- Hardware leakers report that NVIDIA has stopped allocating GB202 chips to the GeForce RTX 5090, redirecting all future supply to the more profitable RTX Pro AI and workstation GPU lineup.
- A second leaker shared what appears to be a factory notice stating NVIDIA will no longer manufacture any RTX 5090 models, including China-specific D variants.
- NVIDIA has not officially confirmed any of these reports and has made no public announcement regarding RTX 5090 discontinuation.
- RTX 5090 prices have already surged to as high as $9,500 on the secondary market in the US, well above its $1,999 MSRP at launch.
- A separate leak suggests NVIDIA is preparing an RTX 5080 24GB to replace the RTX 5090 as the consumer flagship — but the RTX 5080 has roughly 30% lower 4K performance than the 5090.
- The same GB202 chip used in the RTX 5090 powers the RTX Pro 6000, which sells for $16,000 — eight times the 5090’s launch price — making the chip far more lucrative in the AI market.
If you’ve been eyeing the RTX 5090 and waiting for a good deal to come along, this update is going to sting. Reports emerging over the past week suggest that NVIDIA may be quietly winding down production of its flagship consumer GPU — and the reason comes down to cold, hard profit margins. The same chip that powers gaming’s most powerful card is apparently worth a lot more to NVIDIA when it’s sitting inside a $16,000 professional AI workstation GPU instead.
What the Leaks Are Saying
The story broke on October 10, 2026, through hardware leaker MEGAsizeGPU on X, who posted simply: “No more GB202 for 5090, confirmed.” In a follow-up, the leaker added that “all future GB202 supplies are RTX Pro exclusive.” A second leaker, hongxing2020, added weight to the claim by sharing what they described as a factory notice stating NVIDIA will no longer manufacture any RTX 5090 models — including the China-specific RTX 5090 D variants. The claim was subsequently picked up and reported by multiple hardware outlets including Tom’s Hardware, VideoCardz, TechPowerUp, and Guru3D.
To be clear: NVIDIA has not confirmed any of this. The company has issued no statement on an end-of-life date for the RTX 5090, and neither leak has independent verification beyond the factory notice document. That said, the financial logic behind the move is hard to argue with, and the pattern fits a well-established NVIDIA playbook of redirecting premium silicon to higher-margin professional products as consumer demand normalises.
Why NVIDIA Would Do This — The Business Case
The RTX 5090 uses the GB202 chip — NVIDIA’s largest Blackwell architecture die. In the RTX 5090, it’s actually a slightly cut-down configuration, running 21,760 CUDA cores and 32GB of GDDR7 memory, with a launch MSRP of $1,999.
The same family of GB202 chips also powers the RTX Pro 6000 Blackwell, a professional workstation and AI GPU that runs the chip in a fuller configuration — 24,064 CUDA cores and 96GB of memory — and sells for $16,000 on NVIDIA’s official website. That’s eight times the RTX 5090’s launch price for the same underlying silicon.
The financial arithmetic is straightforward. AI data centres, animation studios, and enterprise customers are paying a massive premium for GB202-based professional cards, and demand from that sector has been enormous throughout 2026. Every wafer NVIDIA redirects from the consumer GeForce line to RTX Pro generates dramatically more revenue per unit. With AI infrastructure spending continuing to accelerate, the incentive to deprioritise the $1,999 gaming card in favour of the $16,000 professional card is obvious — even if it leaves gamers out in the cold.
It’s also worth noting that this isn’t the first sign of strain for the RTX 5090. Earlier this year, reports surfaced that AI firms were bulk-buying RTX 5090 cards for server use, further draining consumer inventory. In September, Hardware Unboxed reported a melted RTX 5090 power connector reaching 175°C, and first-party US retail stock essentially vanished from Amazon and Newegg.
What Happens to RTX 5090 Prices and Availability
The RTX 5090 launched in early 2025 at $1,999. By September 2026, first-party stock had dried up from major US retailers. Third-party sellers are now listing RTX 5090 cards at up to $9,500 — nearly five times MSRP. European markets have seen similar spikes, with cards trading above $6,000 in Germany. In Japan, models that were occasionally restocked through September have surged to around 1.2 million yen.
If the production wind-down is accurate, these prices are not coming down. Existing retail and secondary market inventory is all that remains, and historical precedent with limited-run flagship GPUs suggests prices firm and rise as remaining stock sells through rather than falling. Anyone who was waiting for prices to normalise before buying should treat that hope as effectively over if these leaks prove true.
Importantly, ending new GB202 chip allocations is not technically the same as recalling or discontinuing existing cards. Board partners like ASUS, Gigabyte, MSI, and others can still sell through their existing RTX 5090 inventory. The cards don’t stop working; they just stop being manufactured.
What About the RTX 5080 24GB Rumour?
Alongside the RTX 5090 discontinuation leak, a separate claim from leaker hongxing2020 suggests NVIDIA is preparing a 24GB version of the RTX 5080 to step into the consumer flagship role. The alleged factory notice reportedly states: “Going forward, the highest-end GeForce model will be the RTX 5080 24GB, and the RTX 5080 16GB version will be discontinued.”
This isn’t a completely new rumour — speculation about a 24GB RTX 5080 (variously labelled the RTX 5080 Super or RTX 5080 Ti depending on the source) has circulated since as far back as September 2024. What’s new is the framing of it as a direct replacement for the RTX 5090 as the consumer flagship, rather than a mid-cycle refresh.
Here’s the problem: the RTX 5080 uses the GB203 chip, not GB202. Even with VRAM bumped from 16GB to 24GB, the card would still have roughly half the CUDA cores of the RTX 5090 (10,752 vs. 21,760) and sits approximately 30% below the RTX 5090 in 4K gaming performance. Calling it a replacement for the 5090 is technically accurate in the “there’s now nothing better to buy” sense, but performance-wise it’s a significant step down for consumers who were paying the 5090 premium specifically for that top-tier gap over everything else.
As for price — no confirmed MSRP exists yet for the 24GB variant, but given that the current 16GB RTX 5080 launched at $999 and the 5090 launched at $1,999, speculation is pointing toward a retail price somewhere in the $1,999–$2,999 range for the 24GB model. Given current GPU market conditions, that seems optimistic for anything near MSRP at launch.
What Does This Mean for PC Gamers Right Now?
If you own an RTX 5090 already, nothing changes for you immediately. The card continues to function and receive driver support, and there’s no indication NVIDIA is doing anything but quietly redirecting chip supply — not abandoning the product line or cutting software support. Your card isn’t going to become a brick.
If you were in the market for one, the window has likely closed unless you’re willing to pay secondary market prices that bear no relation to the original MSRP. And if you were hoping a new high-end NVIDIA GPU would arrive soon to push the 5090 into more affordable territory, the news of the RTX 5080 24GB as a likely replacement rather than a true 5090 successor is genuinely disappointing for anyone who wanted that top-end performance.
The broader context is a GPU market that has been trending badly for consumers for most of 2026, with prices up across the board thanks to component shortages, AI demand eating into supply chains, and limited competition at the very high end. For the gaming audience watching all of this, it’s one more frustrating chapter in what has been a rough year to be shopping for high-end PC hardware.
For the rest of this week’s gaming news, our coverage of Car Parking Multiplayer 2’s impressive free-to-play Steam launch is a refreshing contrast — a game that costs nothing and has been overdelivering on content. And if you’re following the AI conversation in gaming more broadly, check out Dragon Quest creator Yuji Horii’s AI-built game and our piece on No Man’s Sky’s upcoming iOS launch on October 28 — two very different stories about technology being used in genuinely interesting ways.
