The Venture Capital Power Law

In brief

Most investing advice is built on managing downside: diversify, avoid ruin, and settle for the average return of a broad basket. Venture capital runs on the opposite instinct. The returns of early-stage startups are not spread evenly around a typical outcome; they follow a power law, in which a handful of companies grow so large that they dwarf everything else in the portfolio combined. The investor Peter Thiel put the point bluntly in Zero to One (2014): the best investment in a successful fund tends to equal or outperform the entire rest of the fund put together. The consequence is a strange discipline. A venture partner expects most bets to return little or nothing, treats that as normal rather than as failure, and evaluates each new company not by how likely it is to work but by how large it could become if it does. The general shape of this distribution is described in the entry on power laws; this entry is about what that shape does to the logic of building and running a fund.

The full treatment

The problem it answers

Early-stage investing faces a distribution that ordinary finance intuition mishandles. If startup outcomes clustered around an average, like the heights in a population, the sensible move would be to hold a broad, balanced portfolio and earn something near the mean. But they do not. A large fraction of venture-backed companies fail outright or return less than the capital put in; a smaller group returns something modest; and a very few return tens or hundreds of times the investment. Return studies from firms such as Correlation Ventures and Cambridge Associates through the 2010s repeatedly found this shape: the majority of individual investments lose money or barely break even, and a thin slice of extreme winners accounts for most of the total gains across the industry. In such a world, a portfolio built to be safe is a portfolio built to be mediocre, because the average investment is a loser and only the outliers pay.

How it works

The power law converts a familiar rule of thumb into a hard operating principle. Because a single company can return more than the entire fund, the arithmetic of the portfolio is dominated by its right tail. Thiel's formulation is that a venture investor should only back companies that have a credible path to returning the value of the whole fund on their own, because statistically that is where the returns will come from. This produces several counterintuitive moves. First, the relevant question about any startup is not "how likely is this to succeed?" but "if it succeeds, how big can it get?" A safe company with a ceiling is worse than a risky company with no ceiling. Second, losses are capped at the money invested, while gains are effectively unbounded, so the asymmetry rewards maximizing exposure to the extreme upside rather than trimming the downside. Third, a fund cannot dilute its way to safety by spreading tiny amounts across hundreds of names, because doing so guarantees that even a home run barely moves the total. The correct posture is concentration on a smaller number of companies that each could be enormous, followed by the willingness to keep funding the winners as they grow, a practice sometimes called doubling down on the outliers.

The key thinkers and the evidence

The clearest modern statement belongs to Peter Thiel (born 1967), co-founder of PayPal and Palantir and an early investor in Facebook through Founders Fund. In Zero to One, written with Blake Masters, he describes learning the lesson from his own fund's history: the single best investment ended up worth more than all the others combined, and the second best was worth more than everything except the first. He drew from this a rule against "spray and pray" diversification and in favor of deep conviction in a few companies. The journalist Sebastian Mallaby (born 1964) made the pattern the organizing idea of his history of the industry, The Power Law (2022), which traces how successive generations of investors, from Arthur Rock in the 1960s through the firms of Sand Hill Road, either understood the skewed distribution of outcomes and were carried by their few giant wins, or misunderstood it and were sunk by the many losses. The economist William Janeway, himself a longtime venture practitioner, connects the pattern in Doing Capitalism in the Innovation Economy (2012) to the deeper source of the outliers: the Schumpeterian gales of creative destruction that occasionally throw up a company capable of remaking an entire market.

Distinctions that matter

Two confusions are worth clearing. First, the venture power law is not the same as simply "most startups fail." High failure rates alone would only mean investing is risky; the power law adds the crucial second half, that the rare successes are so extreme they more than compensate. It is the combination that inverts the strategy. Second, this is a specific application of a general phenomenon, not a separate law. The mathematics, the fat tails, the scale invariance, and the unreliability of the average, belong to the study of power-law distributions at large. What is distinctive here is the portfolio prescription that follows: in a fat-tailed outcome space with capped downside and uncapped upside, you optimize for the tail. Note too that this logic bears directly on ownership: because the winners must carry the fund, investors care intensely about how much of a company they hold and how much that stake shrinks over successive financing rounds, which is the subject of the entry on dilution and equity.

