A dangerous analogy, because the sole criterion of success for a startup is to be financially successful, whereas there are lots of research ideas that cannot be translated into a profitable business, yet are still worthwhile to pursue. Think fundamental results in maths and physics.
I am all for universities to have sources of revenue other than taxpayer money. But I fear that making those the only sources of revenue (you did defend that in your original post) will instill a climate where researchers are prevented from pursuing research topics that cannot possibly bring in any money for their universities.
And this is not just about "knowledge for knowledge's sake", either. Think about, say, quantum mechanics or complex analysis. Would anyone have paid for the early research in these areas with the expectation of financial returns? And yet, here they are today at the basis of a number of highly profitable industries.
All I want is better odds they'll stay, based on the assumption that there is more supply of worthy foreign students who'd stay if they could, than demand for them. There's no good reason we shouldn't maximize the benefit to the US taxpayer, by cherry-picking the supply and paving the way to their citizenship in advance. That the PhD students do real science (i.e. a job) isn't a good enough reason to leave taxpayer money on the table.
If some foreign student is so much a wunderkind that they'll do a $million in work for their $250K outlay, but plans to leave the US, that could be an exception.