I'll also weigh in, because I have opinions 
, but they're also my opinions.
I like a Mac Book, personally. I've been working on a 16GB macbook for years and do pretty well with 16S analyses. (I bought a 13" a couple years ago as a work/personal computer before I started a recent job.) I feel like the M1 chip does a decent job allocating memory and I haven't had any performance issues. (I do occasionally leave my computer to "cook" overnight, but I put an oven rack underneath so it stays cool 
).
I'm also not computationally savy enough to work between two OS. I like the fact that my Mac has a unix backend and I can do a lot of Unix stuff, but I can also get microsoft word/adobe which I need personally and professionally. It's a more integrated experience for me.
You might also look at something like EC2 or Azure if you have big computing projects. It will save your machine and might be easier to bill (I have burned out laptops.) If you configure it correctly, you can use it as a pop up server: run DADA2, train your classifier, and do beta diversity on the server and then pull your files down locally for visualization and analysis.
Best,
Justine