Yeah, we're not particularly good at those out of the box.
1) Horizontal scaling: In some cases it's doable using streaming replication, but it depends if you need to scale reads or writes. Or if you need distributed queries. There are quite a few forks and/or projects built on PostgreSQL that address different use cases (CitusDB, Greenplum, Postgres-XL, BDR, ...). And the features slowly trickle back.
One reason why it's like this is extensibility/flexibility - the project is unlike to hard-code one particular approach to horizontal scaling, because that would not work for the other use cases. So we need something that does not have that effect, which takes longer. It's a bit annoying, of course.
2) Storage systems: We don't really have a way to do that now - there are extensions using FDW to do that, but I'd say that's really a misuse of the FDW interface, and it has plenty of annoying limitations (backups, MVCC, ...). But it's something we're currently working on so there's hope for PG12+: https://commitfest.postgresql.org/20/1283/
That and perhaps global consistency aka "sort of but not really defeating the CAP theorem using atomic clocks and all that" in Spanner.