To see if this actually worked, I downloaded the (x86) Efinix FPGA tools (they're a lot smaller than Vivado!), which installed fine, and Efinity launched great.

What it doesn't say is that this is the Arm version of Windows 11, however it lets you run x86 applications via the Windows x86 emulator.

I let it do so, and it installed it seamlessly

Parallels installed fine, and asked if I wanted to go grab Windows 11.

I thought the same as you, but on a whim, tried to install it since it said it ran on the M1. I have a Parallels subscription, and just got a new Mac Studio Ultra. Progress is being made along those lines. I say 'get what works now' and just be happy.

At that point, I'll have to rebuild everything, move off of my Eagle based world, and KiCad will be the new thing, and the new CPUs will be even nicer than M1. My summary is that a modern, fast i7 Mac and VMware Fusion will give you the best of both worlds for a few years more until it all falls off the grid. On the laptop, the CPU heats up too much for my comfort, so I have the desktop for brute force goofiness like this. Still, I do use the desktop when I cannot figure out how to fix the simulation and I need to grind on it repeatedly while I figure out how to make it un-broken. Fixing the simulation always provides the best speed-up. I have a 6 year old i7 MacBook Pro that's 2x slower than a fancy new i7 10700 Win10 desktop, and it's almost never that I notice any significant speed difference. It's not like a factor of 10, and honestly, unless your simulation is really a mess, it's not often that they grind that slowly.

To the OP, I don't think Parallels will be able to provide a Windows VM on M1, and honestly, an M1 is only about 2x faster than a fast i7. Regardless, the tools are 'what they are' and you can either get work done with them or not. What would be more ideal is some sort of ability to efficiently use one networked library, or even a remote directory.

The models had to be on local disks for performance, and the same tedious 'make a part' procedure had to be used for both platforms. The basic installation was tedious for both Mac and PC, and not fundamentally different for either platform. Is it really much different between Mac and Windows? I have around 450 models that are on my Mac and also on a Windows 10 machine (both of which run LTspice XVII) that I had to install by hand onto each machine and sync by hand between the two.

Altium designer mac os x