20VC：走进 Carnegie Mellon 的 400亿美元捐赠基金｜为什么90%的LP不该投VC｜多阶段基金的1400亿美元难题｜Miles Dieffenbach 揭秘 DPI、TVPI 与非流动性背后的数学

My message here to all venture capitalists is: now is the time. Please take your companies public. I breathe investing. These business models these GPs are creating are some of the best high-margin businesses ever created. My question to any new allocator or investor is: do you think you're going to have access to top-decile managers? Because at that point, with top decile, you are achieving returns above the PME consistently. But below that, even top quartile, you're not.

Miles, dude, I'm so excited for this. Listen, we've been friends for a while, and I'm so excited that we could also make it happen in person. What no one knows is I dragged you around London for a walk last night, and it poured with rain. You were so patient and great, but thank you for joining me, man.

Thank you for having me. It's a pleasure to be here. You've had some incredible guests on the podcast, and I'm honored to be one of them.

It's amazing, given the fact that I've known you for a while, and then also, in the research for this, learning more and more about you, because I didn't actually realize this, but at 26 you went through a cancer experience, and you're a cancer survivor now. It's pretty unbearable to think about, given the fact that I'm 29. It's just the most incredible strength. How did having cancer and facing your own mortality change your mindset? I've never asked that question to start a show before.

Well, let's dive into it. We'll dive into the heavy and hot. It's a surreal moment when that happens. I think everyone at that age thinks you're invincible. I did. You get that news, and you're in a bit of shock, right?

It was so abnormal to me when they told me I had lymphoma. I said, “Oh, great. What's lymphoma?” I thought it was a cold. I didn't even know what it was. They said, “It's cancer, and it's progressed quite substantially, and we need to start a chemotherapy process here within the week.”

Like most people, I sulked for about 12 hours, went home, was mad at the world, and didn't want to speak to anybody. Why me? I woke up that next morning, and one of my college football coaches had a great quote that really stuck with me: “Success in life is 10% what happens to you and 90% how you react to what happens to you.”

I took that running. That next day, I said, “I'm going to attack this. I can't change the situation I'm in, but I can change how I react to it moving forward.” I basically said, “Cancer can't kill me if I don't stop moving.” So I started a pretty insane regimen of workouts. When I would go in and get my chemotherapy, that was my R&R. That was my recovery period. I'd get out, and I'd start that again. Flash forward four months, I was cancer-free, and I have been so ever since.

Do you remember the moment you were told you were cancer-free?

Yeah. It was crazy because I got in there. You get a scan right before, and then you go into the office, and I waited two hours in the office after the scan. Usually, it's about 30 minutes. I'm sitting there thinking, “I might be biased toward negativity. It's got to be bad news if he's waiting two hours.” He came in with his arms wide open and gave me a big hug. It was pretty incredible.

Wow, that must be the most special moment.

Yeah, it's special. Looking back on it, everyone's had adversity. You've had adversity in your life. A lot of people do. Everyone does. No life is perfect. But there's beauty in the struggle, right? That makes you who you are as a person, and it builds you into a stronger person. The trials of life are many, and I wouldn't change anything.

Did it set a benchmark of shit that now makes everything else seem kind of okay?

Oh, I mean, the perspective you have moving forward after that is one of the great blessings of it, right? Life is an incredible joy and a blessing, right? There aren't many things that can take me down mentally at this point.

How on earth does one go from surviving cancer, beating the odds—amazing—to the endowment model?

There we go.

I mean, it's a pretty smooth transition for me. Give me credit. I do want to start with laying the landscape and framework for how CMU operates and is structured today. If you think about a construction that's easy for everyone to understand, what does that portfolio construction look like for CMU today from a top-down perspective?

From a top-down perspective, we manage $4 billion on behalf of the university. Starting at the highest level, we think of equity and fixed income as the 2 parts of the endowment. 85% of the endowment is equity, and 15% is fixed income. That is our allocation, and we manage to that on a quarterly basis.

One step below that are the sub-asset classes within it. Our target is for 50% of the portfolio to be in privates. That's a mixture of venture capital, private equity, real estate, natural resources, and private credit. The other 50% is hedge funds and liquids, which are public equities and fixed income. That is the top-down management of the portfolio.

Within that private bucket, we have free rein into the underlying allocations. We call it a best-athlete portfolio. How do we find the best risk-adjusted returns globally across all of those different private asset classes so we can have the best risk-adjusted return for the portfolio?

When you look at it today, how has that makeup changed over time in terms of where the private commitments lie?

Yeah.

So, from a liquidity perspective, we've been fortunate compared to most endowments, where that private equity book has been self-funding the past 3 years. Our distributions have paid for our capital calls over the past 3 years.

The sub-asset classes within that have had very different performance. Our buyout portfolio, our private equity portfolio, has contributed the most to those distributions. Venture has been the largest detractor of those, but it's been self-funding, right? So our private equity book, at that 50% number, has stayed relatively consistent for the past 6 or 7 years. As distributions from venture have slowed down dramatically over the past 3 years, venture has risen as the NAV has risen, but there have been markdowns along the way as well.

So, when you think about commitment to venture as a whole, what is the percentage commitment to venture as a whole of the endowment?

For us, venture globally is a little less than 25% of the total endowment, so almost half of that private equity book.

How does that compare to others like you?

I'd say we're overweight venture by, call it, anywhere from 5 to 10 points versus most other endowments of our size. We're underweight hedge funds and real assets, which would be real estate and natural resources. On privates as a whole, we're right on par with most endowments, plus or minus 5 points.

When you think about all of those different asset classes that you can allocate to, how do you think about opportunity cost? I think you said it before, which is a unit of return per unit of risk.

