I think it's unlikely that Office for iPad is a billion dollar investment either. It doesn't change that it might not actually generate any direct returns for them.

Loss-making strategies can ultimately be profitable in a broader sense (if it increases the stickiness of the product on it's primary platforms, keeping away competitors, a strategic moat).

I don't think there's any evidence either way to be able to make any assumptions about the level of interest of the product, and whether or not that justifies it's existence.