The trouble for AMD is that because of chip fabrication there isn’t really as much of a viable “pile them high, sell them cheap” option, and if you look at AMD’s market share over 20 years they’ve never seen a lasting benefit from being the budget brand. If their GCN (and earlier) based cards were trying to build up a large enough loyal audience for future higher priced GPUs, they failed.

Personally I’d say a related issue has been nvidia introducing ray tracing, rasterization was hitting a wall in terms of increasing the visuals offered each generation so they spent a few years of product development making it and the complimentary DLSS tech. Going back to around 2018 when it was a bleeding edge ‘vision of the future’ thing, IIRC Jensen Huang said something along the lines of “by around 2023 everyone will want to get a GPU with RT”, except right now it’s still only higher-end GPUs where you’d want to enable it for good framerates, it’s not a factor through the whole range (i.e. you’d ignore it exists at the low-mid range). And importantly while RT/DLSS has clearly given them a competitive advantage they seem to overestimated how quickly AMD would get that capability up to ‘decent’ performance to where the broader market including developers, and including cross-platform console games will lean into it. If nvidia’s graphics GPUs slow down in selling because there’s nowhere for advancement to go, they don’t care because they are selling tons to ML (and so are AMD apparently)

I wouldn’t be surprised if this is the first stage of AMD throwing in the towel for competing in dGPUs, taking their foot off the gas. They’ll continue with radeon for iGPUs in Ryzen and the rumored PS5pro RDNA3.5 refresh, but beyond that it seems like they’re just conceding second place to intel, assuming intel’s Arc Battlemage and beyond is competitive.