According to YouTuber “Moore’s Law is Dead”, NVIDIA is stockpiling its GeForce RTX 3060 graphics cards to overwhelm the launch of AMD’s Radeon RX 6600 XT. As you might have noticed, AMD’s supply problems, especially on the GPU side have been much more severe than NVIDIA’s. The reason being that the former relies on TSMC’s N7 node for all its CPUs and GPUs, as well as the custom PS5/Xbox Series X|S SoCs which puts a lot of strain on the capacity.
NVIDIA, on the other hand, is one of the few chipmakers using Samsung’s 8nm LPP node, allowing it much more flexibility with respect to the capacity allocations. Going by leaked benchmarks, it would seem that the Radeon RX 6600 XT is faster than the RTX 3060 and in some applications, even the RTX 3060 Ti.
However, if NVIDIA can procure more supply alongside a price cut, it won’t have much trouble suppressing AMD’s Radeon RX 6600 XT. The company’s RTX-DLSS marketing (to a little extend, genuine) has been rather controversial, but the worst part is that it actually works. Most gamers believe that ray-tracing “RTX” was invented by NVIDIA (it wasn’t, it has been around for as long as computer graphics).
Additionally, it appears that AIBs are prioritizing the supply of the non-LHR variant of the GeForce RTX 3070, simply because it sells at a better price. NVIDIA doesn’t seem to mind this and hasn’t placed any restrictions on its board partners over this, and they are taking full advantage of that. In fact, most stocks of the RTX 3070 arriving at retails continue to be standard variants, when we thought that the production of those was stopped back in late June.