We take everything from a lens of risk first. When you think about the different private asset classes, you've got real estate, natural resources, private equity, and venture capital, which is a mixture of growth and early stage.

Take real estate, for example. You could have an industrial building that does a triple-net lease, with rents being paid by Amazon. Those rents increase 3% a year. It's a very stable asset. There's a replacement cost to that asset. It's not nearly as risky, and so the returns will compensate for that. It is not as risky of an asset, right?

Venture—picture early-stage venture—it's a $100 million fund investing into 2 or 3 people with an idea. It could be a completely new idea. It could be an idea going against big incumbents. The company is not going to be profitable when they start out. It's probably the riskiest asset class you could have. So you want to get compensated—you need to get compensated—for the risk you're taking within that asset class.

Do you think LPs are getting paid for the risk that they are taking investing in venture?

Absolutely not.

Why not?

We take a very hard look at the data that comes out of the asset class. There's really good data from about 1998 to today. You look at the median IRR for the asset class over that time period for mature funds, right? We'll look at the 10- and 15-year returns for every one of those vintages, stopping at 2016, as that's going to be the closest to a mature vintage you're going to get.

The median IRR is about 8% net for that asset class, and the top quartile is a bit higher, at 15%, but the MOIC is about 2.5x, right? The big difference is when you look at those performance numbers on a DPI basis. We'll stretch that from 10 to 15 years. Top-quartile DPI for 15-year-vintage funds from 1998 up until 2015 is 1.8x.

When we think about those underlying asset classes and our public equity portfolio, we have a public market equivalent for every private asset class we invest in, right? For real estate, it could be VNQ, Vanguard's REIT index. For our buyout portfolio, it could be a small- and mid-cap value index. For venture, it's the QQQs, the Nasdaq-100, and that's been the best-performing PME globally over the past 25 years.

When we think about “absolutely not, you're not getting paid for the risks that you're taking,” and a statement that you said to me before, which is 90% of LPs shouldn't be investing in venture, who should and who shouldn't, then?

That's the million-dollar question. I think you need to have a frank conversation with—say you're a new endowment or a new family office, and you say, “We want technology exposure.” You've got 2 options. You could do that through the public markets, or you could do that through the private markets.

My question to any new allocator or investor is: Do you think you're going to have access to top-decile managers? At that point, top decile, you are achieving returns above the PME consistently, but below that, even top quartile, you're not. That is the question, and I think most people, clearly by the data, especially as a new entrant to a mature asset class, are not going to have top-decile access.

That instantly suggests, though, that you're working on historical, lagging data, which is obviously their prior returns, not a first-time fund or smaller micro-funds that are in their first vintages. That is where we see a lot of family offices and even smaller endowment funds playing today. How do you think about that?

A strategy that a lot of people are taking is first-time funds and smaller funds, as the incredible performance of the now-multistage venture firms has allowed them to scale. We spend time in that space as well, but it is a place that is quite risky. New funds, small funds, and a hypercompetitive part of the market. There are thousands and thousands of managers: specific seed funds, angel funds, and operators.

You know what I find funny? Sorry, I want this also to be an open and free discussion.

Yeah.

I find it really funny how all LPs love $50 million to $100 million seed funds, and when you actually run the math on average seed-round sizes, that's the worst place to be. The average seed round is $4 million to $5 million. To write a check with ownership, you need $3 million to $3.5 million. If you want enough diversification, you need 30, and so you need $3 million to $3.5 million checks. Well, that's $90 million. You're not going to have that with a $50 million to $100 million fund. It's impossible.

So then you either have subscale ownership or subscale diversification, or you do what everyone does, which is that they end up writing tweener checks, like $1.5 million checks. It is fucking hard to get a $1.5 million check into a $3 million to $4 million seed round when the best in the world want it. Put $50K in, but $1.5 million? Mm-mm.

My response to that would be: Consensus seed deals—either a consensus founder or a consensus idea—are extremely hard to play in because the multistage firms have all planted a flag at seed and have essentially said, “We're going to— all these seed funds are our shrapnel. We're going to blow your model up at a much cheaper cost of capital than you, and we can deploy $5 million, $10 million checks at seed when the model traditionally was $2 million to $3 million.”

But if you're doing nonconsensus founders and nonconsensus ideas, those rounds are usually noncompetitive, and that shows up in price and ownership. So I'd say that's the question I would pose back.

Do you actually see that in your portfolios? Because I don't actually know what is nonconsensus anymore. The rounds that were in the old days, they're kind of not now. Find non-AI deals, but non-AI deals are still priced incredibly rich. Actually, when you push now, it's impossible. It's such a mature asset class; I don't think you have that luxury on price.

The true moat of early-stage venture capital is the picking skill. You look at some of the most incredible companies that have ever come out of the venture asset class: Airbnb, Uber, SpaceX, and Amazon all struggled mightily to raise their seed rounds.

To your question, is there so much capital available at seed today that that's never going to be the case moving forward? I hope and pray not as an allocator to the space. I still believe there is a moat around picking, but we'll see.

So, unpack for me. Do you think venture's an access game or a picking game? You're in some of the best brand names. Is it access or is it picking?

I think it's both.

You have to weigh it out of 100.

Oof. Weigh it out of 100. I would say if you are a multistage firm that is deploying large checks at scale, 70% access, 30% picking. If you are a small and nimble early-stage fund that is trying to break the mold, I'm going to say it's 70% picking. I'll flip it, yeah, 30% access.

