Ok but why would all infra would go down everywhere, at the same time? Is it all a big server? Are there any Architects left or is it all Golf Players?

It’s a distributed system with thousands of servers with automatic update distribution and config change distribution. Bad changes can propagate across the entire system though Microsoft status updates say they didn’t see any change that seemed relevant. It could also be a bug across all the uniform code that somehow triggered. For example Cisco BE servers once had a firmware bug that after 32767 hours the fans would all run at 100% speed.

Microsoft product reliability has been a running joke since the company started launching products. Why would anything change just because they pivoted to managed services instead of shipping install media?