The demand for hardware to power and train AI models has steadily been devouring computing component supplies. The latest casualty, if a rumor posted by X user Uniko's Hardware is to be believed, could be next-gen consumer GPUs. They claim that the shortage of 3 GB GDDR7 chips will lead to Nvidia outright cancelling the RTX 5000 Super series of graphics cards.
The current RTX 50-series cards use 2 GB GDDR7 chips, and the newer cards are expected to use 3 GB GDDR7 chips. The problem is that not only is RAM fab capacity constrained by AI gobbling up most of Samsung and Micron's production, Nvidia might also save its existing stock of 3 GB chips for high-profit models like the RTX Pro 6000.
[rumor]due to the crazy shortage recently, 3gb gddr7 cant make it to the consumer market for desktop, thus the super series is cancelled.and the current models are expected to be more expensive very soon, because of the increasing cost of 2gb gddr7. pic.twitter.com/yMSMj1MjqJNovember 7, 2025
While GDDR7 production is apparently still ramping up, we would expect that for the time being, the extra capacity will be gobbled up by AI accelerators of various shapes and sizes, including Chinese ones.
Plus, the question remains as to whether Samsung, SK Hynix et al will even bother dedicating fab capacity for GDDR7 when they can adapt existing lines to make HBM at far juicier profit margins.
It wouldn't look good for Nvidia to cancel or delay a potential RTX 5000 Super launch, as the company likely wants to continue to keep its presence known in the gaming space even though it's but a fraction of its income.
The Super cards were never formally announced, though the rumor mill suggests they're meant to come in as direct replacements for the extant 5070, 5070 Ti, and 5080 models, hopefully helping to bring down prices on those cards.
... continue reading