To date, Google’s knowledge graph has collected over 5 billion entities and 500 billion entity properties. This is not that much, considering that entities are not just notable people, but also books, songs, movies, companies, concepts literally anything.

What’s more important is that Google now has the technology to discover new entities on its own not just borrow them from Wikipedia. It uses AI to follow the syntax around known entities and see if they lead to any new ones. These new entities are then logged as related to the old ones. It’s an oversimplification but it’s basically how it works.