So, we mentioned multistage funds getting seed. We mentioned the seed firms, which are $50 million to $100 million. Well, I don't like them. What do you like? When you see a fund come through the door, where are you like, “That's straight down the fairway for me size-wise, geo-wise”? Hit me.

For us, the sweet spot is dependent, first, on the GP's skill set and what they've done previously. For us and our commitment size—which at the low end is, call it, $10 million—we'll do an $80 million fund at the low end. At the high end, anywhere from $400 million to $1 billion, right? In that range, dependent on the skill set and the track record of the team.

It's very much dependent on the people: what they've done, what they've proven, what they want to do with this fund, and the pattern matching and diligence we can do against that.

We said about access and picking.

We spoke before this about the pillars of venture. I'd love it if you could just unpack the pillars of venture, how you think about them, and where you place more and less emphasis.

Yeah. So the five are sourcing, picking, winning, helping, and selling. Selling is going to be the newest of those five, I think, for the asset class as a muscle as a whole.

Do you think your managers have been good at selling over the past decade?

Some yes, some no. Union Square broadly—and we're not an investor there; we wish we were—but I think they've perennially been the best at selling. They've got a very strict protocol that they run through, from years 8 to 12 on those funds and with those founders, to let them know that they are going to be active sellers.

Do you think managers should distribute shares, stock? Do you think Sequoia is right that, with the evergreen fund structure, they are best placed and have asymmetric information? How do you think about that?

We like them to distribute cash versus stock. The reason being, if they distribute stock to us, there is sometimes a time lag between when we sell it and when others sell it, and so there could be a 1% to 2% pricing discrepancy versus them distributing cash. Day one is quite easy. They sell that entire book immediately, and they distribute that to all their LPs equally.

Do you think the last generation did a good enough job selling in the 2021 and 2022 vintage?

Clearly not. I think that's a pretty easy one. The one thing I'll say is that the reason it got so crazy was that the public markets were pricing growth assets for an 18-month period at—you know, the median ARR multiple for a software company was 20 times, and if you were a top-quartile grower, it was 40 times. Everyone looked at their models and thought their company was going to be worth 2 or 3 times what it was in 3 years. You had public-market comps to support your reasoning for holding stock, but that all changed very quickly.

Do you believe managers' books? We all come back with these prices in terms of the marks on our books, which is where we mark our portfolios' latest valuations. Do you think managers are accurate enough in how they price their books?

Certain ones, yes; certain ones, no.

Who's the best?

Usually the multi-stage firms—think your perennial firms like Excels or Sequoia. They're taking very aggressive discounts on basically all of their securities. Even if it's a great company that is maybe achieving an even higher price on the secondary market, they're still going to hold that at a 20% to 30% discount.

But 2021 caused us to create new muscles in regard to underwriting as a group as well. For any re-up or any new manager we diligence, we'll look at the top 10 company NAVs within that general partnership. We'll underwrite those companies ourselves, and we will, on a rough approximation, determine: Are these assets extremely overvalued, undervalued, or fairly valued?

I think my biggest worry is that we've got a generation of marked books where they're like, "Oh, it may not be the 5X fund. It might be the 2.5X fund." I'm worried that it's not even going to be that. Do you think there's a realization among LPs of, bluntly, the dire nature of some of the books?

Look at the data. A top-quartile TVPI is 2.5X. Top-quartile DPI is 1.8X.

One thing that really pisses me off, because I do some LP checks when I meet managers, is, "Listen. I don't know if we're going to do an 8X, but we'll definitely do a 6X." And I'm like, "Do you know how hard it is to do that?"

Brutal.

Anything that managers say in the early meetings with you where you're like, "Oh, no, just don't say that"?

I've had a few manager meetings where folks come right out and proactively say how easy what they're doing is, how much great access they have, and the great performance that they will have. They say that the market they play in is just like shooting fish in a barrel. That is always, to me, like, "We're going to stop this call early." It's just the kind of hubris. This is one of the most competitive asset classes in the world, and we look at everybody's returns. We see how hard it is, like you said, to achieve a 6X net fund. So that's definitely a big one.

Starting at the start of the—I'm jumping around so much, but I love this. Fuck it. We said about the five pillars. Starting at the start of that process, we've got the access element, or the sourcing element. How many managers do you actually think have proprietary sourcing where you're actually like, "Ah, I see. They see shit that no one else does"?

The premier funds on Sand Hill Road and in London, such as yourself—

Well played. Thank you.

You're welcome.

Yeah.

I think there is no systematic sourcing strategy. The partners and the brands are so strong, and they're so networked in the S-tier founder community. They're just going to be a first call for a lot of these firms.

I think if you are doing a more esoteric strategy, such as bootstrapped companies in Australia or some of these tertiary markets—in Pittsburgh, right?—you can build automated CRMs to maybe track some of those companies that are going to be off the radar of your traditional Silicon Valley firm. But I think for those more traditional firms, the brand and strength of the partners mean that there isn't much of a systematic sourcing strategy there.

The thing is, when you are such a tier-one brand name, you just become a de facto meeting in the fundraising process. Before I sign the term sheet, I'm going to go to Index, Accel, Sequoia—you name your firm. But you just want to be one of the flagposts.

I'll always remember Pat Grady saying a brilliant thing to me. He's so humble, which is why I love him so much. He said, "People think we're so successful. Harry, pretty much every software company that goes public, we've missed—that we're not in—because we do see a lot. That's on us." I thought that was, A, incredibly humble, but B, the flagship. They see everything at some point in the journey.

100%. Those partnerships have clearly stood the test of time.

When you think about proprietary access where you actually buy it, who stands out most to you?

On the sourcing side?

Yeah.

