They might be willing to do things like crawl libgen which google possibly isn't, giving them an advantage. They might be more skilled at generating useful synthetic data which is a bit of an art and subject to taste, which other competitors might not be as good at.