Lineage

The idea sits at the meeting point of two older streams. One is the statistical study of skewed distributions, running from Vilfredo Pareto's observation of concentrated wealth in the 1890s through the twentieth-century recognition that many social and economic quantities have fat tails rather than bell-shaped ones. The other is Joseph Schumpeter's theory of creative destruction, in which capitalist growth is driven not by steady incremental improvement but by discontinuous innovations that destroy incumbents and create vast new value. The venture power law is what you get when the second idea meets the first: if the returns to innovation are radically discontinuous, then the financial returns to backing innovators must be radically skewed. The naming of the pattern as "the power law" within venture capital is recent, popularized by Thiel in 2014 and cemented by Mallaby's 2022 title, though practitioners had lived by the underlying logic for decades before it had a slogan.

The strongest case for it

The strongest argument is that the strategy fits the actual distribution of outcomes, and strategies that ignore that distribution reliably underperform. If a small number of companies genuinely capture most of the value, then a fund that hedges, diversifies broadly, and prunes its risk is systematically avoiding the only investments that could make it succeed. The power-law posture also handles uncertainty honestly: no one can predict which seedling becomes a redwood, so rather than pretend to pick winners with precision, the investor buys enough credible shots at the extreme upside that a few are likely to land, and structures the fund so that a few are enough. Empirically, the funds and eras that Mallaby documents as most successful are consistently those that concentrated on outliers and held their nerve on the winners, while the graveyard is full of portfolios that were too timid to own enough of the one company that mattered.

The strongest case against it

The pattern is real, but the doctrine built on it invites overreach, and serious critics have named the failure modes.

The first objection is survivorship bias in the storytelling. The power-law narrative is assembled after the fact from the funds that won, and it is easy to mistake a distribution that rewards a few for a strategy that reliably produces them. Sebastian Mallaby himself, and other observers of the industry, note that most venture funds do not beat public market indexes after fees; the extraordinary returns are concentrated in a small number of firms, which raises the question of whether the power law describes skill or merely luck compounded by access to the best deals. The strategy that works for a top-quartile firm with proprietary access may be ruinous for the average investor who lacks it.

The second objection concerns the danger of the doctrine as a self-justifying license. Because the model expects most bets to fail, it can excuse sloppy diligence and herd behavior: any given loss is "just the power law at work," which makes the framework unfalsifiable in the hands of a careless investor. Critics of the venture model during the funding excesses of the late 2010s and early 2020s argued that "chase the outlier" had degraded into overpaying for growth at any cost and inflating valuations, with the power law invoked to rationalize bets that were simply bad.

A third line of criticism, associated with statisticians such as Aaron Clauset, Cosma Shalizi, and Mark Newman who showed that power-law claims are routinely overfitted in other fields, warns that many outcome distributions called power laws are really log-normal or otherwise heavy-tailed without being scale invariant. If venture outcomes are merely skewed rather than truly power-law distributed, the tail may be less extreme than the doctrine assumes, and extreme concentration may carry more risk than the story admits. Finally, the model is domain-bound: it fits capital-light, winner-take-most software and network businesses far better than capital-intensive or steadily profitable ones, and applying its "go big or go home" logic where the distribution is flatter destroys value rather than creating it.

Where it stands now

Among early-stage venture investors the power law is close to an article of faith, and the vocabulary Thiel and Mallaby supplied is now standard: funds speak openly of needing "fund returners," of the futility of spraying capital thinly, and of concentrating on the companies that could be enormous. The empirical shape is not seriously disputed; venture outcomes are steeply skewed, and the average investment is a poor one. What remains contested is how far the prescription should be pushed and by whom. The consensus that has settled is roughly this: the distribution is real and the naive diversify-and-hold instinct is wrong for this asset class, but the power law is a description of where returns come from, not a substitute for judgment about which companies can actually reach the tail. Used as a discipline it clarifies; used as an excuse it corrodes.

Test yourself

Think of a portfolio of bets you control, at work or in your own choices, where the losses are capped but a rare success could be transformative. Are you managing it to avoid the small losses, or to maximize your exposure to the large win? If you found yourself instinctively trimming risk and spreading thin, ask whether the outcome you actually care about lives in the tail, and whether your strategy is even reaching for it.