Well, I mean—

I'd say the Partovi's—Ali Partovi. The dude is in, like—

Yeah, that fund is incredible.

What the fuck?

Cursor.

What the fuck?

A few others, yeah.

Amazing for him. I'm so pleased for him, but that stands out to me. Any for you?

It's become such a crowded market. There are so many alternatives. You've got South Park Commons, you've got Ali Partovi and his network, you've got YC, you've got Techstars, and you've got a thousand seed funds.

Outside of maybe a few like Ali, I think sourcing broadly—and now I'm willing to be wrong here—but I think there's a lot of luck in sourcing. You're just hustling. You're going out, you're getting emails from friends, you're getting emails from partners, and you're taking as many meetings as you can. You're on a call with a Harry, and he's like, "Wow, Harry is fucking unbelievable. I'm going to dive into this." That's the magic of venture, right? That's how I see most of it.

I agree with you, which is why in some respects I do think it is a young person's game, because it's about pounding the pavements—

Yeah.

—being there, showing up at 7:00 a.m. That takes youth in a lot of ways. Picking is the next element. Difficult to unpack in a lot of ways. Who do you think is the best picker that you know?

I love the way Mike Maples discusses picking—the way he thinks about companies that are going against the grain of the universe and are inherently not going to be super attractive or super hot because it is against the grain and dysfunctional against the way our human minds work today.

I'll never forget when I first heard of Uber. I thought it was the stupidest idea I'd ever heard. That's how you know I'd be a bad venture capitalist. I was late in college. I'm going to get in some random person's car and they're going to drive me somewhere? Same with Airbnb. I'm going to go to some random person's house and I'm just going to sleep in their bedroom? This is the craziest idea ever, right?

Those are the people and investors—Cyan Banister, another one—whose ability to see into the future is something that not a lot of people can do. It's a superpower.

How do you unpack whether someone's a good picker? Is it just looking at track record?

I think it's looking at track record, understanding the true thought behind what they were thinking when they made that investment and when they met that founder. Then we speak to founders, and so we want to hear their side of the story as well. What was that pitch like with the broader community? They'll usually tell you, "No one would even pick up the phone for us. No one would respond to our emails." Cyan or Harry sat down, and they had a blink in their eye and saw the idea.

They believed in us before everyone else did. We really want to understand the depth and granularity of those stories.

Do you often get bad references?

Yes.

Do you?

Yes.

Wow.

The way we think about referencing, when we do a new fund, we're looking for at least 20 reference calls, right?

20 reference calls.

We'll take 5 from the GP, which are the worst references we'll get, right?

Yeah. Miles was great.

Well, yeah, exactly.

Miles was great. Yeah.

And by worst, I mean they're going to be patting Harry on the back, right?

He's also the godfather of my children.

Right. Exactly.

He's my best friend from school.

So those references we don't spend too much time on. The golden references are the off-sheet references. Thankfully, venture is such a networked community that if you spend enough time in the asset class, you're able to build those networks pretty quickly.

Do you give a shit about other people's perspectives on other GPs, like venture to venture? Does that make much—

Perspectives on strategy, not so much. We're very much trying to find interpersonal risk and partnership risk. Those are 2 things that we're really digging into. We want to know: are they a good person? Have they created a bad persona among other people? Have they wronged others in a pretty malicious way?

Then we want to understand the partnership dynamic—things that they will never tell us on a phone call. We could ask them bluntly, to their face, “Is there any risk in the partnership? Does Harry like Sally? How is the mesh?”

They'll say, “Oh, it's incredible. This is the best partnership ever. We love each other. We sit down every day. We've never disagreed on a deal.” We spend a lot of time trying to understand that partnership risk.

What is the number one reason you think partnerships break down?

Incentives. Incentives and who's working the hardest. Those are going to be the 2 every time.

Do you think we have a generation of venture firms where the partnerships are staying together for the kids?

I personally think you've seen partnerships…

Implode?

The amount of change you've seen in partnerships over the past 2 years is the most I've seen combined in my 8-year history as an LP.

How do you justify that? How do you reason that?

I think there are a lot of reasons. First, folks who had made a lot of money didn't want to deal with the crap that you're dealing with today: 3 years of no liquidity, dealing with broken cap tables, dealing with founder transitions. It's just a lot of hard, gritty work that, if you made a lot of money, why do it?

Second, if you are a newer GP, you were promised a certain amount of compensation for your role, and part of that was variable carried interest. That carry has evaporated as performance has come down. Now you're getting paid 70% less than what you thought you were, so why not start fresh? Why not start with a new book, or why not start my own firm?

Why not start your own firm? We're seeing a lot of spin-outs too. Do you love spin-outs? I think they're drastically overrated.

Yeah. We historically have not done many, if any, spin-outs. Call it your tier-one, clean spin-outs. Kevin Hartz at Astar is a new partner of ours, and he was at Founders Fund for a few years. He wasn't there that long. He was still—

And he was kind of tinkering on the side.

Yeah.

We both love Kevin. He wouldn't mind.

Yeah.

He was always a founder.

Yeah, exactly. We traditionally have not done many spin-outs.

When you think about getting a good read on that, time helps. How do you think about your willingness to write checks fast versus the need to build the relationship over time, with the knowledge that they might scale if you wait 3 funds?

It's a risk we take openly. I'd say half of the new funds that we commit to, we will not invest in right when we meet them. We'll spend a 6- to 12-month period and invest in their fund that year. For the other half, we'll take either 1 fund or 2 funds into the future. So, over 3 to 6 years, we'll build that relationship over time.

