The tech behind whatsapp was pretty cool IIRC .. that's how I got interested in Erlang/Beam. These people were not your average CS grad or even your average googler.
WhatsApp could have been coded in any language. The first version was a very simple app but it promised free SMS alternative with some great marketing campaign. Still amazing how they managed to grow the user base using this marketing when there were tons of apps doing the same thing.
I don't believe the backend would be as scalable (with the same amount of ease) on JVM (Akka) or straight C/C++. Erlang, Beam bring a lot of value to the table for this specific application:
The brilliance wasn't in building the core software but rather leveraging the right technology for scalable low cost messaging. WhatsApp's core messaging infrastructure is built on top of a modified version of the open source project Ejabberd. The technology heavy lifting has largely been done by Erlang and Ejabberd. In my opinion WhatsApp was brilliant to leverage those existing systems - but it wasn't the computer science skills that made them successful.
I thought they did some fine tuning to the BEAM VM to get an extreme amount of scale. I didn't know people had Ejabberd deployments at such high scale (2 million+ concurrent connections per server) in that time period. I never got into details of scaling XMPP so you might be right.