Nvidia’s GeForce RTX 5090 may be on its way out. Multiple hardware outlets reported on October 10 and 11, 2026 that the company has stopped feeding new GB202 silicon to its gaming flagship, with the chips instead routed to the professional RTX Pro lineup. The same reports claim a 24GB version of the GeForce RTX 5080 is being lined up to take the 5090’s place at the top of the consumer stack. Nvidia has not confirmed any of it.
The story broke through a mix of leaker posts and board-partner chatter rather than a press release, which is exactly why it matters how it’s reported. Here’s what outlets like TweakTown, TechPowerUp, VideoCardz and Wccftech actually said, what Nvidia has gone on record with, and what every gamer weighing an RTX 5090 purchase right now needs to know before the rumor mill drowns out the facts.
What the RTX 5090 discontinued reports actually claim
The chain of reporting traces back to two X-based hardware leakers, MEGAsizeGPU and hongxing2020, who separately posted on October 10 that Nvidia had cut off new GB202 chip allocation for the RTX 5090. TechPowerUp picked up the MEGAsizeGPU claim first, noting it had circulated in a quieter form back in July before resurfacing with new details this month. VideoCardz and TweakTown followed with their own write-ups the same week, both attributing the core claim to the same pair of leakers rather than to any Nvidia statement.
The claim itself has two parts. First, Nvidia is said to have told board partners it will no longer sell RTX 5090 models, including the China-specific RTX 5090 D variants built to comply with export rules. Second, the GB202 die that powers the 5090 would not disappear. It would simply get redirected entirely to the RTX Pro workstation and AI-server cards, where demand has outpaced what Nvidia can currently supply. That distinction matters: this isn’t a story about a chip going extinct, it’s a story about who gets to buy it.
The alleged factory notice, word for word
According to the notice relayed by leaker hongxing2020 and cited in subsequent coverage, the message read in part: “NVIDIA will no longer sell any RTX 5090 models, including the D variants.” That line is doing a lot of work in this story, and it’s worth being precise about where it comes from. It is not a quote from Nvidia. It is not a quote from a named Nvidia executive. It is a secondhand factory notice relayed by an anonymous leaker and repeated by outlets that, to their credit, flagged it as unconfirmed from the start.
The same alleged notice reportedly went further, stating that a GeForce RTX 5080 24GB would become the new highest-end GeForce card, and that the current RTX 5080 16GB would be phased out in the process. If true, that would mean Nvidia isn’t just trimming a flagship, it’s reshuffling the entire top half of the RTX 50 lineup in the middle of its product cycle.
Why GB202 silicon suddenly became this valuable
GB202 is the largest die in Nvidia’s Blackwell consumer and workstation stack, and it’s the same silicon that underpins the RTX Pro 6000, RTX Pro 5500 and RTX Pro 5000 cards used in AI training and rendering workloads. Those professional cards carry far thicker margins than a gaming GPU, and demand for them has climbed through 2026 as enterprises race to build out local AI infrastructure. Several reports on this story, including coverage from 4Gamers in Taiwan, note that RTX Pro cards built on GB202 ship with configurations as large as 96GB of GDDR7, aimed squarely at large-model training rather than gaming.
Put plainly: when a single die can be sold into a 4K gaming card or a 96GB AI workstation card, and the AI card sells for several times more, a chip allocation squeeze tends to go one direction. This isn’t the first time that dynamic has shown up around the RTX 5090 this year, and it won’t be the last as long as GB202 stays Nvidia’s largest available die for both markets.
RTX 5080 24GB: the rumored replacement flagship
What changes if the 24GB card is real
Reports describe the proposed RTX 5080 24GB as carrying 24GB of GDDR7 on a 256-bit memory bus, roughly matching the existing RTX 5080’s bus width but with more total memory. One detail circulating among the reports is that Nvidia could reach that higher capacity by switching to 3GB GDDR7 memory modules, up from the 2GB modules used on the current RTX 5080 16GB. That’s a real manufacturing option available to memory vendors today, but none of it has been specified by Nvidia, and no price or launch date has surfaced in any of the reporting.
What stays confirmed
What is confirmed, straight from Nvidia’s own GeForce RTX 5080 product page, is that the card currently ships with 16GB of GDDR7 memory. Nvidia also confirmed back in January 2025 that both the RTX 5090 and RTX 5080 were “out now” as of January 30 of that year. Everything about a 24GB variant, a new memory module size, or a wider rollout remains outside anything Nvidia has published.
What Nvidia has and hasn’t confirmed
It bears repeating because this story has a habit of getting flattened into certainty as it spreads: Nvidia has not announced that it discontinued the RTX 5090 or any RTX 5090 D variant. Nvidia has not announced a GeForce RTX 5080 24GB, its specifications, a launch date, or a price. No Nvidia spokesperson, executive, or official channel has been quoted anywhere in this reporting cycle confirming any part of the discontinuation claim.
