I'd say these two factors:

1. Most practitioners had to go to graduate school, while many entrepreneurs try to get out there asap

2. ML requires a lot of data, and that's something that big companies have more than startups