Key Takeaways
- • Scott Wilson, CIO of Washington University's endowment, told Harry Stebbings on the 20VC podcast that direct co-investments, not fund commitments, are what move WashU toward the top of its peer group.
- • WashU turned over more than 75% of an $8 billion portfolio in its first 18 months, liquidating all 36–37 hedge funds, and the endowment now manages just under $20 billion.
- • SpaceX reached more than 20% of WashU's portfolio at cost at one point, per Wilson, with the team using several continuation vehicles to hold through the IPO.
- • Wilson's stated biggest mistake was failing to add capital to winners, which he says has "cost us billions of dollars."
- • Wilson expects an AI shakeout, doubts model providers have durable moats, and says WashU avoids funds above $10 billion.
Summary
Scott Wilson, Chief Investment Officer of Washington University's endowment, argued in a roughly 57-minute conversation with Harry Stebbings on 20VC that private markets are structurally less efficient than public ones and that direct co-investments, rather than fund commitments, separate top-quartile endowments from average ones. According to Wilson, if WashU had held no directs or co-investments over the last decade it would have landed "right at the average of our peer group." He says the endowment has generated roughly $17 billion of profits on its original $8 billion while paying about $5 billion to the university.
Wilson described a model in which roughly half the private book sits in directs and co-investments, with a stated target of 80% if unconstrained. The team has no separate direct-investing group, and 80–90% of its time goes to meeting individual companies. He said the team invests in roughly 0.1% of what it sees. Wilson acknowledged a capacity ceiling: a 1% position at current size is close to $200 million, so the team often starts with a $50 million check. Wilson said the single-name governance requires notifying the board at 2–3% of the portfolio at cost and board approval at 4%.
On SpaceX, Wilson credited Lee Fixel, the board, and the team for a position that at one point exceeded 20% of the portfolio. The team argued to hold through multiple liquidity opportunities. For post-IPO sales, Wilson described a gradual, multi-year approach to taking chips off the table. On Cerebras, he said the valuation is hard to justify on the last 12 months' KPIs even at $25–30 billion, so holders are "betting on the come."
Wilson named failing to add to winners as his largest error across the $11 billion of value created, attributing it to frame-of-reference bias. He also cited a mid-2020 secondary package that the team declined to sell, expecting higher marks; it was offered at roughly 80–90 cents on the dollar before venture markets re-priced. Fund selection is limited to roughly 20–30 funds, with a bias against mega-funds above $10 billion, though not an absolute rule.
Wilson expects an AI shakeout, estimating roughly half of today's AI-related companies will be gone in ten years, and said the top five large language models could be hard to distinguish on 99% of tasks. He also said he is "not convinced that the public market's going to give" Anthropic a valuation north of $2 trillion.
Wilson described a model in which roughly half the private book sits in directs and co-investments, with a stated target of 80% if unconstrained. The team has no separate direct-investing group, and 80–90% of its time goes to meeting individual companies. He said the team invests in roughly 0.1% of what it sees. Wilson acknowledged a capacity ceiling: a 1% position at current size is close to $200 million, so the team often starts with a $50 million check. Wilson said the single-name governance requires notifying the board at 2–3% of the portfolio at cost and board approval at 4%.
On SpaceX, Wilson credited Lee Fixel, the board, and the team for a position that at one point exceeded 20% of the portfolio. The team argued to hold through multiple liquidity opportunities. For post-IPO sales, Wilson described a gradual, multi-year approach to taking chips off the table. On Cerebras, he said the valuation is hard to justify on the last 12 months' KPIs even at $25–30 billion, so holders are "betting on the come."
Wilson named failing to add to winners as his largest error across the $11 billion of value created, attributing it to frame-of-reference bias. He also cited a mid-2020 secondary package that the team declined to sell, expecting higher marks; it was offered at roughly 80–90 cents on the dollar before venture markets re-priced. Fund selection is limited to roughly 20–30 funds, with a bias against mega-funds above $10 billion, though not an absolute rule.
Wilson expects an AI shakeout, estimating roughly half of today's AI-related companies will be gone in ten years, and said the top five large language models could be hard to distinguish on 99% of tasks. He also said he is "not convinced that the public market's going to give" Anthropic a valuation north of $2 trillion.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.