That doesn’t make the reports worthless. MEGAsizeGPU and hongxing2020 both have a track record of surfacing accurate supply-chain details ahead of official confirmation in past Nvidia product cycles, and four separate outlets corroborated the core claim within the same 48-hour window. But a pattern of accurate leaks in the past is not the same thing as a confirmed announcement today, and readers deciding whether to buy, sell, or wait on an RTX 5090 right now should weigh the claim accordingly.
GeForce RTX 50-series: confirmed specs vs the rumored lineup
| Card | Status | Memory | Memory bus | Source |
|---|---|---|---|---|
| GeForce RTX 5090 | Currently shipping | Not specified in this reporting cycle | Not specified in this reporting cycle | Nvidia, launched Jan. 30, 2025 |
| GeForce RTX 5080 (16GB) | Currently shipping | 16GB GDDR7 | 256-bit (current model) | Nvidia.com product page |
| GeForce RTX 5090 D (China) | Reportedly being discontinued | Unchanged from standard 5090 | Unchanged from standard 5090 | VideoCardz, citing leaker reports |
| GeForce RTX 5080 24GB (rumored) | Unannounced, reported in development | 24GB GDDR7 (reported) | 256-bit (reported) | Wccftech, TweakTown, VideoCardz |
| RTX Pro 6000 / 5500 / 5000 | Currently shipping (professional line) | Up to 96GB GDDR7 (reported range) | Varies by model | 4Gamers, citing GB202 allocation reports |
How this fits the GB202 story that’s been building all year
This isn’t the first time GB202 allocation has made headlines in 2026. Earlier reports throughout the year tracked a slower squeeze, with Nvidia gradually redirecting GB202 capacity toward RTX Pro cards while RTX 5090 retail availability thinned out and street prices climbed well above the card’s original suggested price. What’s different about this week’s reports is the scale of the claim. Earlier coverage described a supply shift. This week’s reports describe an outright production stop, with a named replacement card already being discussed internally, according to the leakers.
If the earlier allocation-shift reports turn out to have been the first phase and this week’s discontinuation claim is the second, that would track with how semiconductor companies tend to retire a SKU: quietly reduce supply first, then formalize the exit once the replacement is ready to ship. That sequencing is circumstantial, not proof, but it is consistent with the pattern several outlets have now reported independently.
Historical context: Nvidia has done this before
Nvidia trimming a flagship SKU mid-generation isn’t new. The company has reshuffled GeForce lineups between major launches before, swapping memory configurations, adjusting core counts, and retiring cards once newer silicon or supply conditions made the old configuration less profitable to keep building. What makes the RTX 5090 situation distinct is the driver behind it. Earlier lineup changes were mostly about yields, cost, or making room for a refresh. This one, if the reports hold up, is about a single chip being pulled in two directions by two very different markets at once: gamers who want the biggest GeForce card available, and enterprises buying RTX Pro cards by the rack for AI workloads.
That tension has been building since Blackwell launched in January 2025. GB202 was always Nvidia’s most capable consumer die, and it was always going to be tempting to lean on it for professional cards once AI demand outstripped supply. This week’s reports suggest that tipping point may have finally arrived for the RTX 5090 specifically.
Market impact: what happens to RTX 5090 prices now
If the discontinuation reports are accurate, the immediate effect on the secondary market is predictable. Cards already in retail channels become the only way to get a new RTX 5090, and once board-partner and retailer stock runs dry, as TweakTown’s report suggests it eventually will, prices on remaining inventory tend to rise rather than fall. That pattern has already played out this year with GB202 allocation tightening, and a confirmed discontinuation would likely accelerate it rather than reverse it.
Retailers and board partners such as Asus, MSI and Gigabyte have not issued statements confirming an end-of-life timeline for their RTX 5090 cards, and none of the outlets covering this story reported a specific sell-through deadline. For buyers currently shopping, that leaves a genuinely uncertain window: the card isn’t confirmed dead, but treating current stock as finite rather than restockable looks like the safer assumption given how the reporting has trended.
Competitive landscape: where AMD fits into this
Nvidia pulling its halo gaming card out from under enthusiasts, even temporarily, opens a narrow but real opportunity for AMD. AMD’s Radeon RX 9000 series has positioned itself as the value alternative to Nvidia’s RTX 50 lineup rather than a direct flagship competitor, and AMD has not announced a card built to go head to head with the RTX 5090 on raw performance. Still, a gap at the very top of Nvidia’s consumer stack, even a short one created by a supply reshuffle rather than a true successor launch, is the kind of opening AMD’s marketing has leaned into before.
Whether AMD moves to capitalize on it depends on timing details none of the current reporting addresses: how long the RTX 5090 stays off shelves, whether the RTX 5080 24GB actually materializes as a true flagship replacement, and how close that card’s performance ends up to the 5090 it would replace. None of those answers exist yet.
