It's 100% our mistake(s). It's only AWS fault indirectly, in that AWS is complicated and requires a lot of non transferable knowledge.

As a small example, we currently pay $750 for Route53. We don't know why (it isn't traffic). It has something to do with Route53 resolvers that our "lead sre" setup before leaving. AWS support doesn't understand how it's setup, and since $750 is relatively small, we've just left it.