Graph DBs I believe are just useful when your data and the value you can get from your data is heavily reliant on relationships. This lets you construct a graph, and the field of CS/Math that has evolved around graph structures is implemented in a DB software that abstracts useful features in common functions etc.

Graph implies lots of nodes, which means lots of data. Only certain businesses use lots of data. Most small/medium businesses won't use it. That's why it seems like graphs are mostly in use at big tech etc.