By the way, I also know a few things about MMUs, cache coherency, and the like, and your point is still not well taken because x86 is as complex or more so in each case. (For work I've done see, for example:
http://ieeexplore.ieee.org/document/641683/
https://www.ecse.rpi.edu/frisc/theses/MaierThesis/
I was also one of the designers of AMD's first 64-bit opteron/athlon 64, UltraSparc V, and Exponential x704, among others.
and for giggles
https://www.law360.com/articles/784676/lg-gets-amd-graphics-patent-claims-axed-by-ptab)
Interesting work.
I did x86 cache controller design.
I was at Amdahl in the late 80's doing design in the computer development group.
But I have written papers on cache coherency, audio digital signal processing and unified flow for FPGA to ASIC transition.
My area of specialization was processor architecture, but have spent more time doing custom processors for video encode/decode and compression and not general purpose computing.
But, credentials aside I think you missed my point.
My overall point is that an A11 does not have the equivalent complexity of an Intel desktop processor.
By that I mean the additional peripheral interfaces needed for a balanced system.
Compare an A11 to the newly announced i9.
That's
not an A11 with a few tweaks.
Can Apple design a processor to compete? Probably, if they hire enough people.
Does it make sense? Only if the ROI is in the billions of dollars because that is the kind of commitment they are going to need to make. It's not a single processor, they are going to need a family of processors and a processor roadmap.
Right now the logistics and people aren't in line with Apple replacing Intel in 2020.