Many of these outages are due to pushing broken artifacts or configuration to production.
A single provider can pretty easily offer geographic or network topological redundancy, but administrative and/or technological independence is pretty hard to achieve in a single company.