I think you're missing a huge point there in terms of time to market and stuff an organisation wants to deal with. No matter how good I get at server ops, Amazon will always be way better than me. They probably have patents on how to cool data centres. A small company, like mine, has to choose its battles if it's to launch products quickly, and competing in server maintenance isn't what will make those products better.

And sure, I can self-host something and it's going to be cheaper if you just measure the cost of hardware. but count in the cost of humans maintaining that, patching the OS, dealing with security, working the network etc. You might be particularly good at that (and if you're running all your stuff you probably are), but I'm not. And I don't want to be. I want to focus on building stuff that I care about. So paying a bit more for server hosting works out a lot cheaper than managing all that myself, as I can focus on other things.

On the whole serverless thing - yes, it's a buzzword, but people tend to look at it from a technical perspective which is all wrong. the financial side is what matters. the tech isn't really that exciting there -- sure, there's a lot of new tech like containers, but the major change is that I'm only paying for stuff when it does something useful, instead of paying for servers to sit idle. But that's a topic for a completely different discussion. I published some of my thoughts on it here: https://gojko.net/2016/08/27/serverless.html