The problem is in almost all of these cases the workload doesn't match the financial capacity.
Basically, there is an absurd lot of stuff you have to know for each of these fields, which is why domain experts develop in the first place - it's just too much to handle for a generalist unless you happen to run into a person on the autism spectrum with eidetic memory (these are rare and worth their weight in gold). So, assuming at least basic levels of redundancy, you'll need at least two people of each specialty (at least network, storage, virtualization/cloud, Linux, k8s, AD, macOS, Windows) in a company that has IT needs beyond "they're shuffling papers and emails". That's 16 people, but many organizations don't have enough work to justify employing that many people.
Most end up with going the "hire a generalist or two and have them do everything", which can work for an incredibly long time but then it collapses in a dumpsterfire (the generalist calls it quits and no sane potential new hire wants that workload for that paycheck, the company gets hacked because the generalist only has rough concepts about client-side IT security, ...).
Some end up with outsourcing their IT, either to some local company or to body farms from India. That usually has better capacity to account for vacations, sick days and whatnot, but comes at the price that you don't have someone you can call for a quick firewall change but have to wait days or weeks for the ticket to make its way around, leading to frustration and eventually attrition from the non-IT staff.
And a very select few end up hiring a bunch of specialists, which produce very good quality and high amounts of internal satisfaction... but as these few companies usually do that because the CIO pushes for it, some day when the CEO wants to count beans and the CIO can't defend the effort ("preparedness paradoxon"), the department gets slashed, and they end up like the first sort.
They come from all different walks of life. That's part of what makes hiring specialists difficult. Specialists don't start out that way, they become that way, mostly due to intense focus on a particular area for a long period of time. Some people get that way quickly because they are pre-disposed with passion for a particular area, other people get that way due to their environment needing them to be there and sticking to it.
There's consistent way, that I know of, to "create" a specialist. Probably the closest is by rote memorization and drilling from detailed and intensive instruction, which is the method militaries use to staff specialty roles, but quality and capability vary greatly. I don't necessarily classify these folks as "specialists" in the same way as someone who naturally gravitates towards a subfield.
Of these:
- the first group tends to have the greater holistic view about their certification topic and know a ton about best practices, but lack detailed knowledge about dependencies, prerequisites and other experience you only gain during real world experimentation and work.
- the second group, the ones I like most, may lack best practices, but their experience makes more than up for that
- the third are one hell of an explosive mixture. Either you get a combination of the first two groups (basically, someone with tons of certificates because you're not worth anything in public service without a paper proving your worth, and a lot of experience working in constrained and otherwise troubled environments) which is the best you can get, or you get a serial bullshitter whose best experience is in kissing leadership arses (as said, all that matters in public service is the papers, not knowledge or performance).
Of course, it's also reasonable to ask if they really need that complicated infrastructure.
So in the end you are at +/- 0 resources anyway. However, what you get out of these same resource count is potentially a lot more, because the cloud gives you a longer leverage arm and a lot of the underlying tech in the cloud is more best practice than what you would have cobbled together yourself. For example IAM is a higher abstraction level, that a manager can actually understand, whereas ssh keys is black magic tech talk.