Nvidia could be really gunning for AMD with its rumored trio of refreshed RTX 4070 and RTX 4080 Super graphics cards, or that’s a possibility according to a well-known YouTube leaker.
This is Moore’s Law is Dead (MLID) who in his latest video talks about said Super spins, although we should make it crystal clear from the outset that pricing decisions haven’t yet been nearly finalized, the leaker informs us. (Obviously they never are until close to launch, anyway).
However, if Nvidia goes to the lowest price tags that are purportedly being mulled by the firm – apply salt here, and lots of it – these inbound graphics cards might prove a big old thorny problem for AMD and its popular RX 7900 models.
Okay, so down to the nitty-gritty: MLID gives us potential names, specs and performance levels, and prices of the three graphics cards that you’re likely familiar with by now, as they have all been bubbling in the GPU rumor cauldron of late.
MLID sources indicate that these cards will most likely be an RTX 4080 Super 16GB, with an RTX 4070 Ti Super 16GB just below it (both using the AD103 chip), and then a straight RTX 4070 Super 12GB (AD104). This mirrors a lot of what we’ve heard on the grapevine lately – including the ‘Ti Super’ theory, a name we’ve admittedly dismissed in the past, but it’s now seeming more likely to happen (we still think it’s confusing, but hey).
Notably, though, it waves away any speculation that the RTX 4080 Super is going to use the AD102 GPU (hence no 20GB of VRAM, either) rather than AD103 (which there has been a fair bit of chatter around lately). AD102 is the GPU that powers the RTX 4090, and MLID makes it clear enough why a cut-down version of this chip wouldn’t be used in the RTX 4080 Super.
The leaker argues that even duff AD102 chips, ones with some wonky cores that aren’t good enough for the 4090, which could theoretically be used in a 4080 Super (with those faulty cores disabled, of course) won’t be employed in this way because they can be better repurposed elsewhere. Namely in Nvidia’s RTX 5000 (Ada Generation) GPU, a graphics card for AI and professional use (not to be confused with the next-gen Blackwell cards for consumers, of course).
What about those all-important specs and prices? MLID gives us upper and lower bounds for what Nvidia might do with these Super models, so let’s start with the worst-case scenario.
If this is the way Nvidia goes, we’re told we are looking at an RTX 4080 Super 16GB which is pushed to be 5% to 10% faster than the vanilla RTX 4080 (using the full CUDA core count of AD103, and clocks being upped). Pricing will be $1,199 to $1,249 in the US (and appropriately scaled to that level in other regions, no doubt, with the usual little bit extra thrown on top).
The RTX 4070 Ti Super 16GB will be primed to sit just below the current RTX 4080, within about 10% of the latter’s performance, priced between $849 and $999 in the US.
Finally, the RTX 4070 Super 12GB will be pitched between the RTX 4070 and 4070 Ti for performance, costing $649 to $699. As a result, the vanilla RTX 4070 will get an official price drop to make more room for this graphics card, and MLID even theorizes that in this scenario, the current RTX 4080 could be discontinued (which isn’t the first time we’ve heard that idea, either).
Right, onto the good stuff – the best-case scenario for how Nvidia could tune the performance and pricing of these products.
In this case, we’ll seemingly get an RTX 4080 Super 16GB that’s about 10% to 15% faster than the RTX 4080, priced between $999 and $1,119.
Then the RTX 4070 Ti Super 16GB would come within 5% of the performance of the 4080, priced at $799 to $899.
And finally, the RTX 4070 Super 12GB could theoretically perform 5% off the pace of the 4070 Ti, so be very close indeed, while priced at $599 to $649.
This scheme of things would likely usher in the discontinuation of both the RTX 4080 and 4070 Ti, we’re told, and indeed it’d be difficult to see how those graphics cards would fit in well with the new range if price/performance is realized at those levels.
Analysis: Decisions, decisions…
Which of these paths might Nvidia follow? Well, that depends on how worried Nvidia is about this upper mid-range GPU territory, and how much Team Green feels it needs to strike back against AMD’s RX 7900 – perceptions which could obviously change over the Holiday season and as we stride boldly onwards into 2024.
On the one hand, personally we have trouble swallowing the idea that Nvidia will come up with really competitive pricing, mainly because the company has consistently pushed the boundaries of acceptable price tags with Lovelace – though that attitude is relenting somewhat of late.
On the other hand, though, consider these two factors: the RTX 4080 is seemingly an embarrassment in terms of how bad sales are, if the rumor mill is to be believed – and there’s lots of chatter to indicate this is the case. Plus AMD is really cleaning up and doing well with its RX 7900 offerings (prices fell quite a bit for these GPUs, but have since risen somewhat).
So maybe Nvidia feels it needs to take action to save its reputation as the top dog graphics card provider, and at least put up a fight in the upper mid-range space. And it will be a fight – AMD will have to respond to the lower prices theorized for these Super cards, for sure, with sizeable RX 7900 cuts presumably (or maybe even a full RDNA 3 higher-end refresh to hit back).
Whatever the case, this is shaping up to be good news for consumers hunting for one of the best graphics cards out there. Although the lower bounds MLID is talking about – an RTX 4080 Super that’s markedly faster than the 4080, but pitched at $999… is that really going to happen? Or an RTX 4070 Super that’s very nearly a 4070 Ti, but sits at $599?
We find that very difficult to believe indeed, but the overall gist that a more competitive – maybe considerably more – angle is going to be pursued by Nvidia is perfectly believable given the situation with the RTX 4080 right now.
And that’s definitely a hope to hold onto, while firmly bearing in mind that firstly, this is all speculation, and secondly, Nvidia hasn’t made any of these decisions yet, and won’t do for quite some time (at least for pricing anyway).