The amount of money the fellowship gives is about the same as two years worth of tuition at Stanford. In what ways is it not scalable?
The 20 fellows chosen in each batch are people who are not only sharp, but have projects and visions that they are already tackling. They tend to teach themselves what they need to learn to make those things work, and have the resourcefulness and skills to make their projects at least somewhat viable.
That doesn't quite capture the whole of it, but my point is that there aren't huge numbers of people who are a good fit for the fellowship, and my experience thus far has been that the fit criteria are narrower than you might expect.
(Also, the vast majority of students don't pay full tuition anyway because of the extensive financial aid systems at these schools.)
http://talk.collegeconfidential.com/college-search-selection...
http://www.deltacostproject.org/resources/pdf/Delta-Subsidy-...
Let's look at Stanford. Wolfram Alpha claims 6532 undergrads, $37380 tuition per year. If student cost was truly double tuition, Stanford would be operating at a loss of $244.4MM per year, and (all other considerations aside) the $12.6BB endowment would be gone in 50 years. Ultimately, the reported cost is often based on some key assumptions that aren't always stated, and I'm pretty sure we as a population don't actually have much clarity on the true cost of educating a student, if it's even possible to really quantify such a thing (try measuring the monetary cost of creating a supportive environment or a caring teacher).
You're right that the CC link misses revenue sources -- but isn't that the point? Stanford seems to be using money from non-student revenue sources to subsidize a majority of per-student expenses. (That is, students are getting "paid.")
Or are you assuming that all that other revenue still comes from students (e.g. parking), or directly from an unchanging endowment? Stanford raises hundreds of millions of dollars in donations from people who aren't students every year, more than enough to cover that $244.4m shortfall directly. Direct income from students is only a small part of the overall university budget: http://facts.stanford.edu/administration/finances
Anyway, I'm only making a factual spending claim here. I make no claim about the effectiveness of the spending in educating students -- that's a separate issue.