Tail recursion: Even better. Erlang has Tail-call-optimization. Any tail-call reuses the stack-frame.
Lightweight processes: Large chunks of (immutable) binary data is shared among processes. This is crucial for keeping memory usage down. Also, there are ETS-tables which can be seen as a shared in-memory datastore among processes.
Messaging: Sending a message is fast. But remember you are doing a copy of the data into the other process memory space (unless you are sending a binary, see above). For small messages, this doesn't matter, but you shouldn't be tempted to send that one million node binary tree ;)
Hotpatching: The hotpatching is mostly centered around the ability to runtime-upgrade between versions of phone-exchange software. People tend to think of Common Lisp style hotpatching where you can dynamically build up your application from scratch without restarting the CL environment. Erlang doesn't really support that, but you can hack it to give you somewhat of the same.
Fast: Erlang uses a threaded-code bytecode interpreter called BEAM (by default, there is the HiPE project). Now, the rule of the thumb is that interpretation is a factor of 10 slower than compilation. The claim of a factor 3 is wildly exaggerated. Floating points are all boxed so it is easy to get a factor 30 in difference there. The persistent memory model also costs compared to C. Some data structures are impossible to get directly. A string of 100 bytes takes up 1600 bytes on a 64-bit machine.
And my own add:
Parallelism: The language semantics maps really nice onto a parallel world where we explicitly control the parallel processes. But the SMP-interpreter in the Erlang kernel is still pretty young. Note that the language was originally built for concurrency rather than parallel computation.
Conclusion: Erlang is a really cool language, but it is not built for everything. If people think they have solved the multi-core problem by using Erlang, the world of parallel computation is just dawning on them. Remember, there is no silver bullet.