Again all this is so premature, but it is probably not too early to start learning a very, very important point about PhDs.
In most fields, including Physics, there are too many PhDs being produced already. Meaning there are more PhDs being produced annually than annual job openings where that PhD is really an important credential. And in fact some facing that reality end up dropping out of their PhD programs before getting the degree, so as to get started on something else sooner.
Even so, competitive PhDs programs–meaning their PhD graduates can actually have a shot at some of those jobs–are collectively tiny in comparison to the number of undergraduates majoring in those fields. So, only a very small percentage of majors are going to end up getting into a competitive PhD program to begin with, and then only a fraction of those will actually complete the program and get a job where that PhD is an important credential.
And the really critical thing to understand is that is true at Yale, Harvard, and so on. They may have a pretty good RELATIVE placement rate in competitive PhD programs (although not usually quite as high as their relative undergraduate selectivity), but in absolute terms it is still quite low, and then even many of those people will not actually make it all the way to the end.
Now in a way, that is actually not really a critique of valuing such colleges (assuming they are affordable). Because one of the nice things about them is the Plan Bs available–which for the vast majority will become a Plan A, sooner or later–are generally robust. Which means you don’t need to know in advance what the Plan B will end up being, you can follow wherever your interests and abilities actually lead as they are developed.
But that simply isn’t unique to those colleges. Like on that per capita list above, the Plan Bs for Haverford, Swarthmore, Carleton, Rochester, and so on are all robust too. The same is true for many of the universities high on the gross list–Cal, Cornell, Michigan, Illinois, Chicago (high on both lists), UCLA, Rochester (again high on both lists) . . . .
OK, so if a kid likes Chicago or Rochester or Haverford or whatever, including for the Physics Department but also because of the rest of the college, great.
Personally, when I get a little nervous for a kid is when they are picking a college based on a specific interest that realistically has a low probability of panning out, and that college would not be high on their list otherwise. But as long as you are picking a college that would be good for you even in that scenario–which again ends up the actual scenario for most–then I think it is fine to use something like PhD placement as a sort of tiebreaker.
But again, the list of such colleges does not stop with Yale and Harvard. In fact it is way, way longer than that.