I disagree, "cloud" is extracting basic Linux functions into as many proprietary services as possible because businesses would rather deal with obscure YAML configurations than ever having to touch Linux-proper.
I'm sure the vast majority of businesses can handle ~10 min of scheduled downtime per week necessary to restart everything.
Now, database replication, not having to waste time to run/maintain clusters (be it Kubernetes or Elastic stack or something else), that I believe is well worth the money to offload to someone else, but even there you can get a much cheaper deal with someone that's not one of the three big cloud providers. I will also concede that Firebase is genuinely nicer to work with than its alternatives (Supabase very much included).
It's frequently simplifying things so that you don't have to worry about managing a server at all.
I run a PowerShell script once a day at noon, and I have no idea what kind of server it's running on, where in the world that server is, how much memory it has, or any other details. I get about a CPU, a dozen MB of memory, and a tiny amount of network capacity for about 5 seconds.
This is a very different experience from "We will rent you a VPS by the month".
That's like saying "Cars are a marketing term for internal combustion engines."
Clouds give you software-definable load balancers, networking, clustering, integrated systemwide security, and a boatload of managed services like message queues, databases, AI training and inference, etc. etc.
No-one sane implements all that using a collection of VPSes, because of a simple principle of business: it's more profitable to focus investment on your core competencies, and for almost all companies, managing a non-trivial computing infrastructure is decidedly not a core competency.