Nvidia reportedly halts GeForce RTX 5090 production in favor of AI data center and professional GPUs — impending supply drought expected to drive up prices, RTX 5080 24GB rumored as new gaming flagship

RTX 5090
(Image credit: MSI)
35

Nvidia has reportedly ceased selling its GB202 graphics processors for consumer-grade GeForce RTX 5090-series add-in-boards (AIBs), including 'D' variants for the Chinese market, according to a leaked notice from an unknown factory cited by @hongxing2020, who tends to have insights into the graphics card market. Another prominent leaker @Zed__Wang confirmed the report and said that from now on the GB202 silicon will be exclusively used on data center and professional graphics cards.

"According to a notification from the factory, Nvidia will no longer manufacture any RTX 5090 models, including the D variants," a message cited by @hongxing2020 reads. "Going forward, the highest-end GeForce graphics card will be the RTX 5080 24GB, while the RTX 5080 16GB version will be discontinued. Please spread the word."

Latest Videos FromTom's Hardware
Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • PEnns
    Oh NO!!! Just as I was about to shell out $25,000 to buy one!!

    I wonder how many fellow gamers will be crying now. But don't worry, Jens loves gamers, he said so himself.
    Reply
  • rluker5
    So will the discontinued card no longer continue to be featured in game performance reviews, or will it at least come with a *discontinued asterisk?
    If so my lowly $599 9070XT is moving up the ranks in those charts.

    Imagine what a 6090 could sell for if it is highly suspected that it is a limited production run only.
    Reply
  • ezst036
    These are the actions that an AI-centric company would take.

    Nvidia 20 years ago would not have done this to gamers. I would consider posting a picture of Linus with the finger since he is specifically talking about Nvidia at that point, but, with respect for the moderators - no.

    On the flip side does anybody see the silver lining? Now you won't have melting 5090s anymore.
    Reply
  • bourgeoisdude
    The 5090 has basically been sold as AI GPU for a while anyway. This really shouldn't be surprising.
    Reply
  • bill001g
    Maybe the game developers should ask the AI to help them make games that are coded more efficiently. They seem to think that they can just let the video card magically solve all their problems. They have for a long time posted so called in game footage using stuff like 5090 that nobody could afford. Now what recommend video cards that no longer exist if you want to use the top settings.
    Reply
  • hotaru251
    i mean nto a shock?
    Jensen yrs ago stated they are an ai company now just havent changed the name.
    Reply
  • Kieran Menon
    bill001g said:
    Maybe the game developers should ask the AI to help them make games that are coded more efficiently. They seem to think that they can just let the video card magically solve all their problems. They have for a long time posted so called in game footage using stuff like 5090 that nobody could afford. Now what recommend video cards that no longer exist if you want to use the top settings.
    What, devs learning the lost art of game optimization instead of using higher specs as a crutch?

    NEVER!
    Reply
  • thisisaname
    Surprised it took them so long when instead of putting the chips in a cards selling for only 2K they can put the chips into an Ai card and sell it for well over 10K.
    Reply
  • emitfudd
    Nobody wants to pay 5K for a damn GPU anyways.
    Reply
  • Stomx
    Translating this rumor into plain English: recent advances in Mixture-of-Experts (MoE) architectures have shown that a single good old RTX 3090 on a PC with 128–256 GB of RAM can handle your AI workloads locally, delivering 30 tokens per second during generation and 60 tokens per second for prompt processing on very competitive AI models with top 5 intelligence ratings — jeopardizing the profits of the entire AI industry. The RTX 5090 almost doubling these numbers. Publicized GPT 6.1 Astra speed is by the way 48 t/s. Two 5090 literally put all on halt
    Reply