I can see only one scenario where it makes sense for Microsoft to buy Canonical: to prevent Amazon or Google from acquiring them and doing something anti-competitive (or as much as can be done legally in that respect) and making life difficult for customers using Ubuntu on Azure.

The comparison of relevance is Github. Microsoft was using Github so much already that the prospect of someone else buying them was an issue (not to mention the joke that buying Github was cheaper than paying for Microsoft's growing use of the platform). In practice, has Microsoft done anything to Github that has limited them in any way? The biggest change is that Github has a super-rich benefactor and they can offer free accounts to everyone now... and their budget for hiring/retaining top talent has been boosted as well.

Microsoft has more than enough money to do the same with Canonical and if they do I think it will largely be a matter of making sure Canonical doesn't need to worry about finances anymore and can hire whomever they need to build and deliver features. And yes, it's possible that Azure Linux applications might be available on Ubuntu first but extending and extinguishing Ubuntu would be massively opposed to making money on Azure -- I don't see Microsoft doing that.

Why acquire Canonical when they can just create a Redmond Linux distro and cherry-pick the best aspects of RHEL, Debian, or Gentoo -- or just plain fork something out there now and add the option of Microsoft corporate support for any manager wanting to buy a throat to choke in case there's a production hiccup?

Or better yet: a Windows-like or Windows-compatible desktop that can run on any number of systems (and maybe corporate support is available for Debian, CentOS)? Let's not forget that the co-founder of Gnome has been working for Microsoft since 2016.

Why do you have to jinx it?