Developer goodwill? Crush OpenAI? Choke Google?
More seriously Meta is experimenting with AI and the cost of $5m you've come up with is as good as R&D, its set to depreciate in value rapidly as the iteration on the models improve, and its just a step along the way. There is no way to monetise it realistically with the business model Meta currently has which keeps developers in their own ecosystem.