Using externally-hosted services solves a myriad of problems for sysadmins and higher-ups, until it goes down for a long time, or data is lost, corrupted, or compromised, and there's absolutely nothing you can do about it.
Even so, "nobody ever got fired for using IBM services" still applies.
As far as highly available, enterprise services, probably the best situation is "we own it and host it, but have iron-clad 24x7 service and support contract with vendor."
Even so, I've seen senior admins roasted for things beyond their control, or for simple human error 1 in 100,000 times they did something.
If someone on my team screwed up I would go to mgmt, this happened & I'm responsible.
"Henry, I'm so tired of your I chopped the cherry tree!"
I am talking about IT departments with a 100 million dollar plus budget. Like Fortune 50 companies or major research universities.
Small businesses buy off the shelf services perforce.
The days of "hire a sysadmin and let them do whatever they want because wizard stuff" are unfortunately gone, except for in my own self-hosted projects, so there's that. I still do cloud stuff too, it's inescapable.
I think to break this cycle IT folks need to start translating IT concerns into business concerns sans the technical explanations. The business people cutting the checks do not want technobabble and don't' share IT's concerns. So its on IT to expand its vocabulary to achieve both business and IT goals. Or get increasingly ignored.
Server run by the main developers of the project It is not focused on any particular niche interest - everyone is welcome as long as you follow our code of conduct!