Start with the position. Ten percent of nearly $15bn is about $1.5bn [2][1], and a book that ends the year near $15bn after gaining more than 30% began it near $11.5bn, which puts the year's gain at roughly $3.5bn [2]. Suppose, for the arithmetic, that the SpaceX mark tripled over those twelve months: the position enters the year around $500mn, adds about $1bn, and that $1bn is 8.7 points of return on an $11.5bn base, leaving the other $11bn of the portfolio to have returned about 22% [3]. Which is roughly what the index itself did, up more than 20% in the 12 months to June 30 [6]. One name at a tenth of the book is the difference between beating the market and matching it, which is a smaller claim than a 30% headline invites.
The closest thing the disclosures offer to a control is Colorado. The University of Colorado Foundation held SpaceX from 2009, and its $3.5bn portfolio returned 20.3% in the year to June, just under the index [10]. Same winner, smaller weight, no outperformance. What separated UNC was sizing rather than access, and sizing is what the institution was arguing about in 2009 or 2010, when Founders Fund came back for more money and then-chancellor Holden Thorp objected that rockets were too capital-hungry for venture, since the saying at the time was that you never invest in bending metal [4][5].
Against the cohort's record, one strong year does less than it looks. Endowments above $5bn compounded at 7.8% a year for the three years to June 2025 while the S&P 500 did 19.7%, according to NACUBO and Commonfund research, a gap of 11.9 points a year [9][5]. Compound the cohort rate and a dollar becomes $1.25; add a 30% fourth year and it is $1.63, while the index's three years plus a 20% fourth make $2.06, so the dollar is still about 43 cents behind, using the floor on both headline figures [4].
The reading that this vindicates illiquid private tech as a class is narrowed by Margaret Chen's own framing at Cambridge Associates, which credits a small number of very successful private companies [7], in years when private valuations fell after 2021 and returned cash slowly because there were fewer IPOs and buyouts to sell into [11]. The counter-thesis worth holding is appraisal lag: if a listing prices SpaceX above where the endowment carried it, the 7.8% three-year number was understated when it was printed and this year's 30% is recognition arriving late rather than skill arriving at all [12]. Both readings produce the same 30%. They imply different things about what an investment committee should do with the next venture commitment.
What would break the concentration thesis is a median above 20% at endowments that own none of the marquee names, which would make this the asset class rather than two or three companies. Chen expects the sector median to be very strong [7], and the evidence so far does not separate the two. Nor has UNC put a cost basis or a dollar contribution against the SpaceX line, so "a huge part" of 30% [1] is carrying weight a number should carry.
