Thanks for pointing this out. He mentions in this one that he didn’t have time to get into Graham’s number, but I still think TREE(3) is worth mentioning.
The theorem is interesting for other reasons though: It allows us to define graph families by forbidding certain substructures.