What is the ultimate locus of value, or 'end in itself'? I see two tempting answers:
(1) Each particular individual.
(2) The world as a whole.
Utilitarianism is, I think, most naturally developed as a type-2 view. We commonly assume that what's good for persons is good for the world (though for some possible exceptions see my post on 'Welfare and Contributory Value'). But Parfit's Non-Identity Problem really brings out the difference, as does the 'value receptacle' objection (advanced most coherently by G.A. Cohen). Utilitarians -- or 'world consequentialists' more broadly -- ultimately act for the sake of making the world a better place, and this is different from ultimately acting for the sake of Tom, Dick and Harry (or any other particular individuals).
Of course, world-consequentialists are in favour of making life better for individuals, just like utilitarians are in favour of helping your friends. It's pro tanto good. But just as, from a utilitarian perspective, helping your friends is ultimately desirable because your friends are people (and we want to make all people better-off), similarly making all people better-off is good because it makes the world better -- and that's what ultimately matters.
Some will consider this a distastefully 'instrumentalist' moral stance. But in light of my above analogy, I don't think this objection should particularly bother utilitarians. Once you accept impartialism, is the step to impersonalism really such a leap?
[P.S. It's worth emphasizing that utilitarianism doesn't recommend we adopt this perspective in our everyday lives!]
Update: I should note that holistic views (e.g. average utilitarianism) can only be stated in impersonal (type-2) terms. But 'total utilitarianism' could potentially -- if awkwardly -- be cast as a person-affecting (type-1) theory, if we were willing to claim that we harm a merely possible person when we fail to bring them into existence.