Cyrus Reed: Iris, hey — you eat lunch yet, or did you just go straight into the spreadsheets again?
Iris Holm: Lunch happened. Barely. I was reading a portfolio table from a 2012 fund and I could not put it down.
Cyrus Reed: Okay, that's — wait, a portfolio table? That's what held you?
Iris Holm: Thirty-one companies. Most of them: write-off. A few at 1.1x. Then one line. A Series A check into a logistics software startup. $1.2 million in. $216 million out.
Cyrus Reed: Hold on — $216 million out of a $1.2 million check?
Iris Holm: 180x. One row on a spreadsheet. Everything else was basically noise.
Cyrus Reed: So I want to — okay, I want to sit inside that moment for a second, because I think that's actually where this whole conversation has to start. Picture the general partner, Tuesday afternoon, staring at that table. Thirty companies she's spent years on. And then one line item just... rewrites the whole story of the fund. That's not a portfolio. That's — I mean, that's a lottery ticket that paid off, except she had to be smart enough to write the check in the first place.
Iris Holm: The question is whether smart had anything to do with it.
Cyrus Reed: Right — but here's the scale of it. Marc Andreessen has a framing I always reference: roughly 15 companies out of every 4,000 that seek VC funding each year generate 95% of all economic returns. Fifteen. Out of four thousand. And 1 to 3 companies per fund are expected to account for 50 to 80% of that fund's total returns. So if you miss those one or two companies — you are just gone. The whole decade, gone.
Iris Holm: That's what the Power Law Distribution actually means in practice. Not a curve. A cliff.
Cyrus Reed: And we're trying to figure out today — is that cliff a feature someone built, or is it just the shape of the ground?
Iris Holm: That cliff — here's how I'd say it to a friend over dinner. Imagine you buy a hundred lottery tickets. You know going in: most are worthless. A few pay back your dollar. And one — maybe — pays for the whole neighborhood. That's the structure. Not a metaphor. That's literally what the numbers look like.
Cyrus Reed: The whole neighborhood. Okay. What are we actually talking about, in fund terms?
Iris Holm: Roughly 50 to 75 percent of venture-backed companies fail — or never return a dollar of cash to investors. Not write-downs on paper. Gone. Then about 30 percent return somewhere between 1 and 3x. You get your dollar back, maybe a little more. And then 5 to 15 percent deliver 10x or higher. That tiny slice is the whole game.
Cyrus Reed: And the 50x to 1000x exits — those are inside that 5 to 15 percent?
Iris Holm: A fraction of that fraction. Which is what makes this so different from classical portfolio theory — that framework assumes a bell curve. Extreme outcomes are vanishingly rare. Flukes. Here they're structurally expected.
Cyrus Reed: Wait — not just common. *Expected.*
Iris Holm: Built in. And there's now academic language for why. Henry Lahr — 2022 study, International Review of Financial Analysis — looked at private equity fund return distributions and found the data fits what he calls a smooth double Pareto distribution. Not a normal curve. Fat tails. And venture capital showed the most pronounced fat tails of any private equity category.
Cyrus Reed: Okay, fat tails I've heard. But then what does — wait, what does Lahr say about variance specifically?
Iris Holm: Potentially infinite variance over a fund's lifetime.
Cyrus Reed: Say that again — slowly.
Iris Holm: Infinite variance. Meaning — mathematically — the upper bound on returns doesn't settle. It's not just unknown. It's theoretically unbounded. The distribution doesn't converge the way stock returns do. That 180x line I was staring at? That's not an outlier in the freakish sense. That's the distribution doing exactly what the distribution does.
Cyrus Reed: So the 2012 fund wasn't lucky. It was just the one that caught the ticket. The structure *requires* someone to catch it — but it never tells you who.
Iris Holm: Which is exactly why the portfolio-of-many-bets isn't guesswork. It's the only mathematically coherent response to that structure. If you know the distribution requires a rare extreme outcome — and you can't predict which company delivers it — you maximize the number of shots.
Cyrus Reed: Wait, so funding twenty, thirty, fifty companies isn't — it's not spreading thin. It's actually the *rational* move.
Iris Holm: Cambridge Associates makes this concrete. Top-decile VC funds consistently hit net returns of 3x or higher. Most funds miss that benchmark. And when you dig into why — it almost always comes down to whether a fund captured one or more power law outliers.
Cyrus Reed: And missing those one or two companies — that's the whole difference? Between top-decile and just... average?
Iris Holm: The difference between top-decile and mediocre. Literally the same fund, same vintage, same market — but one caught a 100x and one didn't.
Cyrus Reed: Okay, but — so Fred Wilson actually said this out loud, right? Like, he didn't just do it. He published it. 2009 blog post, Union Square Ventures, and he just... put the power law paradigm on the table publicly. That's a weird thing to do. You're basically telling everyone your secret.
Iris Holm: It became foundational because of that post. Before that, practitioners were operating on this intuition. Wilson named it. After that, it's a mental model with a name — and a generation of VCs started building to it explicitly.
