Just contingency planning
This is no different from migrating from Oracle 8->9 or 9->10 or 10->11 (you get the idea), only start-ups can migrate without detailed recovery/contingency plans, but there is a grain of truth.
Azure, just like AWS and others was designed for "web-scale" load balanced web-services where network load-balancing, failover and geographically distribution are more important than keeping a single node up.
"Business Critical" is not the same as "Safety Critical": satisfying auditors, compliance & regulators sometimes means sticking to primitive technology because that is what they understand.
Biting the hand that feeds IT
