I feel like I’m taking imagination-b-gone pills. The only potential uses off the top of my head that I’ve witnessed are credentials management and financial instruments arbitrage, and in both cases it was solved well by other measures.
I’m sure there are projects where it’s useful, but not “every company” useful like SQL or even a server supervisor framework. I would be happy to be proven wrong.
I've got a friend who claims that nearly everything you can do with SQL, you can do easier with a graph database. I'm not sure that's true; as long as you need only one or two tables at a time, and/or automatic validation against the database schema is vitally important for you (that's probably the big one), then SQL databases are probably still better. But as soon as relations start to matter, and the number of joins in your queries starts to grow, a graph database becomes much nicer to use.
I’m sure there are projects where it’s useful, but not “every company” useful like SQL or even a server supervisor framework. I would be happy to be proven wrong.