Nvidia reportedly ends RTX 5090 output as GB202 silicon goes to RTX Pro cards

Nvidia's GB202 graphics processor is the largest die in the company's Blackwell family, and until recently it served two very different markets. One is the GeForce RTX 5090, the consumer flagship that launched in January 2025 at a recommended price of $1,999. The other is a growing family of professional and AI accelerators, including the RTX Pro 6000, which uses a fuller 24,064 core version of the same silicon with 96GB of memory, and the recently unveiled RTX Pro 5500, which pairs the RTX 5090's 21,760 CUDA cores with 84GB of GDDR7.

The economics of those two markets have diverged sharply. The RTX Pro 6000 can sell for around $15,000, and Tech Insider reports listings near $16,000. The RTX 5090 launched at $1,999. TweakTown notes that the professional cards are far more lucrative per unit of silicon, and that GB202 already powers them. A wafer of GB202 therefore earns a much higher return when it is finished as an RTX Pro board than when it becomes a gaming card.

Consumer availability has been deteriorating for months. TechPowerUp reports that European listings for the RTX 5090 passed €5,000 and that Walmart sold out at $4,299. Tech Insider reports that the card has already vanished from first party United States retail, with Newegg, Amazon, Best Buy and B&H showing no first party stock, while third party marketplace sellers ask between $6,500 and $9,500. For a product with a $1,999 launch price, that is a striking gap between the shelf label and the street price.

Against that backdrop, on October 10, 2026, a cluster of hardware outlets reported that Nvidia has stopped allocating GB202 silicon to the GeForce RTX 5090 altogether, and that all future supply of the die will go to the RTX Pro professional and AI lineup. The reports rest on a leaked factory notice and on two well known hardware leakers, and Nvidia has not confirmed any of it.

Key Facts

Tom's Hardware reported on October 10, 2026 that Nvidia has reportedly ceased selling GB202 graphics processors for consumer GeForce RTX 5090 series add-in boards, including the D variants made for the Chinese market. The outlet cited a leaked notice from an unknown factory, relayed by leaker @hongxing2020. A second prominent leaker, @Zed__Wang, who also posts as MEGAsizeGPU, confirmed the claim, saying that from now on GB202 silicon will be used exclusively on data center and professional graphics cards.

TechPowerUp reported on October 10, 2026 that Nvidia has reportedly stopped allocating GB202 silicon to the RTX 5090. The outlet quoted MEGAsizeGPU directly: "No more GB202 for 5090, confirmed," the leaker wrote, adding that "all future GB202 supplies are RTX Pro exclusive." The same leaker posted the phrase "5090 EOL, 5080 24GB ready" alongside an apparent factory notice. TechPowerUp notes that GB202 is the largest Blackwell graphics chip and that the RTX 5090 does not even max it out, because the consumer card leaves some of the die's capacity unused.

The factory notice itself has been reproduced across several sites. It reads: "According to a notification from the factory, Nvidia will no longer manufacture any RTX 5090 models, including the D variants. Going forward, the highest-end GeForce graphics card will be the RTX 5080 24GB, while the RTX 5080 16GB version will be discontinued." TweakTown reported on October 10, 2026 that the RTX 5090 appears to be nearing end-of-life status, and that the notice was corroborated independently by MEGAsizeGPU.

Tech Insider reported on October 11, 2026 that Nvidia has reportedly stopped assigning new GB202 silicon to the GeForce RTX 5090, redirecting the chip exclusively to the RTX Pro professional and AI lineup. The same report stresses several limits on what is actually known: no leaked document includes a quantified production cut percentage or a unit count, no add-in board partner such as ASUS, MSI or Gigabyte has publicly confirmed or denied the claim, and Nvidia has issued no statement naming an end-of-life date. Additional coverage was cited from HotHardware, Wccftech and HWBusters.

Existing RTX 5090 cards remain on sale until stocks run out. The RTX 60 series, the next consumer generation, has reportedly slipped to 2028, which means the rumored RTX 5080 24GB could sit at the top of the GeForce stack for a long time.

Analysis

What this really means is that the consumer graphics card market is now a byproduct of the AI accelerator market rather than a priority in its own right. GB202 is a scarce and expensive piece of silicon, and Nvidia can either sell it as a $1,999 gaming card or as a professional board that lists near $15,000 or $16,000 with 96GB of memory. When the same die can be sold into a market with that much pricing power, the gaming segment has to wait in line, and the queue is being managed quietly rather than through a product launch or a formal end-of-life announcement.

The wording of the reported factory notice is revealing. It does not describe a shortage or a pause. It describes a permanent change in product hierarchy, with the RTX 5080 24GB named as the highest end GeForce card and the RTX 5080 16GB discontinued. That is the language of a portfolio decision rather than a temporary allocation problem.

There is also a credibility question to weigh. The story comes from leakers and a factory notice rather than from Nvidia itself, and no AIB partner has confirmed or denied it, while no cut percentage or unit count has surfaced. The RTX 5090 was already in severe shortage before any of this. That combination makes the report plausible but not proven, an unverified supply allocation rumor layered on top of a documented shortage.

Either way, the direction of travel is the same. TweakTown's read, that the shift makes financial sense because GB202 powers far more lucrative professional cards, is hard to argue with. The consumer flagship is being treated as the least valuable use of the company's largest chip.

Why It Matters

For PC builders, the practical consequence is price. A card that launched at $1,999 in January 2025 was already trading at $4,299 at Walmart, above €5,000 in European listings, and between $6,500 and $9,500 from third party sellers before the production reports appeared. If supply tightens further, those numbers become the floor rather than the ceiling, and the RTX 5080 24GB inherits the top slot in the GeForce line whether or not buyers consider it a worthy flagship.

For Nvidia, the trade is straightforward but not costless. Every GB202 die routed to an RTX Pro 6000 or RTX Pro 5500 supports a much higher selling price, and the professional cards are aimed at the same AI demand that has driven the company's data center business. The cost is the enthusiast segment, which has historically been the source of brand loyalty and the halo effect that sells lower tier cards.

For the wider market, the episode is another sign that consumer GPUs are no longer the main event. Memory and advanced packaging capacity are being consumed by accelerators, and the consumer product stack is being reshaped around whatever silicon is left. The RTX 5080 24GB becoming the fastest GeForce card in production would be a striking outcome for a generation that was introduced with the RTX 5090 as its headline part.

Next Up

The immediate signal to watch is inventory. Existing RTX 5090 boards remain available until stocks run out, and Nvidia has issued no statement on an end-of-life date, so the market will be reading retail listings and distributor notices rather than press releases. Any confirmation or denial from ASUS, MSI or Gigabyte would settle the question quickly, and so would a formal Nvidia announcement about the RTX 5080 24GB.

Further out, the reported slip of the RTX 60 series to 2028 suggests the reshaped stack could persist for a long time. If the RTX 5080 24GB does become the top consumer GeForce card, it may hold that position for well over a year, with GB202 living out its life almost entirely inside RTX Pro boards.