NVIDIA is reportedly discontinuing production of its most expensive GeForce gaming card, potentially leaving the RTX 5080 as the new flagship of its consumer lineup. The change has not been confirmed by the company.
NVIDIA’s GeForce RTX 5090 may be reaching the end of its production far sooner than gamers were expecting. According to reports, the company has stopped allocating its GB202 graphics processor to consumer graphics cards and intends to reserve future supplies for professional RTX PRO products on October 10, 2026.
If the reports turn out to be true, the decision could have an impact on the world of gamers, PC builders, indie developers and creators that depend on the 32GB of graphics memory on the RTX 5090. Retail supplies may become scarce, while remaining cards may sell for higher prices.
NVIDIA has not publicly confirmed that the RTX 5090 is dead. Currently, the information comes from industry leakers and an alleged factory notice, not from an official product announcement.
The Facts About the NVIDIA RTX 5090 Production Reports
The first question is whether NVIDIA has pulled the plug on the RTX 5090 or has just stopped providing new processors to graphics card manufacturers. These are related developments, but not necessarily on the same day.
On October 10, Tom’s Hardware reported that a hardware industry source, @hongxing2020, leaked an alleged factory notice, stating that NVIDIA will halt production of RTX 5090 models, including those for China.
Another industry leaker, MEGAsizeGPU, tweeting under the name @Zed__Wang, said the GB202 silicon will be reserved for professional and data center products. The same account has appeared in TechPowerUp coverage and other hardware outlets.
The distinction is important because multiple articles reporting the same information do not necessarily mean multiple independent sources have verified it. The factory communique has not been officially confirmed and the company has not released any official timeline for the end of production.
There is also a separate claim about the GeForce RTX 5080. The most powerful gaming card will be the RTX 5080 24GB, which will replace both the RTX 5090 and the current RTX 5080 16GB, according to the supposed notice. We don’t know if the 16GB model has been discontinued or if there’s a replacement, and NVIDIA hasn’t said either way.
Why the GB202 Processor is More Important Than the RTX 5090 Name
The RTX 5090 is based on the GB202 processor from the Blackwell architecture by NVIDIA. This is the silicon that powers much of the graphics and compute performance of the card. It includes processing units for traditional rendering, ray tracing and machine learning calculations.
“This class of processor isn’t just what we’re building the RTX 5090 around. The company’s professional Blackwell graphics cards also use GB202-based configurations, but are sold to customers with different memory, reliability and software requirements.
For example, NVIDIA’s RTX PRO 6000 Blackwell Workstation Edition has 96GB of GDDR7 memory with error-correcting code, while the GeForce RTX 5090 has 32GB of GDDR7. The professional card is designed for demanding workstation applications like local model development, engineering simulations, rendering and scientific computing.
The reported change in the allocation is economically plausible. The same silicon, sold as part of a consumer graphics card, may make less money than a GB202 processor used in a professional product. Memory capacity, error correction and software support may be more important to the pro customer than the gaming buyer.
That does not establish what NVIDIA’s real motivation is. There are no public plans that show how many GB202 processors the company used to send to gaming cards or how many it now plans to set aside for professional products.
NVIDIA RTX 5090 vs. RTX PRO 6000: Why the Same Chip Can Have Different Buyers
NVIDIA’s published product specifications shed light on why a decision on chip allocation might matter regardless of whether the consumer and professional products are aiming for different markets.
| Specification | GeForce RTX 5090 | RTX PRO 6000 Blackwell Workstation |
|---|---|---|
| Architecture | Blackwell | Blackwell |
| GPU family | GB202 | GB202 |
| Graphics memory | 32GB GDDR7 | 96GB GDDR7 ECC |
| CUDA cores | 21,760 | 24,064 |
| Board power rating | 575W | 600W |
| Original US Pricing | Price starting at $1,999 | Professional price range |
| Target use | Gaming and content creation | Professional computing and visualization |
Sources: NVIDIA, Official GeForce and RTX PRO specs. The RTX PRO 6000 Workstation Edition is one of the professional GB202 configurations. These figures are for different products, not an apples-to-apples comparison of performance.
The amount of video memory isn’t the only difference. The two products share architectural technology, but have different configurations and target workloads. ECC memory can detect and correct some memory errors . This is useful for calculations and professional work where the integrity of the data matters .
