I run several services on a VM, with no automation to restore them if there's a hardware issue with the VM host. (There is some resiliency: a managed datacentre, the servers have RAID, dual PSUs, dual network, etc.)

For over 10 years, this has not been a problem. We understand the risk, and know how to restore services onto a different VM host if necessary.

I think the more complex system (e.g. K8S) has problems when people don't fully understand it. We've had outages there, caused by misconfigurations or unexpected behaviour.