Developers usually desire a fast platform. While it is true most OS hosts can virtualize Linux, there are at least three important caveats when considering macOS. The first is that the file system performance is poor and chews up CPU when using file sharing [1]. The second is that if you are targeting x86_64 containers, you'll want to avoid full-virtualization and/or the Rosetta x86 to ARM translation later. Finally, the third is that if you work in machine learning, chances are you use NVIDA on the server, so you may want to do simple smoke testing of your containers locally, but Apple no longer ships computers with NVIDIA graphics cards.

Now some of that is alleviated if you are building non-machine learning ARM images, but you may run the risk of not having ARM versions of third-party containers you need.

I personally regret not thinking ahead about this and buying a MacBook Pro a few years ago. Either with Linux directly, or with Windows WSL2, I don't think developers need to buy a macOS machine unless they need to build iOS or macOS apps. And even then, you can rent time on GitHub-hosted Actions macOS runners instead.

[1]: https://docs.docker.com/desktop/mac/#file-sharing

Virtualization is a far cry, performance and experience wise, from running the OS natively.

Not anymore. Not on the M1s. They're far and away faster than anything but the beefiest desktop. A lua parser benchmark I ran in a VM on my M1 beat my 5950X by a factor of two.

The only remaining limitation are missing some ARM docker containers (which is slowly being solved) and the machines not being suitable for ML (in which case you need a desktop or to rent server time on the cloud).

What work are you even doing to push virtualization performance to its limit?

I cab run a Ubuntu server VM at good performance on VMware.

Is this satire?

As in? I'm serious.

It is definitely not easier to passthrough an NVIDIA GPU that I require for machine learning to a virtual machine on an M1 Mac versus just buying a laptop with an NVIDIA GPU. I don't even know if M1 supports eGPU, have not heard anything about it, and I doubt PCI passthrough or whatever you use for eGPUs works on macOS. Sometimes hardware does matter, and it glaringly matters (esp. w.r.t. CUDA).

Also consider gaming. Considering that GPU virtualization is not really a thing still, I cannot imagine, for example, gaming being easier on macOS, and the VM requires additional setup too.