Confirmed vs unconfirmed: a claim-by-claim breakdown
| Claim | Status | Primary source |
|---|---|---|
| RTX 5090 and RTX 5080 are official GeForce products | Confirmed | Nvidia.com |
| RTX 5090 and RTX 5080 launched Jan. 30, 2025 | Confirmed | Nvidia |
| RTX 5080 currently ships with 16GB GDDR7 | Confirmed | Nvidia.com product page |
| Nvidia has discontinued the RTX 5090, including D variants | Unconfirmed | Leaker notice via hongxing2020, reported by VideoCardz |
| GB202 allocation is being redirected to RTX Pro cards | Unconfirmed | MEGAsizeGPU, reported by TechPowerUp |
| A GeForce RTX 5080 24GB is planned as the new flagship | Unconfirmed | Wccftech, TweakTown, VideoCardz |
| RTX 5080 24GB would use 3GB GDDR7 modules | Unconfirmed | Reports circulating alongside Wccftech coverage |
| Price or launch date for RTX 5080 24GB | Not reported anywhere | None available |
What gamers should actually do right now
For anyone shopping for a high-end GPU this week, the practical advice is less dramatic than the headlines suggest. If an RTX 5090 is in stock at or near its original price, the discontinuation reports alone aren’t a reason to panic-buy, since nothing about performance or driver support changes even if the reports are accurate. If the only available 5090 cards are already marked up well past list price, waiting to see whether the RTX 5080 24GB materializes, and at what price, is the more defensible move.
Buyers who specifically need 32GB-class VRAM for local AI workloads, rather than gaming, face a tighter calculation. If GB202 really is being pulled toward RTX Pro cards, that line is the one most likely to see Nvidia’s attention and new stock going forward, for a price far above anything in the GeForce lineup.
Predictions: how this likely plays out
- Nvidia issues some form of acknowledgment within weeks rather than months, given how quickly this story spread across TweakTown, VideoCardz, TechPowerUp and Wccftech within a single 48-hour window.
- If a GeForce RTX 5080 24GB does ship, it arrives priced closer to the outgoing RTX 5090 than to the current RTX 5080, given GB202’s die cost and the memory upgrade required.
- RTX 5090 street prices rise in the near term regardless of official confirmation, simply because buyers now perceive the card as finite.
- AMD does not announce a direct RTX 5090-class competitor in response, continuing its current strategy of competing on value in the upper-midrange rather than at the absolute top of the stack.
- Any RTX 5080 24GB launch, if it happens, slips into 2027 rather than arriving before the end of 2026, given that none of the current reports include a firm date.
The bigger pattern: AI demand is reshaping GeForce itself
Strip away the specific product names and this story is really about a chip shortage that has nothing to do with gaming demand. GB202 was designed as a flagship consumer die first, and its second life as the backbone of RTX Pro’s AI-oriented cards wasn’t the original plan so much as a consequence of how fast enterprise AI infrastructure spending grew through 2025 and 2026. When a company has to choose between selling a chip to a gamer once or selling a far more expensive version of the same chip to an enterprise buyer repeatedly, the economics tend to win the argument eventually.
That dynamic isn’t unique to Nvidia or to this chip. It’s the same pressure showing up across the GPU and memory supply chain this year, from GDDR7 pricing to DRAM allocation more broadly. The RTX 5090 discontinuation reports, confirmed or not, are a symptom of that pressure rather than an isolated product decision.
Frequently asked questions
Has Nvidia officially confirmed the RTX 5090 is discontinued?
No. As of October 12, 2026, Nvidia has not made any official statement confirming that the GeForce RTX 5090, including the China-market RTX 5090 D variants, has been discontinued. The claim comes from leaker reports relayed by outlets including VideoCardz and TweakTown.
Is the GeForce RTX 5080 24GB a real, confirmed product?
No. The RTX 5080 24GB has not been announced by Nvidia. It exists only in reports citing leakers, with no confirmed specifications, price, or launch date.
What does the current GeForce RTX 5080 actually ship with?
According to Nvidia’s own product page, the currently available GeForce RTX 5080 ships with 16GB of GDDR7 memory. Any 24GB version remains unconfirmed and unannounced.
Why would Nvidia redirect GB202 chips away from the RTX 5090?
Reports attribute the shift to rising demand for Nvidia’s RTX Pro professional cards, which use the same GB202 die and reportedly carry configurations with far more memory, aimed at AI training and enterprise workloads that pay significantly more per chip than gaming cards do.
Should I buy an RTX 5090 right now if I see one in stock?
That depends on price and need. Nothing about the card’s performance changes based on these reports. If it’s priced near its original level, the discontinuation rumor alone isn’t a strong reason to avoid it. If it’s already marked up significantly, waiting to see how the RTX 5080 24GB situation develops is the more cautious path.
Will RTX 5090 prices go up if this is confirmed?
No official pricing data exists yet for any outcome, but historically, when a high-demand GPU is perceived as being phased out, remaining retail stock tends to see price increases rather than discounts, a pattern already visible in GB202-related supply reports earlier this year.
Where did this story originate?
The claim traces back to posts from hardware leakers MEGAsizeGPU and hongxing2020, picked up and reported independently by TechPowerUp, VideoCardz, TweakTown and Wccftech between July and October 2026, with the most detailed reporting landing on October 10 and 11.
Does this affect RTX 5090 warranty or driver support?
None of the current reporting suggests any change to warranty coverage or driver support for existing RTX 5090 owners. A production discontinuation, even if confirmed, would not retroactively affect cards already sold.
