This is true but we are referring to the GPU. Will future NextGen games and engines benefit from more GPU cores vs a smaller number and higher clocks? That seems to be the case when it comes to CPUs.
On CPU’s it depends on how the high clock was achieved. High clock with a deep pipeline and bad branch prediction yes while in other cases you might talk about less than linear scaling but you are scaling nonetheless.
CPU and GPU are both under the same virtualisation layer and both are affected albeit MS always promised a very small performance hit they never promised it was free.
It depends on the workload you run on the GPU I guess, graphics are still very parallelism friendly but complex shaders using lots of registers and having branchy logic you end up having more problems keeping all those thousands of processing elements fed and being more latency sensitive and having a higher clockspeed can help (and if the shading results is dependent by previous rendering steps or neighbouring pixels you introduce dependencies where doing something faster gets the job done more efficiently).
TL;DR: some scenarios will show off the much higher CU count of the XSX (albeit held back a bit by XSS which seems to have a considerably lower geometry processing rate too) and some others will be tuned to PS5’s strengths while tons of third party titles will be evenly split with minor differences between consoles.