GoKawiilHardware leaker Kopite7Kimi, who has an exceptional track record, posted on X that Nvidia's RTX 60 series GPUs, based on the Rubin architecture, might not be released until at least 2028. The leaker previously believed that we might see a new generation of graphics cards in late 2027, but now says: "Unfortunately, GR20x is a product from 2028. That means they have postponed again." We reached out to Nvidia regarding the post, and a company spokesperson stated that they "don't comment on rumors or unannounced products."
While Nvidia issued its standard boilerplate response, Kopite7Kimi has a historically near-spotless record when it comes to speaking about Nvidia hardware leaks, and it seems that this time, consumers may have to wait a little longer. Both the Ada Lovelace and Blackwell architectures in RTX 40 and RTX 50 series GPUs share the same TSMC 4N process node, so if the leaker's words are to be believed, it will take Nvidia a total of six years to move its consumer GPUs to the more advanced TSMC 3N or 3NP node that its Rubin accelerators currently utilize.
I hope Jensen can see everyone's passion and launch new products for gaming as soon as possible. Unfortunately, GR20x is a product from 2028. That means they have postponed again. 🙃🙃🙃I will try to minimize updates until there is definite news. (Lessons learned)September 21, 2026
Now, the potential reasons behind the delay could be twofold: The first is the numerous reports we've heard about a potential RTX 50 Super series Blackwell refresh being sidelined, despite reportedly being ready. The claimed reason behind the delay or potential cancellation of the rumored RTX 50 Super series is the use of denser 3GB GDDR7 VRAM chips, which just so happen to have appeared in Nvidia's recently released RTX Pro 5500 and share many similarities with the RTX 5090.
Latest Videos From Tom's Hardware Watch full video here:
So, it's quite possible that the delays in delivering a GeForce product based on the Rubin architecture, which has already begun shipping as "the fastest ramp in company history," is down to careful allocation of resources on Nvidia's part, as the company expects to make $20 billion from Vera Rubin in Q3 2027.
Secondly, Nvidia can only acquire so many resources to supply its products, which demand vast amounts of memory, and has also locked in a certain amount of wafer supply from TSMC. So, it's highly probable that Nvidia is focusing on enterprise and workstation products ahead of a consumer launch, especially as component costs remain eye-wateringly high. Nvidia's highly profitable data center arm is currently experiencing soaring demand from the ongoing AI buildout, meaning there's less onus to deliver a potentially less profitable consumer GPU, while there's still cash to be made from hyperscalers.
It's not unusual to see enterprise products use more advanced nodes before arriving in consumer products, but this time around, at least according to Kopite7Kimi, the Blackwell product cycle is poised to be unusually long. This is particularly bad news for consumers, as the RTX 5090 is experiencing a pricing squeeze, which is seeing costs soaring all the way above $9,000 — if you're lucky enough to find one listed online.
For now, your best bet is to just hold on to the GPU you currently have until the storm blows over and a new generation of graphics cards is released, unless you're willing to pay the dizzyingly high costs that the market demands right now.