The way we think about it is, if you're an early-stage venture fund, it's going to take at least 15 years for that fund to be wrapped up—probably 18 to be fully done, with all positions liquidated. When we back a new manager, we want to back them for at least 3 funds. Call it 25 years of an illiquid relationship. It's twice the length of an average marriage in the US. I don't know what marriages are like here in Europe, but—

I think we're less. I think we're 7.

Yeah, probably less.

Yeah. It's the one thing we're more proactive than Americans in.

Congrats.

You know what's fascinating about that, given the duration you mentioned there, is also LP churn. LP churn is fricking real right now.

Yeah.

Oh my God. How should GPs think about LP churn?

First, it's good to have a relatively diversified LP base, which protects you from that. So, a mixture—but not everyone can choose their LP base, right? Sometimes it's, “Take whatever—”

Money's money at the end of the day.

Yeah, money's green. In a best-case scenario, you've got a mix of endowments, foundations, family offices, founders, maybe a couple of GP checks from venture funds—a mixture of folks who are aligned to your long-term vision.

Inherently, stuff's going to happen. Folks are going to have a liquidity crunch. A family office's family is going to say, “Fuck venture. We don't want to play in this asset class anymore.” You're going to have some things come up. Being open to that and trying to still be as good of a partner as you can is pretty important.

How much is the right amount in terms of concentration from your biggest investor?

Anything more than 30%.

30%?

Mm-hmm.

Wow.

Yeah.

I'll never forget Mickey Malcher, I think it was, telling me 10%.

Yeah. Best-case scenario, you don't have anyone at more than 10%. But if you're raising a $50 million or $100 million fund and you can secure a $10 million, $20 million, or $30 million check from someone who's aligned for the long term, that still makes sense. But best-case scenario, yeah.

Yeah, I was lucky we did 10% on the back of Mickey. Fantastic advisor. Okay, so I totally get that. We have the 10% there. In terms of stability, I was always taught that endowment funds are the blue chip for stability. Is there a rubric? How do you think about advising managers on stability among different asset classes of LPs?

Yeah, I think you're right historically, with endowments being quite long-term-oriented. The endowment model in the US today has headwinds. In particular, certain endowments are going to start getting taxed. They call it the 8% range. That's 5 endowments. That's a headwind to their model, in a sense. It's not as bad as the 20% that it was going to look like a month ago.

Would you expect them to cut positions, downsize?

It's all going to depend on what they do with their draw.

What does that mean?

An endowment is mandated every year: 5% of the endowment goes to campus to support scholarships, professors' salaries, buildings, and so on. That can range anywhere from 4% to 6%, but most endowments have stayed right at that 5% number forever.

If it turns out we're going to start getting taxed 8%, we could lever our drawdown to 4.5% versus 5%, because the real risk you run as an endowment is eating into the purchasing power of the endowment.

The way an endowment works, you have a 5% draw every year, and then inflation is, call it, 3% for higher education here in the US. To maintain the corpus—the purchasing power of that endowment—you need an 8% return.

Most endowments are targeting an 8% to 10% return over the long term, over 10 or 15 years. When you start getting closer to that number, you run into some real risks. It will depend on what they do with their draw. If they don't reduce the draw, I think venture broadly will be okay. It will still be the idiosyncratic headwind that there's just no capital coming back from venture. That's the headwind to the asset class for LPs re-upping today.

What do you advise managers in terms of closes? First closes, many closes, one close?

I think GPs should be spending the least amount of time fundraising as possible. That's not your job, and you make your money investing. But some people are not as fortunate to just do the one-and-done closes, so I think it's very much dependent on your situation.

Best-case scenario, you have a very crisp timeline. You know, we're going to do our first close here. Lining up your LPs and being sure they're committed to that process, and doing the work on the subdocs and the legal work prior to that, is really important. It's about setting clear timelines.

Do you mind if a manager's ever sold part of the management company?

Yes, absolutely. A massive red flag for us, and I would say most institutional LPs. I'm not going to speak for everybody, but—

No, it is, but it's just one of those things where I see so many first-time GPs bullied into it by one large investor, often a family office, and then really regret it over time. It's the one thing where I'm like, "No, no, no, never."

The magic of a partnership is the carried interest, and you are now giving that carried interest away to a silent partner who is not going to be, like we said, grinding and taking 100 calls a week and working 9/9/6 like you.

Yeah, baby.

How do you— And so, how do you feel when you deliver incredible returns and a silent partner is getting a decent chunk of that carried interest? It's a problem.

How do you think about the rise of multi-stage platforms, dude? You mentioned the 8% to 10% that these kinds of endowment funds, and the endowment model, rely on to keep that kind of corpus the same. Everyone says, "Oh, well, it's going to be fine," because, basically, yes, they will have worse returns being multi-stage funds—8% to 12%, say—but the LPs they have are different now, and that's good enough for them. How do you think about that?

It worries us. The funds are extremely large today, and I think it's hard to assume the same returns you had from 2010 to, call it, 2017. I think Masa and SoftBank—I would put the flag in the ground at Vision Fund 1—were when all the other venture firms saw that as the opportunity to absolutely scale their capital base.

It's wrong to assume the returns you had from those years, when most all venture funds were basically raising a $400 million Series A fund—all the premier funds—and maybe they had a $400 million growth fund attached to it. But the fund sizes stayed basically the same for a decade, and so it worries us tremendously.

Do you think they will post as good returns, then? No.

No. I'll walk you through a very simple math that other LPs can put in their back pocket, but for how we underwrite these big funds today. Simple math, but I'll walk you through it.

