Custom RX 7000 graphics cards from third-party card manufacturers are set to release later than reference boards, and can be quite expensive to boot, if the latest from the GPU vine is to be believed.
VideoCardz (opens in a new tab) reports that the Board Channels forum (in China) states that the only RX 7900 XT and 7900 XTX graphics cards that will initially appear on AMD’s launch day (December 13) will be the reference boards.
Those are the ones built to AMD’s reference specs, but there will also be models with beefier settings (higher clocks, better cooling), as usual. However, the latest third-party variants won’t debut until a week or two after the reference boards hit shelves. That would point to a late-December on-sale date for these custom boards (and perhaps things could even be pushed back to early 2023, since the very end of December is a bit of an odd place in terms of time).
Worryingly, the Chinese source (treat all of this very cautiously as with any speculation) further states that custom boards can be “much more expensive” compared to initial reference models.
Analysis: A tough dilemma for AMD?
Such high-end custom graphics cards naturally feature accelerated clocks as mentioned, and often better quality components, cooling solutions, etc., to keep temperatures in check with those faster clocks. In short, it’s not a surprise to see the beefier 7900 XT and XTX models be more expensive, it’s perfectly expected.
However, what is a bit worrying here is the way things are phrased as if they were much more expensive, and that sounds sinister. In the same way that the RTX 4080 ended up priced heavily with some third-party variants compared to MSRP, will we see the same sort of thing for AMD’s new RDNA 3 GPUs?
That seems to be the suggestion, which would be disappointing as we expect AMD to keep its RX 7900 variants relatively affordable, at least when compared to Nvidia’s RTX 4080. (Remember, Team Red isn’t going to try and compete with the RTX 4090, but rather the 4080, that’s clear from what we’ve seen of AMD’s positioning and press materials.)
There’s another key element here, and that’s the kind of volume that AMD and its card-making partners will be mustering in the initial release of RDNA 3. Obviously, that’s going to be critical to pricing, because if inventory isn’t plentiful and they sell out fast, well it becomes a vicious cycle as cards will sell out even faster as scalpers put on their price gouging hats and buy stock to resell.
According to Board Channels’ source, the supply of RDNA 3 graphics cards is initially expected to be “sufficient”, but that’s a difficult judgment to make at this stage, and one we’d take with several handfuls of salt.
AMD and its card makers could be in a difficult position in terms of evaluating stock, as Nvidia has been selling pricey Lovelace GPUs for some time now, swapping out a lot of units from the RTX 4090 in particular. (And also the RTX 4080 to a lesser extent, though given its price relative to the 4090, that GPU has been received in a much less enthusiastic fashion, shall we say.)
Expensive graphics cards are a niche proposition, particularly when a cost-of-living crisis is in full swing, with sky-high energy bills, so there’s a danger to AMD that Nvidia has already eaten too large a slice of this. pie. ? Remember that while the RX 7900 XT and XTX are certainly more affordable than Nvidia’s RTX 4080, they’re still pricey boards, reference or not. And perhaps this kind of thinking is fueling concerns at AMD that the company doesn’t want to go overboard by pushing stock RDNA 3 too much initially.
This is pure speculation, of course, but that could tie in to why you want to hold off on third-party custom models until a bit later, and spread out supply to perhaps better match anticipated demand early on.
Overall, it’s pretty clear that AMD might have some tough judgments to make and several possible considerations to balance. So it’s not unimaginable that there might be a last-minute timeline shakeup behind the scenes as the Red Team looks at Nvidia’s sales sheet with Lovelace, which is apparently around 160,000 units and counts for the RTX 4090. and 4080, and tries to calculate the best approach to launch the strategy and volume.