What he's saying is that perhaps Dropbox can buy a user for $1 (via adsense, whatever) and make $2 off them. In that case, it makes sense to take all the money you can and plow it into customer acquisition.
At some point the law of big numbers becomes a limiting factor, but probably not at any point that would make it unwise to take $7m.
Not to mention that there is a lot more that can be done in that space from an engineering perspective. Files syncing could be just the beginning. What about consumer MVC from files to web services? Edit a photo on iphoto and it automatically changes on every service that uses it.
The "data keeper" role is like an octopus, big in the middle, tentacles everywhere. I could imagine a world where that role becomes as important if not more than the OS / services.
At some point the law of big numbers becomes a limiting factor, but probably not at any point that would make it unwise to take $7m.