Even working on a 'small' set of data with 1 table of 15 million rows, the fact it was doing lots of user string pattern matching throughout the data and lots of reporting by different users, partitioning brought a 50* performance increase in queries.

So it's definitely worth learning, multi-billion row datasets aren't essential to get benefits from it. Anything where you can greatly reduce the dataset to be queried or modified when using intensive queries can benefit from it.