That actually makes a lot of sense - rather than "Apple designed" chips meaning ARM, what if it means custom Ryzens?
As for why you don't want to emulate x86 on ARM, it has to do with the complexity of the applications. Every iOS application has to run on everything back to an iPhone 6S - requiring iOS 13 gets rid of anything earlier, but leaves open the possibility of a 4+ year old phone with 2 GB of RAM and a Geekbench 4 score around 4000 - this limits how complex the application could possibly be. I don't believe the App Store allows a developer to say "newer devices only"?
Emulation requires that the emulating machine be something like 2-3 times as powerful as the machine being emulated to get the same performance.
Catalyst relies on the fact that most Macs that turn up running Catalina will be powerful enough to emulate an iPhone 6 at reasonable performance - there are only a few edge cases of Macs capable of running Catalina that aren't several times as powerful as an old iPhone (mostly 11-inch Airs).
Going the other way would require the ARM Mac to be several times as powerful as the x86 Mac it's trying to emulate. Given that Adobe apps will end up emulated (unless you're willing to run translated iPad versions - Adobe will almost certainly base native ARM Mac versions on the iPad codebase, not the x86 Mac codebase), that could mean emulating something like a 2015 iMac or a 2017 MacBook Pro for the larger apps to perform decently (not extremely fast - just acceptable). Those Macs Geekbench (4) somewhere around 15000, meaning that the machine that emulates them would have to be around 30,000 - 45,000 (I know Geekbench isn't perfect, but not much else compares Macs to iOS).
The fastest iOS devices also have Geekbench 4 scores around 15,000 - meaning that an ARM Mac would have to be two or three times that fast merely to emulate running Photoshop or Premiere on what users of that software would consider an old, slow computer. A top-end 16" MacBook Pro Geekbenches (using version 4 scores to maintain compatibility with older hardware that lacks GB 5 scores) very close to 30,000, with a top-end 2019 iMac slightly faster and an 18-core iMac Pro brushing 50,000. These are the machines the big Adobe apps tend to run on. What's the chance Apple can come up with something 2-3 times that powerful (let alone two or three times as powerful as a Mac Pro that might Geekbench 100,000)?
AMD chips make a lot more sense - there's not a 2-3 time emulation penalty...
As for why you don't want to emulate x86 on ARM, it has to do with the complexity of the applications. Every iOS application has to run on everything back to an iPhone 6S - requiring iOS 13 gets rid of anything earlier, but leaves open the possibility of a 4+ year old phone with 2 GB of RAM and a Geekbench 4 score around 4000 - this limits how complex the application could possibly be. I don't believe the App Store allows a developer to say "newer devices only"?
Emulation requires that the emulating machine be something like 2-3 times as powerful as the machine being emulated to get the same performance.
Catalyst relies on the fact that most Macs that turn up running Catalina will be powerful enough to emulate an iPhone 6 at reasonable performance - there are only a few edge cases of Macs capable of running Catalina that aren't several times as powerful as an old iPhone (mostly 11-inch Airs).
Going the other way would require the ARM Mac to be several times as powerful as the x86 Mac it's trying to emulate. Given that Adobe apps will end up emulated (unless you're willing to run translated iPad versions - Adobe will almost certainly base native ARM Mac versions on the iPad codebase, not the x86 Mac codebase), that could mean emulating something like a 2015 iMac or a 2017 MacBook Pro for the larger apps to perform decently (not extremely fast - just acceptable). Those Macs Geekbench (4) somewhere around 15000, meaning that the machine that emulates them would have to be around 30,000 - 45,000 (I know Geekbench isn't perfect, but not much else compares Macs to iOS).
The fastest iOS devices also have Geekbench 4 scores around 15,000 - meaning that an ARM Mac would have to be two or three times that fast merely to emulate running Photoshop or Premiere on what users of that software would consider an old, slow computer. A top-end 16" MacBook Pro Geekbenches (using version 4 scores to maintain compatibility with older hardware that lacks GB 5 scores) very close to 30,000, with a top-end 2019 iMac slightly faster and an 18-core iMac Pro brushing 50,000. These are the machines the big Adobe apps tend to run on. What's the chance Apple can come up with something 2-3 times that powerful (let alone two or three times as powerful as a Mac Pro that might Geekbench 100,000)?
AMD chips make a lot more sense - there's not a 2-3 time emulation penalty...