As hard as it is to believe, we’re already some 18 months into the life-cycle of the Nvidia RTX 50 series of graphics cards, with cards such as the RTX 5090 hitting store shelves in January, 2025. Since the launch of Blackwell series of GeForce cards, we’ve had several ‘delays’ for the Nvidia RTX 50 Super cards, which largely shared the same specifications as their non-super cousins, albeit with slightly cranked clocks and higher density memory.
Earliest this year, Kopite7Kimi leaked that the RTX 60 series of Nvidia GPUs would be due to release in 2H 2027, but, everything I’m hearing is that this date is looking less and less certain. About a week ago, I spoke on the YouTube channel that I’d a pretty reliable source tell me Nvidia were thinking to delay the Rubin GeForce cards until 2028 simply due to costs around memory and other components.
I reported this as of course, I should, as purely a rumor – but I kept doing a bit more digging. Unfortunately, much like the situation surrounding RTX 50 Super, the situation of the next-generation GeForce parts isn’t completely clear. For reference, I’d first reported RTX 50 Super would likely be canned or heavily delayed last year (2025), and since then the cards have been off and on again a few times. With last report of them planned to make an appearance in 2027’s CES.
But, that would mean that by the time those cards were released and on store shelves (likely late January/early Feb), assuming the 2H 2027 of RTX 60 was true, it would be perhaps 6-8 months before the next-generation was in full swing. Nvidia (and AMD actually) have certainly done similar in the past, but it would also gobble up a lot of 3GB density memory.
So, I’m still hearing that RTX 50 Super isn’t certain – and indeed, I’d be very surprised if it made an appearance, especially due to AMD having nothing to counter it. Indeed, back in January I reported RTX 50 Super had been cancelled, yet again. This was on top of my other reports from 2025 it was dead, so it being finally laid to rest at this point could well make sense.
But, what about GeForce RTX 60?
| SKU Name | SM Count | Memory Bus | RAM Config |
|---|---|---|---|
| RTX 5070 | 48 | 192 Bit | 12GB GDDR7 |
| RTX 5070 Super | 50 | 192 Bit | 18GB GDDR7 |
| RTX 5070 Ti | 70 | 256 Bit | 16GB GDDR7 |
| RTX 5070 Ti Super | 70 | 256 Bit | 24GB GDDR7 |
| RTX 5080 | 84 | 256 Bit | 16GB GDDR7 |
| RTX 5080 Super | 84 | 256 Bit | 24GB GDDR7 |
Well, following up from my initial conversations last week, a few other sources have (paraphrasing) told me “Yeah, 2027 isn’t happening”.
There’s several problems driving this, the first is Nvidia’s own memory supply isn’t infinite. 3GB memory is still not in high enough densities to be ideal, and memory costs are also very expensive. There’s been several price increases for GPUs – not just Nvidia ones.
Plus, Nvidia ideally doesn’t want the same situation it’s had before with certain SKUs. Let’s pick on the RTX 5070 12GB shall we? The card is perfectly serviceable, but – as someone who reviewed the card, I can tell you that 12GB does bite on occasion. Higher quality textures with Ray Tracing and other settings can start leading to the shortage of VRAM becoming obvious. The RTX 5070 Super is an 18GB card, but ideally, RTX 6070 would have 16GB (for example) of RAM.
Plus, other issues such as higher MSRP on launch is obviously something Nvidia would be keen to avoid, along with shortages. It’s just bad PR. There’s also the problem that if you start loading the cards with lots of VRAM, they become more interesting for folks wanting to run local AI models, which then starts eating into Nvidia’s line up of other cards (albeit, this isn’t an issue so much for say, 16GB cards, but let’s say a 24GB RTX 6090).
Nvidia also want to simply make a priority of silicon (memory, GPU chips etc) for now on AI. As it’s what makes the most money.
So what does this mean, will we definitely not get an RTX 60 series of cards in 2027, and gamers will have to wait for GeForce 60 in 2028? That brings me back full-circle, it depends. If the bottom starts falling out the AI market, Nvidia can procure more VRAM etc, then it makes it more likely. But as of right now… the several people I’ve spoken to have all told me a delay is happening.
As for RDNA 5, I’ve been told that several RDNA 5 based products are releasing next year from one person, but they weren’t able to clarify if this is for gamers, or say professional use. Another source told me that yes, RDNA 5 gaming variants are releasing for gamers in 2H 2027 (to be precise, later in the year, so Q4 maybe?) but they didn’t offer up which variant that is.
Basically, AT0 is the highest end cards – the flagship products. The rumors indicate that this will indeed launch for gamers, but a cut down variant. AT2 etc reduce VRAM, shaders and other specs – so very similar to what we’ve seen with other launches.
The uncertainty of hardware releases isn’t just GPUs though, I mean – to give you an idea of how messy this is, Sony themselves have confirmed they simply don’t know the release date of the Playstation 6. So, if they’re unsure if the PS6 is going to release next year, it stands to reason Nvidia, AMD etc are also a bit unsure what’s happening.