This is a live manager; I won't share their name, but this is a manager we underwrote a year ago. This manager was targeting a $7 billion fundraise. What we do is dollar-weighted entry ownership across their different funds. This had a $1 billion early-stage fund, a $2 billion to $3 billion growth fund, and the rest was an opportunity fund. As an LP, most LPs have to invest pro rata across those funds—equally, as a percentage of the fund across those funds.

Inherently, your smallest check is going to be to that early-stage fund. Your largest checks are going to be to the growth and opportunity funds. What we do is look at the early-stage fund. This fund, call it, had 15% entry ownership; the growth fund had about 6% to 7%, and the opportunity fund had about 2.5% to 3% ownership.

We dollar-weight that across the funds, and then we look at our check: What is the average entry ownership our check is getting within those funds? This fund was about 5% across those vehicles, dollar-weighted. The very simple math there is $7 billion divided by 5%, which is $140 billion, right?

That's the enterprise value. That is the market cap of the companies—the size of the companies they are deploying that fund into: $140 billion. For us, when we do a venture fund, our target is a 4x net. That's our goal. These funds—the early-stage funds charge 2.5 and 30, and growth funds charge 2 and 20—so you're going to need at least a 6x gross to get a 4x net on that fund.

$140 billion times 6, you're close to $800 billion of market cap needed to return a 4x net for those multi-stage funds. For reference, in 2021, the best exit year of all time, there was $850 billion-ish of market cap exit value from that year. You need an entire year of IPOs and M&A just for this one manager. Clearly, it's going to be broken up over numerous years, but that's a staggering number.

My counter to you there would be you're assessing performance today on the current outcome size, not projecting forward to what it could be in 10 years' time. In other words, now we have 9 or 10 $1 trillion companies. We didn't have any 10 years ago. The outcome sizes are so much bigger than they've ever been.

If we project forward a decade, there's a very real chance that Microsoft is worth $10 trillion, and actually, we have 50 trillion-dollar companies. If that's the case, we could see that play out.

It could. We acknowledge that we could be wrong, and SpaceX, OpenAI, and Anthropic go public at $1 trillion valuations.

What we look at—and like I said, this is backwards-looking data—but we'll give you a few data points. There have been 11 venture-backed $50 billion IPOs—11. The 2 largest venture-backed IPOs ever were Facebook in 2012 and Alibaba in 2014.

So we've gone a decade, through one of the greatest venture bubbles of all time, in 2021, and we still haven't had a bigger exit than we were getting in 2012 and 2014. My guess is that a $100 billion IPO over the next 10 years is still going to be a generational outcome. The question I throw back is: Do you think there are going to be 10, 20, $100 billion-plus IPOs? I do not think so. You look at the trillion—

Well, I think there'll be 10, 20, way more, actually, $100 billion-plus outcomes, because what I'm finding so worrying right now is, bluntly, there are so many exciting companies that I would love to be a part of, whether it's your Anthropic, whether it's your OpenAI, whether it's your SpaceX. I can't get access to them given the extension of private markets. These are all companies that would be in the $100 billion IPO price range.

Oh, Stripe, SpaceX—

For sure.

OpenAI—those are all $100 billion companies today, for sure.

But where does the rubber meet the road there? At some point, the liquidity can has to be passed to someone who goes, "Fuck it, I need it."

Yeah. And even still, 2021 is a good learning opportunity. Most, if not all—except maybe Palantir and a few others—of these very large 2021 IPOs are still down significantly from that price today. These were the greatest venture assets of that vintage.

To say that it's a guarantee that OpenAI is going to be worth $1 trillion in 5 years, there is a lot of risk involved in that. What we posit back to our team is: What is the margin of safety that great investors Warren Buffett and Benjamin Graham coined these terms—what's the margin of safety we want when investing in a fund, given what we have to believe in to achieve our desired return?

I would rather not have to believe in $800 billion of market cap IPOs and M&A transactions to get a 4x net, versus other funds where maybe we have to believe in—maybe it's a $1 billion fund, but its entry ownership is 10%, and we have to believe in $10 billion, $20 billion, $30 billion, right? Anything above that is where you get the real alpha.

It's hard for us to imagine, on these very large multi-stage funds, having that kind of alpha.

Who is the single best performer to you at scale?

Index, I think they have to be. In a market that is as bad as you hear in the news and from all the folks on the podcast, the performance they've delivered and are delivering into the future is unbelievable.

They're the largest shareholder in Figma, the largest shareholder in Dream Games, the largest shareholder in Wiz, and the second-largest shareholder in Scale AI and Revolut. It's unbelievable. I give Index all the credit in the world for not scaling. They even reduced their latest fund size after the 2021 era.

I give them a ton of credit for not aggressively scaling. They could raise as much capital as they want to, and they don't. They are the most performance-driven culture that we see, and so I give them a ton of respect for that.

Danny has been unbelievably good to me since I was very, very young—18 or 19 years old—which I think is a testament to him helping the next generation amazingly. My question to you on the back of that is: Do you think they're in for a hard time?

I'm not singling them out, but the funds that are in that $1 billion to $2 billion range, where they're sizable but they're not that sizable—when you're General Catalyst, Lightspeed, or SoftBank, your cost of capital is just, to throw out a $10 million check, "Thanks for the coffee." When you're in the Index range, you're not one or the other. You're in the middle ground. How do you assess and think about that?

I absolutely think they will continue to survive and thrive at that range. I think you have enough capital to write big checks, right? So you can participate in the abnormally large seed, Series A, and Series B rounds. You have enough capital, and it's a limited amount where you can still drive extreme parallel outcomes within the fund.

And I think the performance-driven culture and what that brand stands for—being the backer of some of the most generational companies of all time. You had Vlad on the show recently. You should ask him why he went back to Index for his new math company. He could have gone to probably a cheaper source of capital and gotten funding from Masa at SoftBank or General Catalyst, or you name it.