For gamers, the RTX 5090 offers a ton of processing power and 32GB of memory without having to buy a professional workstation. Developers have also used consumer GPUs for local machine learning experiments, rendering and other compute-heavy tasks.
That common audience means that the reported GB202 allocation shift has wider implications than a typical graphics card retirement. If RTX 5090 inventory dries up, users who need 32GB of local GPU memory may have fewer options at consumer pricing.
NVIDIA’s official RTX 5090 specs and RTX PRO 6000 product details still serve as the technical reference points for those products.
Will The RTX 5090 Prices Go Up in USA?
Production stoppages do not instantly clear all graphics cards from retail shelves. Manufacturers may still have unused chips or completed boards, distributors may still have some shipments left, and retailers may still be selling through existing inventory.
The effect on US prices would depend on remaining inventory and whether buyers still want the RTX 5090 when new shipments become less certain.
NVIDIA has launched the RTX 5090, starting at $1,999. That is original launch reference, not a guaranty of current availability or retail pricing. NVIDIA’s store also lists individual products and availability, but just because something is out of stock doesn’t mean it’s been discontinued.
If supplies do in fact dry up, buyers could face less inventory, higher premiums on the remaining units, or more activity in the used-car market. None of those outcomes is certain, and the October 10 reports don’t create a validated shortage timeline.
For anyone wanting to build a high-end gaming PC, the pragmatic advice is to separate rumor from purchasing decision. “ The fact that future production is uncertain doesn’t mean that a card offered at an inflated price is a sensible purchase. Warranty, seller credibility, real gaming performance, and overall system budget still count.
Would an RTX 5080 24GB be a replacement for the RTX 5090?
One of the less proven parts of the story is talk of a 24GB version of the GeForce RTX 5080.
NVIDIA revealed the current RTX 5080 with 16GB of GDDR7 memory and a beginning price of $999. A 24GB model would give you 50% more memory capacity, which might help on workloads that go beyond 16GB. But more memory would not suddenly give it the processing power of an RTX 5090.
The two products have different GPU setups. NVIDIA lists 10,752 CUDA cores on the current RTX 5080, and 21,760 on the RTX 5090. Memory capacity alone does not alter that difference.
A 24GB version of the RTX 5080 would be a good option for creators working with large assets, as well as developers needing more memory than the current RTX 5080. Its value will depend on final specs, memory bandwidth, price and performance as tested.
There are no official specifications, US release date, or pricing for a 24GB RTX 5080 in the sources checked for this article. Claims that it will replace the RTX 5090 aren’t part of the same unconfirmed report.
What will happen to the owners of RTX 5090?
Existing RTX 5090 cards would not be rendered useless if production was halted. Owners would keep their installed hardware and these reports show no signs of an announced end to NVIDIA’s gaming driver support for the model.
Production, retail availability, warranty service, and software support are different things. A product may exit manufacturing, but existing customers will continue to receive driver updates and warranty support under the terms applicable to their purchases.
If you’re already gaming, video editing or doing local computing on an RTX 5090, there’s no technical reason to replace the card just because a discontinuation rumor has come up.
The bigger worry is with buyers who have not purchased one yet. If NVIDIA confirms that it’s done shipping new GB202 to consumer manufacturers, the existing stock could gradually run out as it’s sold.
What NVIDIA Has Yet to Confirm
The report issued October 10 raises several questions. It remains unclear if the GeForce RTX 5090 is permanently discontinued, if the allocation reportedly being put in place applies to all markets, or if existing board partners will continue to receive limited quantities of GB202 processors, as reported by VideoCardz.
The proposed replacement, the RTX 5080 24GB, is also not confirmed. We aren’t sure of its final specifications, price, availability, or potential tie-in to an RTX 50-series refresh.
Unless NVIDIA has officially said so, or there are confirmed communications to board manufacturers, or there is consistent evidence of orders for supply being terminated, I don’t believe there is a solid enough basis to report the RTX 5090 as being discontinued.
Until we get that information, the best we can do is say that the RTX 5090 is under a speculative production shutdown report, rather than NVIDIA having officially halted its flagship GeForce graphics card.
The bigger problem is what happens when a processor serves both gaming and professional computing customers. If NVIDIA were to dedicate future GB202 production solely to workstation and server products, the decision would impact more than one graphics card. That will change the customers who can get the company’s biggest Blackwell GPU in the consumer market and how much they pay for it.