Sysadmins who manage on-prem, private clouds are an aging population. they are fed up with the constant refresh cycle, fw updates, failures, backups, testing, security exploits, change control and the countless other waist of time activities that are invisible to the business and add no real value. Bunging all that backend work into the cloud for some other spod to manage isn't such a bad idea. They can then get on with developing better services that actually add value (and not have to respond to that dreaded alert saying the SAN is down). But don't get me wrong, all eggs is a valid argument, your critical business systems should be hybrid or hosted on 2 different public clouds, just like you'd design fault domains into your on-prem infrastructure.