Zen 4 sells incredibly poorly, so that strategy requires accepting far lower margins on Ryzen. Not sure whether they can even compensate enough for the higher platform costs.
AMD might be better off doing deals with MB / Ram manufacturers to offer rebates for people who buy CPU + Mobo + RAM rather than further discounting the CPUs.
I expect AMD may have wanted to charge $1,200 for the XTX and $1,000 for the XT. The XTX with higher clocks would probably offer 4090 level raster performance and about 4080 level RT performance* for the 4080 price. The XT would sit between the 4090 and the 4080 in raster but be faster than the 4080 12G in RT.
If the bug is true then it just means AMD will accept less margin on these initial parts and they can refresh later with a beefier 7950 range than they originally planned for. It does mean that N32 could come in pretty close to 7900XT performance but if that is not launching until mid next year to clear RDNA2 stock then AMD might refresh the 79xx range and release N32 at the same / similar time.
* If you look at the AMD slide the 4K Native RT gain for the 7900XTX is about 1.66x that of the 6950XT. Based on meta reviews of the 6950XT and 4080 the 6950XT is about 52.9% of the RT performance of the 4080 in Native 4K. 1.66x that is 87.8% of the RT performance of the 4080. At higher clocks that number would be closer to 95-105% so AMD could have offered perf/$ parity in RT with a big advantage in Raster. Which is what they are doing anyway by the looks of things but at a slightly lower price point than they may have initially wanted. They would have done the cost/benefit analysis and calculated that doing this means they make more in total because they can sell the broken N31 dies then when they do fix the bug the top XTX bin will be really easy to hit as long as the die has no defects.