delivery for your gear, too
XDA Developers reports an unconfirmed rumor that Nvidia may release a 24GB RTX 5080 Super, potentially leaving it as the GeForce card gamers can readily buy while the RTX 5090 remains scarce. The rumored card would retain the RTX 5080’s GB203 chip, whose CUDA core count is less than half the RTX 5090’s; Nvidia has not announced the Super model or confirmed the leak.
Nvidia has not announced a new GeForce flagship, but an unconfirmed report about a possible 24GB RTX 5080 Super has raised the prospect that a more available 80-class card could head the company’s gaming lineup while the RTX 5090 remains hard to buy. XDA Developers reported the rumor on October 22, 2026; if accurate, the card would use the RTX 5080’s GB203 chip rather than the RTX 5090’s larger GB202, leaving it with less than half the flagship’s CUDA cores.
The reported 5080 Super would keep the GB203 silicon used by the existing RTX 5080, according to the October report. Leaks described a move from 16GB to 24GB of memory and a power rating of 415 watts, up 55 watts from the standard card’s 360 watts. Nvidia has not confirmed those specifications, a launch date or even that the model will be released.
The core-count difference is substantial: the RTX 5090 has 21,760 CUDA cores, just over twice the RTX 5080’s count. Since the RTX 5080 already uses the full GB203 die, a memory increase would not itself bring it closer to the 5090’s compute resources. The report says Nvidia would need to use a cut-down GB202 to offer a different configuration, and that chip is also used in higher-priced workstation products.
XDA Developers put the RTX 5090’s original MSRP at $1,999, but said the lowest listed price it found at the time of writing was about $8,000, compared with roughly $1,700 for the RTX 5080. Those are reported market prices, not fixed retail prices, and can vary by seller and availability. The report also said Nvidia senior vice president Jeff Fisher told Taiwanese media the 5090 was not being discontinued. That statement addresses whether production is ending; it does not establish how many cards are being made or where available chips are allocated.
GeForce Rumor Watch · October 22, 2026
Nvidia’s Next GeForce Flagship Is Rumored To Lose Half Its CUDA Cores Overnight
An unconfirmed report points to a 24GB RTX 5080 Super that could be easier to find than the RTX 5090. More memory may help some workloads, but it would not close the reported compute gap.
01 / What the report says
More memory, same reported chip
XDA Developers reported a possible 24GB RTX 5080 Super using the RTX 5080’s GB203 silicon. The rumor describes a memory increase and higher power rating, while Nvidia has confirmed neither the model nor its release.
Memory · reported
16GB → 24GB
The extra capacity could matter for some demanding workloads, including 4K gaming with path tracing. It does not guarantee a broad gaming performance gain.
Power · reported
360W → 415W
The rumored rating is 55 watts above the standard RTX 5080. No official specification or independent measurement is available.
Compute · reported
GB203 remains
The report says the Super would retain the RTX 5080 chip. More memory alone would not give it the RTX 5090’s GB202 compute resources.
02 / Core-count context
The 80-class gap has widened
Reported CUDA core counts show a growing distance between Nvidia’s 80-class and 90-class cards. These figures describe core counts, not game performance, which also depends on architecture, clocks, memory, software and the title tested.
| Generation | 80-class cores | 90-class cores | 80-class as share | Reported dies |
|---|---|---|---|---|
| RTX 3080 / 3090 | 8,704 | 10,496 | 83% | GA102 / GA102 |
| RTX 4080 / 4090 | 9,728 | 16,384 | 59% | AD103 / AD102 |
| RTX 5080 / 5090 | Less than half | 21,760 | Under 50% | GB203 / GB202 |
03 / Availability and trade-offs
Stock could shape the practical flagship
The report describes the RTX 5090 as difficult to find and raises silicon allocation as one possible factor. Availability, reported asking prices and the rumored card’s real performance all remain unsettled.
Availability · uncertain
Scarcity changes choices
If the RTX 5090 remains scarce and a lower-priced 80-class card is broadly available, buyers may favor what they can actually purchase. The cards would still not be equivalent.
Memory · workload-specific
Capacity is not compute
24GB could help when a game or creative workload needs more memory. CUDA core count, memory capacity and measured frame rates are different things.
Supply · unverified
Allocation is unknown
GB202 also powers the RTX Pro 6000 Blackwell. The report suggests workstation demand could compete for silicon, but Nvidia has published no allocation figures.
“The 5090 isn’t going anywhere.”
Jeff Fisher, Nvidia senior vice president, as reported by XDA Developers. The statement addresses discontinuation, not production volume or retail availability.
04 / What remains unknown
Rumor, report, and evidence
No independent testing exists for the rumored card. Reported market prices are snapshots from the article, not fixed retail prices or verified averages.
Rumor reported
XDA Developers described a possible 24GB RTX 5080 Super on October 22, 2026.
Specs unverified
Nvidia has not confirmed the card, memory, power rating, price or release timing.
CES 2027 checkpoint
The report points to CES as a possible moment for more information; Nvidia has not confirmed an announcement.
Testing decides value
Retail stock, street prices and independent reviews will show whether the lineup changes for buyers.
05 / Key questions
What buyers can say today
Has Nvidia confirmed the RTX 5080 Super?
No. The 24GB configuration and 415W rating come from reported leaks and remain unverified.
How many CUDA cores does the RTX 5090 have?
XDA Developers reports 21,760 CUDA cores, just over twice the RTX 5080’s reported count.
