x86 vs. ARM/RISC-V has fewer registers which can mean more spill/fill to the stack which results in more instructions to do the same work.
The instruction decode length for x86 has been creeping up with 64-bit encodings and SIMD extenstion (prefix bytes) as you can see it's 3.96 for x86 vs. 4.00 for ARM.
They should drop their custom ASICs and just stick a Qualcomm Snapdragon as a replacement. They could use Android and create a platform with an app store with APIs to access their sensor. Opening up the hardware in this standard way to developers would really add a lot of value to their cameras.
I have no idea why they haven't done this already.
You need to be competitive on single thread performance to have a chance at datacenter. Amdahl's law is still very relevant. Up until very very recently, the CPUs were not up to par.