I think it’s pretty obvious that, now that AMD CPU aka Ryzen mindshare is basically no longer an issue, they’re trying to solve their AMD GPU aka Radeon mindshare issue, and this is especially useful when they can use their CPU mindshare as a sort of “trojan horse” to then do their aforementioned CPU+dGPU bundling and get their foot in the door with dGPUs, especially when they currently have an efficiency advantage on both the CPU and GPU side vs Intel and Nvidia! (and of course, efficiency is much more important in the laptop space than the desktop space)
I mean, even in the more enthusiast market and not just “normies”, Nvidia mindshare is kind of insane. As stated by people like Tom of Moore’s Law is Dead, you’ll have people in real life asking him about the Geforce RTX 4000 series (which hasn’t even been acknowledged by Nvidia publically) and yet, when Tom asked the person if they’ve any interest in the Radeon RX 7000 series, the person in question didn’t even know what he was referring to.
And for those that haven’t been paying attention to the rumor mill, there’s a very real chance that Radeon RX 7000 series is going to basically do a clean sweep against Nvidia since AMD is going to be using chiplets for Navi 31 and 32 while Nvidia will still be using monolithic.
…and if Nvidia doesn’t figure out chiplets by the time Radeon RX 8000 launches, I think we can safely say they’ll be in big trouble since, much like server CPUs, GPUs also easily scale with “moar cores!”, and we know how much of an advantage that is providing AMD in the server market vs Intel.
DISCLAIMER: it’s possible that AMD wanting to solve their GPU mindshare issue isn’t actually obvious and I simply cannot remember what my stance was before Tom of Moore’s Law is Dead basically put out a video stating that solving their GPU mindshare issue is one of the next big goals beginning with the Radeon RX 7000 series and they’re currently laying the groundwork on the software side of things with the likes of the recently-released FSR 2.0, improved h.264 hardware encoding quality, and AMD Noise Suppression as a counter to existing Nvidia software technologies that people like to point out as an advantage in favor of Nvidia, the latter two being particularly a thing with the game-streaming crowd.
And for those that are wondering, CUDA and/or GPU-accelerated Physx has been basically dead in the gaming world for a while now, so CUDA’s entrenched-ness isn’t really a concern in that market. Also my personal theory is that AMD is looking to use CPU-based accelerator on future CPUs to counter-act the likes of CUDA in the long-term (see also: their Xilinx acquisition), especially since Intel is also going that route.