I'll give you an example. An issue with a single data centre can cause multiple AWS APIs in the region to fail across services (I've seen this several times).

I'll give you an example. An issue with a single data centre can cause multiple AWS APIs in the region to fail across services (I've seen this several times). At that point, if you're only in a single region, you're stuffed. Networking may be affected and you want to spin up in another region, but RDS APIs fail so you can't copy over your backups, for some reasons AMIs won't copy between regions and the R53 control plane APIs fail too, so even if you could bring up a replacement you can't update DNS anyway... These can be planned for and mitigations put in place in advance, but that involves similar reasoning as might lead you to decide multi-cloud to be a safer option.