Caching and/or DB hosted in a region closer to your users, or perhaps even in their browsers (if the data is not sensitive).
Naturally, if you use secondary DBs you will have to deal with data replication.
Is hitting a real RDBMS absolutely necessary? Consider producing precomputed data sets which can be cached as JSON or with other non-dynamic, storage-only methods. A distributed CDN can be much cheaper than a VPS.
Also consider if Postgres is really your best option. You could use various other database engines that can optimize certain factors: some could be cheaper, or make replication easier etc.
For example (just one example out of many, may not even be applicable to you) using CouchDB on server and PouchDB on browser would make data replication and propagation extremely easy, and it's also a very efficient and scalable query engine.