Venture is Dead (Again)
The piece argues that venture firms like Benchmark and USV raising far larger funds per partner breaks the core logic of venture investing, which depends on many small bets and close involvement in each company. It backs this with the 1980s precedent, when pension money flooding in pushed up prices and drove median returns down from over 30% to about 8%.
- Benchmark raised $2 billion in June, including a $1.25 billion growth fund, with the same five partners managing about $400 million each, up from roughly $120 million each in 2024.
- Venture works on a power law with radical uncertainty, so it needs a large portfolio of small bets and staged funding. Partners can only actively lead about 10 to 12 companies, so bigger per-partner checks force either less involvement or bigger, lower-return bets.
- The 1979 ERISA rule change let pension money into venture, and the 1980s boom showed the pattern: capital per professional tripled, valuations rose, and median IRRs collapsed to about 8% by 1988.
Benchmark, one of the most successful early-stage firms ever, kept its fund size flat at $425 million in 2024, telling LPs that a fixed size would "force discipline." This June it raised $2 billion, including a first-ever $1.25 billion growth fund, with the same five partners each managing roughly $400 million. Union Square Ventures went the same way: its core funds had usually sat between $125 and $200 million, then in October it raised $900 million across a core fund and an opportunity fund. Fred Wilson had written a decade earlier that he could never sleep if USV managed a billion dollars. The author's point is that these firms, famous for seed and early-stage bets, are drifting toward the late-stage business they were never built for. Concentration is accelerating across the industry too. The top 10 VC funds took a third of the money in 2025, up from 13% in 2021, and funds over $1 billion are pulling 72% of capital raised so far in 2026, versus 34% in 2025 and 16% a decade ago.
The argument rests on the economics of uncertainty. Frank Knight distinguished risk, where odds are known, from uncertainty, where they cannot be calculated, and venture capital exists to absorb the second kind. Its method is a portfolio of many small bets, with money staged into the companies that prove out, so that a single big winner covers the losers. Returns follow a power law, and the biggest winners often create markets that did not exist, as Apple did when Arthur Rock's early stake returned 240x in three years. The author notes that a partner can only actively lead perhaps 10 to 12 companies, sitting on boards and guiding founders through failed experiments and thin evidence. Growing assets per partner therefore forces one of two bad choices: spread across more companies while free-riding on other investors' work, which costs the firm deal flow and founder trust, or write bigger checks into each company, which either means paying higher valuations and lower returns or shifting toward later stages where the uncertainty has already been reduced. Either way, the model drifts away from the thing that made it work.
What makes this worth reading is that the author is not dismissing large funds as greedy. They concede that Benchmark and USV earned the larger checks, and that the incentives of VC management fees, which scale with assets under management, push in exactly this direction. The open question the piece leaves is whether the industry can keep calling itself venture capital once its largest players are mostly doing growth-stage deals with early-stage brands. For anyone following startup financing, it is a useful lens on why later-stage concentration and early-stage shortfalls might be connected.