Power Law
The power law in venture capital is the empirical observation that returns from a venture portfolio are not normally distributed but follow a heavy-tailed distribution: a small number of investments generate almost all the returns, while the majority lose money or barely break even. The pattern is so pronounced that the single best investment in a successful venture fund typically equals or exceeds the combined return of every other investment in the same fund.1
The term is borrowed from statistics, where a power-law (or Pareto) distribution is a heavy-tailed probability distribution in which the probability of an outcome falls off polynomially as the outcome gets larger — the opposite of the bell-shaped normal distribution that describes things like human height. As Peter Thiel put it in a 2014 Forbes interview, "we don't live in a normal world, we live under a power law."4
The idea reframes how venture funds are constructed. If outcomes were normal, a portfolio of "average" startups would produce average returns; in practice, a fund that fails to invest in even one outlier almost cannot return capital to its limited partners. This makes venture capital, in Founders Fund partner Peter Thiel's framing, a search for outliers rather than a search for averages.
The data
Two widely cited datasets are the foundation of the modern power-law argument in venture.
Horsley Bridge. Horsley Bridge Partners, a limited partner in hundreds of US venture funds since 1985, shared aggregated returns from over 7,000 underlying investments with Andreessen Horowitz partner Chris Dixon for a 2015 essay. About 6% of investments, representing 4.5% of the dollars invested, generated roughly 60% of the total returns. The remaining ~94% of investments collectively produced about 40% of the returns, and a large fraction lost money outright.2
The Horsley Bridge cut also revealed a counterintuitive feature of the best funds: great funds lose money more often than merely good funds do. Top-quartile funds had more "home runs" (defined as ≥10× returns), but they also had more strikeouts than middle-of-the-pack funds — the U-shaped curve that Dixon and others have called the Babe Ruth effect, after the baseball player who led the league in home runs and strikeouts in the same seasons. The home runs in good funds returned around 20× capital invested; in great funds they returned almost 70×.2
Correlation Ventures. A separate 2014 study by Correlation Ventures looked at 21,640 US venture financings between 2004 and 2013 — a much broader, finance-level dataset rather than a curated LP portfolio. It found:3
| Outcome | Share of financings |
|---|---|
| Lost money or returned <1× capital | ~65% |
| Returned 1×–5× | ~25% |
| Returned 5×–10× | ~6% |
| Returned 10×–20× | ~2.5% |
| Returned 20×–50× | ~1% |
| Returned >50× | ~0.4% |
The Correlation data flatly contradicts the older industry heuristic — popularized by Union Square Ventures partner Fred Wilson — that a venture portfolio "loses money on a third, gets capital back on a third, and makes the bulk of its returns on a third." The actual distribution is far more skewed: the majority of financings simply do not return invested capital.3
Seth Levine, who published the Correlation data, modeled a hypothetical $100M fund with 20 investments distributed in line with these rates. Such a fund would gross approximately $206M before fees — an internal rate of return near 10%, comparable to public equities and well below the threshold institutional LPs expect from venture. Crucially, almost all of that return came from a fractional share of one or two outsized winners; missing them entirely would have produced bond-like returns despite the equity-like risk profile.3
"The biggest secret in venture capital"
The power law is the throughline of Peter Thiel's 2014 book Zero to One, which adapted his Stanford CS183 startup lectures with Blake Masters. Thiel argues that most VCs and most founders implicitly assume returns are normally distributed and therefore systematically misallocate attention.5
The biggest secret in venture capital is that the best investment in a successful fund equals or outperforms the entire rest of the fund combined.1
The power law means that differences between companies will dwarf the differences in roles inside companies. You could have 100% of the equity if you fully fund your own venture, but if it fails, you'll have 100% of nothing.1
Thiel's prescription, drawn from his experience at Founders Fund (early-stage investor in SpaceX, Palantir, and Facebook), is that a fund should make few, large, deeply researched bets in companies plausibly capable of becoming the dominant firm in a large new market — rather than spreading capital across many "safer" companies that, in expectation, will all fail to move the fund.5
Implications for fund construction
The power law shapes several practical features of how venture funds are designed and operated.
- Concentration over diversification. Beyond a certain point — typically a few dozen positions — additional diversification dilutes the impact of an outlier without meaningfully reducing risk. Many top-tier early-stage funds run portfolios of 25–35 companies rather than the 100+ that a Markowitz-style portfolio theory might suggest.
- Pro-rata rights and reserves. Because the fund's return depends disproportionately on doubling down on the few winners, GPs typically reserve 50% or more of fund capital for follow-on investments and negotiate contractual rights to maintain ownership through later rounds.
- The Babe Ruth tolerance. Funds that "swing for the fences" need partners and LPs who can tolerate losses on most positions. Funds optimized for low loss rates tend to systematically underperform, because they screen out exactly the high-variance bets where the power-law winners hide.2
- The "fund returner" test. A common partner-meeting question is whether a single investment, in its best plausible outcome, could return the entire fund. If the answer is no, the deal often fails to clear the bar — not because the company is bad, but because it cannot move a portfolio governed by the power law.
Implications for founders
The power-law framing also explains a set of behaviors that founders raising venture capital often find puzzling.
- VCs prefer ambitious markets to safe ones. A startup with a clear path to a $50M acquisition can be a fine business, but it cannot be a power-law winner; venture investors backed by power-law math will push founders toward larger, riskier markets even when smaller ones look more achievable.
- Pattern-matching on outliers, not averages. Investors are looking for the few companies that look like the next Stripe, Airbnb, or SpaceX — not the median good company.
- Why VC is the wrong capital for most businesses. Roughly 99% of US small businesses do not raise venture capital, and the power law is part of why: most companies cannot plausibly produce the 10×+ outcome venture math requires, so venture-style equity is a structural mismatch for them, not a missed opportunity.
Critiques and limits
Power-law claims in venture have faced two recurring critiques.
The first is a statistical one: several analysts have pointed out that the empirical distributions reported by Horsley Bridge and Correlation Ventures, while heavy-tailed, do not strictly fit a single-parameter power law. They are better described as log-normal or as a mixture of distributions, and the most extreme outcomes are over-represented relative to a clean power-law fit. The qualitative point — that returns are concentrated in a tiny minority of bets — survives, but the precise mathematical form is contested.
The second is a pragmatic one: Thiel's "best investment returns more than the rest combined" is a post hoc property of successful funds, not an a priori selection rule. Investors cannot reliably identify which deal in a given vintage will be the fund-returner; the power law tells you what the distribution will look like in retrospect, not which specific company at which specific check size will be at the head of it. Critics argue that overemphasizing the power law can encourage GPs to chase the appearance of moonshot deals (large markets, charismatic founders, high valuations) rather than to underwrite individual companies on their merits.
See also
- Venture Capital — the asset class the power law describes
- Carried Interest — the GP compensation mechanism whose payouts depend on hitting power-law winners
- Unicorn — the $1B+ private company class that power-law winners produce
- Pro-Rata Rights — the contractual tool used to concentrate capital in winners
- Don Valentine and Marc Andreessen — investors whose track records illustrate the pattern