Cyrus Reed: And then Andreessen Horowitz — they're the ones who operationalized it most deliberately, right? Like, wide early bets to surface winners, then concentrated follow-on capital to double down on the ones that show traction.
Iris Holm: Broad net first. Then all-in on whatever swims to the surface. That follow-on concentration is the mechanism — small initial checks to cover 20, 30, 50 companies, then deploy serious capital behind the ones signaling outlier potential.
Cyrus Reed: But — wait, that's the part that keeps catching me. The math tells you outliers exist. It does not tell you which one. So how does Andreessen Horowitz — or anyone — know which company to concentrate into?
Iris Holm: Frankly, that's the tension the whole strategy is built on. And fund size, BIP Capital's explicit dissent, whether the durability claim even survives scrutiny — that's what we need to actually stress-test next.
Cyrus Reed: No way — BIP Capital actively rejects this? The whole framework?
Iris Holm: Explicitly. Which is either contrarian noise or the most important data point we haven't interrogated yet. The many-shots strategy is rational — but rational-given-assumptions. The question is whether those assumptions hold.
Cyrus Reed: Okay but — wait, how does BIP Capital even function, then? Like, if you're a VC and you're saying outliers won't drive returns, what are you actually building?
Iris Holm: A tranche-based model. You segment capital deployment by stage and expected return profile — instead of swinging for 1000x, you're constructing for more predictable multiples across tranches.
Cyrus Reed: That's — huh. That's basically saying the power law is wrong, or at least not useful to them.
Iris Holm: Or it's saying the power law is real but not actionable. Which is a sharper critique, actually. And it connects to the fund size problem — because 'outlier' isn't one fixed number. A fund under $50 million can be materially moved by a $30 million exit. That's a 60x on a $500K check — transformative. But a fund sitting between $100 and $300 million? That same $30 million exit is a rounding error. They need at least one $500 million-plus outcome just to hit return targets.
Cyrus Reed: Wait — so the minimum required outlier literally scales with how much you raised.
Iris Holm: The threshold moves. Which means when a mid-size fund says they're running a power law strategy, they're committing to needing an outcome that — statistically — happens maybe once every several funds. That's not a strategy. That's a prayer with a spreadsheet attached.
Cyrus Reed: Okay, but — no, here's what I keep tripping on. If everyone knows the math doesn't tell you which company is the outlier, why are VCs spending millions on due diligence? Like, founder evaluation, market sizing, all of it. That's not cheap.
Iris Holm: That's the contradiction. If the power law is dominant — if the distribution is doing the work — you'd structure funds to be large, broad, cheap to manage. Instead you see intense selection processes. Which implies they believe they have edge in picking.
Cyrus Reed: But wait — maybe the due diligence isn't about finding the 1000x. Maybe it's about filtering out the total disasters. Like, avoiding the write-offs, not chasing the wins.
Iris Holm: That's a different claim than what most VCs say they're doing. Go read how Andreessen Horowitz describes their process — it's not 'we screen for survivability.' It's 'we find founders who can build generational companies.' That's a selection-for-upside argument, not downside filtering.
Cyrus Reed: So the power law is doing — wait, actually, is it doing two jobs at once? Like, prospectively it justifies a broad portfolio. And then retrospectively, when one company wins, it explains why that was always the plan?
Iris Holm: That's the rationalization problem. The math says outliers exist and are decisive — Lahr's fat tails, the Cambridge Associates benchmark, all of it. What the math does not say is whether those outliers can be identified in advance. VCs treat those as the same claim. They're not.
Cyrus Reed: So the durability claim — that the power law persists across market cycles and geographies — that's descriptive. Not causal. It's just... what we saw happen.
Iris Holm: Described, not derived. Lahr's 2022 study fits the distribution — it does not prove the mechanism that generates it is structural rather than situational. And BIP Capital's dissent is worth keeping on the table precisely because of that gap. Maybe the power law is the shape of a specific era, not an eternal law. The stress test hasn't happened yet.
Cyrus Reed: But that spreadsheet keeps nagging at me — thirty-one companies. The one row. Did she know? When she wrote that $1.2 million check into a logistics software startup, was there something she saw — or did she just write thirty-one checks and wait for the math to do what the math does?
Iris Holm: The power law doesn't answer that. It just tells you, after the fact, which row mattered.
Cyrus Reed: And that's — wait, that's actually the whole tension, right? Like, 1 to 3 companies generating 50 to 80 percent of a fund's returns — that number is real whether or not anyone believed it going in. Whether or not anyone planned for it. The distribution doesn't care about the thesis.
Iris Holm: Maybe that's why it holds. Not because VCs are good at finding the outliers. Because someone always gets lucky in the specific, statistically necessary way. And then that fund writes the book on how they did it.
Cyrus Reed: And we read the book.
Iris Holm: And we build the next fund to match the book.
Cyrus Reed: That's — yeah. That's actually where I land. Not that the power law is wrong. Just that it might be a record of what happened, dressed up as a reason why. The 180x row existed. One row, out of thirty-one. That much is just true.
Iris Holm: That much is just true. Good conversation.