Would 24GB make the rumored card as powerful?
No such conclusion follows from memory capacity alone. The rumor retains GB203, while the RTX 5090 uses GB202 and has substantially more CUDA cores.
Why is the RTX 5090 hard to find?
The cause is unclear. The report discusses workstation demand and limited GDDR7 availability as possible factors, without verified supply or allocation data.
Availability Could Define Nvidia’s Flagship
If the 24GB model appears and the RTX 5090 remains scarce, availability may matter more to buyers than the nominal top spot. A card that costs far less and can actually be purchased could serve as the practical high-end choice, even without matching the 5090’s processing capacity. That would not make the cards equivalent: the core-count gap remains, and actual gaming performance cannot be judged from memory capacity or CUDA cores alone.
The rumored memory increase could still be relevant for demanding games. XDA Developers cited 4K gaming with path tracing as a workload where a 16GB card may run into memory limits, pointing to Indiana Jones and the Great Circle as an example. But the article’s writer said they had not personally encountered a game where the RTX 5080’s 16GB capacity limited performance. The benefit of 24GB, then, depends on the workload and should not be treated as a confirmed performance gain across games.
For consumers, the distinction affects what they are paying for: more memory is not the same as more compute. For Nvidia, the choice may also involve allocating GB202 silicon between gaming and workstation products. The report argues that workstation cards can be more profitable, but Nvidia has not publicly disclosed chip allocation figures or said that this is driving RTX 5090 supply.
NVIDIA GeForce RTX 5080 Super 24GB graphics card
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
The 80-Class Gap Has Grown
The comparison reflects a longer shift in Nvidia’s product tiers. Two generations before Blackwell, the RTX 3080 had 8,704 CUDA cores and the RTX 3090 had 10,496, or about 83% as many. Both used the GA102 die. In the next generation, the RTX 4080 had 9,728 cores against the RTX 4090’s 16,384, about 59%; those cards used different silicon, AD103 and AD102 respectively.
On the figures in the report, the RTX 5080’s core count is again less than half of the 5090’s. That comparison describes CUDA core counts, not a direct measure of game performance. Performance also depends on architecture, clock speeds, memory, software and the game being tested. The article’s author, who said they own an RTX 3080 and RTX 5080, described both as capable of high-resolution gaming in the games they had tried, while finding the newer card a meaningful upgrade.
The supply question matters because the RTX 5090 uses the GB202 chip also found in Nvidia’s RTX Pro 6000 Blackwell. XDA Developers said the 5090 has attracted demand for workstation use as well as gaming. The report presents competition for that silicon as one possible factor behind the gaming card’s limited availability, but it does not provide independently verified supply or allocation data.
“The 5090 isn’t going anywhere.”
— Jeff Fisher, Nvidia senior vice president, as reported by XDA Developers
As an affiliate, we earn on qualifying purchases.
Specifications and Supply Remain Unverified
The RTX 5080 Super remains a rumor. Nvidia has not confirmed its existence, its 24GB memory configuration, the reported 415-watt rating, pricing or release timing. The available information does not establish whether the model would be sold broadly or positioned as a replacement for another card.
It is also unclear why RTX 5090 cards are so difficult to find, how many GB202 chips Nvidia is assigning to GeForce versus workstation products, and whether memory supply is limiting a potential 32GB consumer card. XDA Developers noted that limited GDDR7 availability could be an alternative explanation. The report also mentioned an alleged RTX 5090 SE using a cut-down GB202, but said that claim was unsubstantiated.
Reported prices are snapshots rather than verified market averages, and no independent tests exist for the rumored card. Until Nvidia provides specifications and reviewers test it, claims about its gaming performance, value and place in the lineup remain unsettled.
As an affiliate, we earn on qualifying purchases.
CES 2027 May Bring Clarity
CES 2027 is the next expected checkpoint for Nvidia’s GeForce lineup, according to XDA Developers. An announcement there could confirm whether a 24GB RTX 5080 Super is real, and clarify its specifications, price and intended position. Nvidia has not confirmed that it will make an announcement at the event.
Until then, buyers should treat the Super specifications and the separate RTX 5090 SE rumor as unconfirmed. The practical picture will depend on more than a product announcement: retail stock, street prices and independent performance testing will show whether a new 5080 model changes what gamers can actually purchase.
As an affiliate, we earn on qualifying purchases.
Key Questions
Has Nvidia confirmed the RTX 5080 Super?
No. Nvidia has not announced the card. The reported 24GB configuration and 415-watt rating come from leaks described by XDA Developers and remain unverified.
How many CUDA cores does the RTX 5090 have?
The RTX 5090 has 21,760 CUDA cores, according to XDA Developers. The report says that is just over twice the RTX 5080’s core count.
Would 24GB of memory make the rumored 5080 as powerful as a 5090?
No such conclusion can be drawn from memory capacity alone. The rumored card is said to retain the GB203 chip, while the 5090 uses GB202 and has substantially more CUDA cores. No confirmed specifications or independent tests are available for the rumored model.
Why is the RTX 5090 difficult to find?
The reason has not been confirmed. XDA Developers points to possible competition for GB202 chips used in workstation products, while also noting that limited GDDR7 supply could affect consumer card availability. Nvidia has not disclosed allocation figures.
When could Nvidia clarify the rumor?
XDA Developers identifies CES 2027 as the next expected checkpoint for GeForce news. Nvidia has not confirmed that it will announce a 5080 Super there.
Source: XDA Developers
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