Because Mickey told him to fuck off.

He probably could have gone to a cheaper source of capital and raised from Masa at SoftBank or General Catalyst, or you name it.

But just to be clear, you don't actually inherently believe in that fund-size range. You actually just think Index is so good.

What do you mean?

You don't love the $1 billion to $2 billion fund sizes. You just think Index is so good that they'd make anything work.

Well, most people, right? Most funds can't raise $1 billion or $2 billion, so most are inherently going to be in the lower end. And then the ones that can have had good enough performance. Most of them scale.

Have they? Most of the big funds have not got great performance.

I think so. We've looked at all of their returns. These people deserve to raise larger funds, right? They've produced really strong performance.

So when do you say, “Enough's enough. I'm out. It's too big. It's not my game”?

First, we lean on the math. Even if you own 10% of a generational $20 billion outcome, which is still going to be generational—Figma is a generational company. It's probably going to— we'll see where it prices—call it $20 billion to $25 billion.

If you're GC, their last fundraise was $7 billion. Say you own 10% of Figma, which is a generational company: $2 billion. They're going to take 20% of that. I mean, you've returned, what, 0.2x? You need 15 Figmas. It's mind-boggling to me.

My favorite was also Wiz, which was obviously a $30 billion or $31 billion outcome—the GDP of the country.

Yeah.

And it returned a third of Insight Partners' fund.

Yeah.

And you're like, “I'd be really pissed if I was the guy that led Wiz.” And I'm like, “Oh, well, thanks for the third.”

I'm sure he's happy enough that he still did it.

Listen, I'm sure he is, but I'm just like, “Ah.”

Yeah.

So you go back to core math.

We go back to core math. What we really try to understand is this: this is more qualitative, but at some point, the alignment breaks, in our opinion, between the GP and the LP.

Let me put this clearly. I don't ever blame a GP for raising bigger funds. I love incredible business models. I study, I live, I eat, I breathe investing. These business models these GPs are creating are some of the best high-margin businesses ever created, and they're stacking funds.

Think of a firm that has raised $7 billion in this fund. They raised $5 billion in their prior fund. They raised $3 billion before that: $15 billion of capital. They're charging full fees on all of that, right? So they're making, call it, $300 million a year in fees.

For often, like, 5 or 6 partners, where 80% of that fee stream goes.

Yeah.

So really, you're trying to understand whether the magic bond has been broken between GPs and LPs?

Yeah. We really try to understand whether the magic bond has been broken between GPs and LPs, which leads us to think that the fee structures need to change to accommodate that.

Why do you think the fee structures need to change?

Because when you're investing at that size and scale—when you're a fund that big—you are inherently setting up for $100 million checks into very well-established, well-run, well-oiled companies. You are essentially acting as a long-only public-equity investor, right? You're not actively managing the company. They've got their own HR team. They're doing all their own hiring. They've got a 20-person product team and a 10-person BD team. This is a well-oiled machine.

These are what public companies would have been 10 years ago. You're charging 2 and 20 on basically passive investing, right? You're not actively managing most of those positions for most of the time.

I don't think that early-stage managers are actively managing, and I don't think they should be. I work with many, and when they actively manage, they do not make the right decisions. They push managers to do things they shouldn't do. They push them to go enterprise before they should, push them to do more products, push them to scale faster and take on more cash because they want markups. I think you want passive.

Yeah, but you still need the fees for that in a $400 million fund, right? You need a team to go out and meet all of these people.

100%.

You need an office to bring these people in. With a $400 million fund, Harry, you're not going to become a billionaire off of that, right? You're not.

No, sadly.

Sadly, yeah. If you had $15 billion in AUM, which, God bless, I hope you do someday, you're going to become a billionaire off that fund, off those funds, right? That's the difference. You need that capital as a true early-stage venture capital firm. They're utilizing it.

But I get you totally. Fundamentally, leverage is everything, and these firms can raise the money without changing the fees. So why don't they just go, “Dude, thanks for the advice. Fuck you”?

Yes. Absolutely.

So we're never going to get this fee structure changed?

Do you puke when you see 3 and 30?

Yes.

Yeah.

Yes.

And that's nuts.

Yeah. It's remarkable, right?

But you suck it up and pay it.

But this is the one—Renaissance Technologies, the best hedge fund of all time. They were so good that, at one point, I think they were charging 60% or 70% carry and 20% management fees, and they kept increasing it. They kept increasing carry and increasing management fees to incentivize their LPs to get out of the fund because they wanted all the capital for themselves.

The performance was so good that people would pay them whatever they wanted to be in that fund. What's hard for us to understand today is that, when you look at the performance data on a $7 billion fund, you don't think, “God, we will pay whatever we need to get into those funds.”

Our hope is—and the hedge fund industry went through this cycle after the global financial crisis—that there were thousands upon thousands of hedge funds, and they were all charging 2 and 20. Performance was incredible for a very long period of time. All the funds increased competition, returns came down, the global financial crisis happened, and you had a complete bottoming out of the hedge fund industry, which obviously gave rise to these incredible multi-strat hedge funds. But fee structures changed dramatically from 2005 to 2007, then from 2010 to 2013.

What will fee structures be in venture in 10 years' time?

If you're raising that early-stage fund, the core $500 million Series A fund, charge us 2.5 and 20. If you're good enough, 2.5 and 30—we're okay with that, right?

But those growth funds that are really for scaled businesses, that are mature assets, should be charging long-only public-equity fees, which are 1 and 10. If you really love your LPs, it would be 0 and 10, or budget-based, based on the team. But 10% carry—that's the number for a passive long-only investor.

