Also running dig on their endpoint returns a single A record which means they’re terminating all requests on a single server. This is probably not the most highly available setup.

They could be using anycast or managing the fail over on DNS.

https://en.m.wikipedia.org/wiki/Anycast