A 66.9% IRR Wasn't Enough
A look at one institutional LP's long-term activity
This is my evaluation of what I call the Oregon Public Employees Retirement Fund1 / USV paradox. Oregon committed ~$260M across all vintages to USV which returned 6.7x TVPI on contributed capital, yet this was a tiny fraction of the $64.5B total PE book. This quantifies the “needle in a haystack” problem, even legendary GPs can’t move the needle when diluted across a massive portfolio.
The USV numbers are genuinely extraordinary. The 2004 fund alone turned $22.3 million contributed into $307.4 million distributed — a 13.82x multiple with a 66.9% net IRR, powered by early bets on Twitter, Zynga, and Tumblr. The 2012 fund, driven by Coinbase, produced a 23.78x TVPI and a 53.5% net IRR. Even USV's worst mature fund (2008 vintage) delivered a 4.35x multiple. By any measure, USV has been one of the greatest venture capital performers of all time for Oregon.
Sadly, even including USV, Oregon’s PE book underperformed a plain vanilla stock index by over 3 percentage points annually. Even generating $1 billion+ in distributions, USV represented perhaps 2% of total PE distributions. The other 98% included plenty of mediocre and outright terrible performers—funds like Novalpina Capital (-34.9% IRR), Lion Capital Fund II (-7.3% IRR), First Reserve Fund XII (-16.4% IRR), and dozens of others visible in the OPERF PE holdings data. The “2 and 20” fee structure means every one of these underperformers still extracted significant management fees, creating a systematic drag. In essence, approximately 415,000 current and future retirees, teachers, firefighters, police officers, state agency workers, and other public-sector employees across Oregon, were paying PE managers billions in fees to deliver returns worse than a free index fund.
Another portfolio-level observation is that Oregon’s PE overweight (peaking at 28.4% vs. the 20% target) forced the treasury to sell public equities to meet benefit payments and maintain liquidity. This created a perverse dynamic where Oregon was effectively selling its best-performing asset (public equities, up significantly in 2023-2024) to hold its worst-performing one (PE). Divest Oregon calculated this misallocation cost $3.7 billion since 2023. The illiquidity trap of PE—you can’t easily reduce exposure when you need to—turned a return problem into a compounding structural problem.
The math of scaling venture & the institutional LP trap
This is what makes USV so instructive — not just their returns, but their discipline. USV has deliberately maintained small fund sizes throughout its history. Their first fund in 2004 was $125 million; their most recent core fund in 2022 was $275 million. Fred Wilson has written openly about staying small: if you have a fund in the $500 million to $1 billion range, you need many large exits to return the fund even once, but your LPs expect 3x. USV's approach — small teams, high conviction, early-stage focus with check sizes of $1–20 million — is structurally designed to produce the kind of outlier returns Oregon actually received.
The problem isn’t that Oregon chose badly with USV. The problem is that institutions of Oregon’s size — managing $100 billion for 415,000 beneficiaries — face structural constraints that make concentrated venture allocation nearly impossible. A $100 billion fund that allocates even 2% to venture commits $2 billion. Deploying $2 billion into sub-$300 million funds like USV requires an unmanageable number of GP relationships. Each relationship requires due diligence, monitoring, and governance. The staffing, infrastructure, and consultant fees to manage that program eat into returns before the first dollar is invested.
So what happens? The pension gravitates toward larger, easier-to-deploy commitments — $200–500 million checks into mega funds. These are efficient to manage but structurally unlikely to outperform public markets after fees. The pension gets “private equity exposure” on paper. What it doesn’t get is venture-grade alpha.
What this means for family offices
This is where the Oregon story becomes directly relevant to how families should think about building a venture portfolio. I wrote previously that venture is truly an access class, shaped through community — not a box to check on an asset allocation spreadsheet. The Oregon case proves why.
An average family office deploying $5–20 million into venture every 2-4 years has a structural advantage over a $100 billion pension fund. Not despite its size — because of it. A smaller LP can concentrate in high-conviction high-touch GP relationships rather than managing a large portfolio. It can efficiently access sub-$300 million funds where the return data is strongest. It can co-invest alongside those GPs to increase exposure to breakout winners. And it can do all of this without the bureaucratic overhead, consultant layers, and rebalancing pressures that forced Oregon to sell public equities at the worst possible time.
The irony is that the families who feel they’re “too small” for venture are actually the right size for it. The question is not “can we commit $500 million to private equity?” — it’s “can we earn the access to the right 3–5 managers and stay disciplined over multiple fund cycles?” That’s a relationship problem, not a capital problem. And it’s a problem that scales with trust, not with check size.
Does venture scale? The returns say no. But access to venture — earned through conviction, community, and showing up over time — compounds in ways that a pension consultant’s spreadsheet will never capture.
This is an educational post about GEX Ventures investments. It is for informational purposes only and may not be relied on as legal, tax, securities or investment advice and does not constitute an offer to buy or sell interest in any products offered by us or others. Email me at mk@gex.vc or leave a comment if you’d like to exchange ideas.
https://www.oregon.gov/treasury/invested-for-oregon/Documents/Invested-for-OR-Performance-and-Holdings/2025/OPERF_Private_Equity_Portfolio_-_Quarter_3_2025.pdf