At some point, you have to acknowledge that you lost. It's much better to do that than to force people to work on something that everybody knows lost, or at least is worse than the competition. That's what 'follow-through' amounts to. One way that Google is still even vaguely an acceptable place to work is that it allows a fair amount of freedom about what to work on, and nobody good wants to work on something that clearly lost.

Also, it's unclear that success is measured by direct profit from that project. For example, the world with Android is clearly much more promising for Google's continued revenue growth than one without Android. Mainly, Google is the only company with the software chops to write and maintain iPhone-clone software that even vaguely works and the cash and testicles to negotiate with carriers. If Google didn't continue to bang hard on Android, all the non-iPhone smartphones would suck fairly seriously, and Apple would probably cut pretty seriously into Google's ad profits, despite Apple's ads not being as effective, purely because Apple controlled the dominant smartphone platform. Apple might also have a much stronger influence over the tablet web experience, beyond than just having written the first tablet web browser that worked.

(Disclaimer: I like iOS better than Android, having owned both.)