So if a startup wants to buy book PDFs legally to use for AI purposes, any suggestions on how to do that?

Reach the publishers or resellers (like amazon for instance)

Give them this order : "I want to buy all your books as epub"

Pay and fetch the stuff

That's all

For e-books, there will usually be a license agreement that prohibits any kind of nonstandard use.

That's why Anthropic had to scan physical books.