Here is a 6ghz overclock of raptor lake almost hitting 40,000 cinebench and 6ghz is target top speed of the extreme edition 13900ks
How many customers will be doing this? And how many actually sold SKUs are capable of this overlock? Or, to put it more broadly: are we talking about extreme possibilities or are we talking about the average mass-produced device?
For extreme cases, sure, x86 will win in raw performance hands down. Simply because you always have the possibility to tweak your system and choose individual components capable of higher performance. Ultimately, you will be able to reach your 6Ghz overlock even if it means that you have to buy 20 CPUs in order to win the silicon lottery.
But I just don't see how these extreme cases affect the discussion around Macs even in the slightest. The number of users who would go all that way to overclock the hell out of their PCs are negligible even on the PC market, and none of them would ever be interested in a Mac. So in the end the discussion is about ready-made PC boxes. For laptops, there is no competition whatsoever as these record-breaking Raptor Lake scores recorded at 350 watts simply don't matter. For the desktop, it will probably remain like it is now — Apple being slightly slower but 5x less power. Of course, if the M2 Pro is indeed 3nm based (20% more performance than M2 cores) and can run at full turbo in a desktop configuration it should score around 2.2K GB single-core and 20k multi-core, which would put it within the 5% of the Raptor Lake at 5x lower power consumption.
In the end, my current M1 Max is competitive with the desktop Alder Lake for many of the workloads I care about (software development and data wrangling). I expect this pattern to continue unless either Apple or Intel come up with something new. So far, it doesn't seem like Raptor Lake will be anything new.