Recently at TerminusDB, at the behest of an active community member, we decided to do an ingest of the OpenAlex Authors collection. This is a pretty big data set. We found that after the ingest, not only did we have a database with 17 billion triples, but in comparison, our database is smaller than others (only 212GB as compared to 280GB), even though much better indexed. It also has the most compact triple store representation we are aware of, coming in at less than 14 bytes per triple for the tested dataset.

You can search starting from subject, object, or predicate, in any direction, and get results quickly with an extremely low memory footprint, due to the utility of succinct data structures.