
If the latest rumors pan out to be accurate, then NVIDIA could soon be ending production of its current flagship graphics card, the GeForce RTX 5090, only to replace it with an upgraded version of the RTX 5080. So what’s a company to do when its highest-priced consumer product edges closer to its lowest costing profesional offering? Well, in NVIDIA’s case, it looks like it’s going to opt to cease production of said product, namely the GeForce RTX 5090, which originally launched with an MSRP of $1,999 and can now be found for upwards of $9,600. Meanwhile, the RTX 6000 Pro, with its 96 GB of memory and using the same GB202 die, originally launched for $8,655 and now goes for $16,000.
Much like its gaming counterpart, the RTX 6000 Pro is available in different versions and price points, but aside from more GDDR7 VRAM, it features the fully unlocked GB202 die with 24,064 CUDA cores vs. 21,760 on the RTX 5090. The two also share the same 512-bit memory bus. In addition, the RTX Pro 5000 series also uses the GB202 die but with 48GB GDDR7 on a 384-bit bus and goes for between $8K and $10K. According to industry info leaker MEGAsizeGPU, NVIDIA is choosing to use the GB202 die exclusively in its RTX Pro series graphics cards.
All future GB202 supplies are RTX Pro exclusive
— MEGAsizeGPU (@Zed__Wang) October 10, 2026
If true, this would be yet another setback for the consumer gaming market that has not seen any new GPUs in 2026 from Team Green and NVIDIA’s partners. NVIDIA already competes with its partners through sales of its in-house manufactured Founders Edition graphics cards, but by ceasing production of the RTX 5090 and its further cutdown overseas variant, the 5090D, it will mean one less product its partners have to offer. If there can be any silver lining to this, another hardware info leaker has claimed that the long-rumored GeForce RTX 5080 24GB is back in the game and factories are already being notified it will replace its 16GB counterpart. The following, per hongxing2020, translated via VideoCardz:
“We have received a notification from the factory: NVIDIA will no longer sell any RTX 5090 models, including the D variants. Going forward, the highest-end GeForce model will be the RTX 5080 24GB, and the RTX 5080 16GB version will be discontinued. Please take note.”
In a sense, the RTX 5080 24 GB already exists in the form of the RTX 5090 mobile that uses a cutdown version of the GB203 die seen in the desktop RTX 5080. It’s long been said that the current RTX 5080 16GB already uses a fully unlocked version of GB203, so a memory upgrade is the most notable performance improvement, and perhaps higher clocks, that could be made for this card, which has long been rumored to be a “SUPER” product. However, before anyone begins jumping up and down and waving their arms in excitement, pricing is sure to be an issue, and more than a few are expecting this card to launch at $2,000, but good luck getting one even if it does, thanks to scalper bots and an obvious demand for local AI usage.

Discussion (17 replies)
Join Discussion →Can only wonder how high 5090 prices will go if this becomes official. Will they go higher than a Pro 5000 or Pro 6000, who knows but just in case anyone is thinking of getting a Pro 6000 for gaming if it does, well, the 5090 wins in most scenarios according to LTT but that shouldn't really be a surprise since it or its drivers isn't optimized for gaming.
[embedded media]
Honestly this just means the GB202 (or G202 whatever it is) process is refined enough now that they don't have the failure rates of the early chips so there is more capability of making them the FULL chip enterprise models. And with the cost of memory chips being where it is, they are choosing to phase out 32 gig cards for 24 because The cost difference between capacities and availability are such that it just makes sense.
There is no 'loyalty' to the consumers that funded them before. We can't touch the circular money train that they have established that will be their downfall.
They are probably more pissed that the 5090 is too cheap for LLM use and eating into their RTX Pro card profits. The VRAM reduction should make it less desirable to companies buying it in bulk and maybe the focus shifts back to gaming? Would be a nice change if that happens.
I mean I got my 5090 for gaming... but also for tinkering with LLM's. I always thought of the DIY hobby as a gateway to real IT skills. And here we are that being exactly what the makers are afraid of.
That and the memory bandwidth reduction of the 5080.
24GB is still more than 'enough', but the cards will definitely be less desirable.
However, I don't think that having 24GB will do much for the 5080 gaming-wise. This sounds more like fabs are moving on to 3gb GDDR7 and thus are phasing out the 2gb stuff that Nvidia is currently using. Nvidia had planned to use that from the start if the rumors are true.
And whether x90 or Titan, the top-tier card has always been that sort of bridge in between. I remember the same, more or less, being said for over a decade now.
so my understanding is that nvidia is now diverting all 5090 supplies to professionals
AMD has a pro card MI350P (but is HBM 144gb — so very costly)
[URL unfurl="true"]https://www.phoronix.com/review/amd-instinct-mi350p[/URL]
AMD is probably planning on releasing a prosumer version of gddr7 based RDNA 5 AT0 asap to partake the profits
in turn Nvidia might try to pre-empt AMD's schemes by launching a 6090 pro version
effectively no top end gaming card, until & unless Open AI goes bankrupt !!!!!
Lets be honest here, the 5090 is not a top end gaming card, its insurance against AMD, and basically being paid for to be talked about.
More of a technological mandigo against AMD if you will.
Never understood why AMD doesnt do the same, ever; truly must be they cant, because the strategy makes sense.
That said nvidia may cut the card for a bit, but will quickly roll it out if AMD tries some funny business, and or if this AI mess calms down some.
The 7900 XTX just recently went out of stock. Gamers and other users don't want Radeon halo cards in general because of the Nvidia bent neanderthal youtubers hyping up RTX, DLSS5 and other proprietary crap. The situation is improving with the adoption of Linux as Radeons work better than Geforce cards there but the market is still pretty small. My guess would be that the Radeon group is treated as second class citizens at AMD due to which they don't get enough resources and funding to properly compete with Nvidia. AMD is also averse to sinking money in order to compete like Intel did with their ARC series.
I don't see this.
Nvidia has no problem selling what they make with good margins too boot.
AMD really just isn't even relevant.
Not saying people wont buy, not sayin they dont make money, but the 5090 is an absurd card, all around... With a small plug yes.
Absurdly large, absurdly costly... Its great yeah of course.. But It would not exist at all if AMD did not nibble at the 5080 with their efforts.
did you look up the MI350P
I am sure AMD would like to sell a much cheaper gddr7 version of the same. that is where AT0 comes in
[URL unfurl="true"]https://www.phoronix.com/review/amd-instinct-mi350p[/URL]
Nvidia has been roughly two years ahead on performance and features for... a long time.
I don't see AMD as being a motivator. Nvidia owns the market (~90%) and simply competes with their own previous generation.
Why would they bother when they can sell it in considerable volume at twice the price to data centers and businesses relying on AI?
The current GPU mess is mainly because even small businesses are buying up gaming GPUs.
again I bring your attention back to the MI350P
It is for professionals — not data centers
that is whom nvidia & AMD are targetting the pro cards