I don't know what their numbers are exactly, but the math often comes down to something like cost of customer acquisition being $10, and customer lifetime value

I don't know what their numbers are exactly, but the math often comes down to something like cost of customer acquisition being $10, and customer lifetime value being $80 over 10 years. As long as there are customers to be acquired, and money to spend acquiring them, you should get as many as you can as fast as possible. This is clearly long term profitable, but the first many years you do this, you will be in the red. The flip side is that you could stop doing this and become profitable overnight (at the cost of foregoing potential future profits). It's still a bet, it's impossible to say if the lifetime value holds up in the face of competition, screw ups or changes in the business environment - but that's why it's an investment.