Seems like they do, but x86 still has some other advantages. Just in addition to ease of development for ports between console and PC, it's less memory-intensive for x86 instructions compared to ARM ones, from what I've read. And RAM prices aren't necessarily going down while module capacities seem to have been set at 2 GB capacities for the longest time.
So for running certain complex code that games may end up eventually doing, do they trade off CPU performance for a bit more memory and maybe less power used? Or do they stick with more complex CPUs if it means needing less RAM capacity for equivalent tasks? It's not like the CPUs in the new consoles eat that much power anyway, they're based on mobile Zen 2 CPUs IIRC.
X86's main advantage is it's legacy. It's has been used for decades for making and playing games.
But it's not the best CPU arch out there. In fact, Jim Keller said this last year in an interview that the X86 is the worst ISA in the planet.
Considering his experience leading design teams in X86 and ARM, at Apple, AMD, Tesla and Intel, he is one of the best persons to ask.
But then again, he also said that ISA now plays a small role in designing modern CPUs arquitectures. So X86 would not prevent AMD or Intel from making faster and more efficient cores.
This ARM vs X86 efficiency thing has been overblown. Especially when Apple released the M1.
But that efficiency had a lot to do with Apple's access to the most advanced process node, with TSMC's 5nm.
AMD could only use 7nm process node. And Intel couldn't produce anything similar in it's fabs.