Going back to the size, when we scale out of you, so to speak, is there ever a case for LPs where it's like, “You know what? You've made us so much money before. Even though we may not believe in it, we're in for loyalty”?

I think certain LPs, yes. Us, no. We are in an extreme performance-driven culture.

But if a fund does you a 6x or 7x net, which is amazing, and then they raise a big-ass fund, which most would do after that great number, do you ever go, “We've got to come back, and you should have done a 6x or 7x net for us”?

It depends. It depends on how different that fund is and how different the strategy is. It very much is dependent on the situation.

How do you think about this? I have a lot of LPs who are like, “I want to be in X brand name.” And I'm like, “That's not the best risk-adjusted return. There's an opportunity cost to your cash. I think you should be in one of these 3 names.” And they go, “No, no, you don't get it. I don't care about the performance. I just want to be in Andreessen, Sequoia or Index.” How do you respond to the brand-driven nature of LP allocations?

I get why. I think it goes back to an incentive problem in the LP industry. From me personally, I could be your janitor here at the 20VC offices, and I'm going to be the best janitor you've ever had. These are going to be the cleanest floors you've ever had. This is going to be the cleanest table you've ever had.

If my name’s gonna be on it and I’m gonna be a part of it, I am gonna put my 150%. But maybe there are certain people who are incentivized to park capital in brand names that won’t get them fired, right? No one gets fired for buying IBM, right? That’s the classic quote. That’s a problem.

For you, is it easier for you at CMU to get a check done into an X brand name versus saying, “Hey, I love Cyan. I’m gonna go out on a limb and get Long Journey in”?

We have an investment committee that we go to for approval. That’s our governance committee, and we write a detailed memo for any re-up or any new name in the portfolio, and we gotta present our merits and concerns.

But we’ve educated our governance, which is such an important part of any LP that’s wanting to get into venture: having the proper governance set up to allow you to take these very long-term bets. We educate them on the math, the risk-adjusted return of the funds, and the fee structures. And so, they’re very understanding of our strategy and how we think about the world.

Do LPs not realize that some managers are doing $30 million, $40 million, $50 million a year in fees in terms of what’s going back to them at the large multistage? And do LPs hate it?

I think certain LPs choose to just not even think about it, frankly. And so certain LPs, such as ourselves, I will never blame you, Harry, for raising $10 billion. I will never blame you for doing that, and I’ll never blame a GP for doing— I’ll never bash a GP, saying, “Oh, how dare you?” Right? It’s the market. You’ve obviously done something well enough that’s allowing you to raise that capital. It’s our choice to determine: is that the right place, risk-adjusted, for our capital?

But you’re not gonna write me $1 billion.

No.

Dude, come on.

No.

Come on.

No.

I thought we had such a good rapport.

I think not enough people see this as a game of levers. And what I mean by a game of levers is, like, you can have a smaller fund, but deploy it more quickly and actually play that lever game to just amass the fee game and the AUM game. How do you think about temporal diversification? We saw a real shift from 3-year deployment to 2-year deployment. How do you think about that?

I think it’s very important. It all stems from what that GP told you they were going to do. If they told us, “Hey, this is a 2-year fundraising cycle. We’re investing it in 2 years,” and they come back to us 2 years later, we’re okay with that. We underwrote that, right?

But if this is a 3- to 4-year investment period, and you told us it was gonna take 3 to 4 years, and you come back in 2 years, then we’ll have some questions for you, and we’ll want to work with you to understand why. What’s the reason? Because time diversification is extremely important.

What happens if they’re slower? Is that bad?

No, I don’t think so.

’Cause a lot of people say, “Oh, play the game on the field.”

Play the game on the field, right. I think certain folks would have bear-hugged their GPs today for not playing the game on the field in 2021. We’re not investors, but Mark Suster—I give him all the credit in the world. He’s been in the game for a long time. He saw 2021 as an insane period, and he strip-sold a majority of his portfolios and his funds for a very good price.

DPI in the pocket. All of his LPs are bear-hugging him for that. Brutal market. Tough market.

Isn’t it a brutal market?

Yes. You look at the data, both at European venture fundraising and U.S. fundraising. We’ll see what Q3 and Q4 look like. But in the U.S., this is gonna be the lowest year since 2017.

Same.

So it goes back a bit further, maybe to 2016, but it all goes back to liquidity.

What’s the takeaway from that? Is it the lack of liquidity? Is it the concentration of capital to a few names that have scaled and just eaten up more of those dollar allocations? What is the conclusion from that?

A lot of different reasons. I think the main reason is liquidity. From 2002 to 2004, you had more dollars raised in the public markets from IPOs than you did from 2022 to 2024, and with an asset class 10 times the size.

And just for reference, the dot-com bubble took 13 years from the peak of the dot-com bubble to get back to par on your public equity position in the Qs or the Nasdaq. That was a real downturn. It makes 2021 look like pennies. And you had more IPOs raised in the 3 years following that.

Something clearly is broken in the industry, given how bad the liquidity was over the past 3 years. And I think what really frustrated LPs is that you watch the public markets continue, especially as the factor exposure to technology has done tremendously.

I am not ever a believer that the IPO markets are closed. It’s purely a function of price, right? That is the problem. Folks paid significantly too high prices during the peak. Growth has slowed down. There’s not much of a market for a $100 million ARR SaaS company growing 15% with break-even free cash flow when you can buy Microsoft growing top line at 14%, growing earnings at 17%, with real GAAP profits, buying back 1% of the company every year, with the strongest competitive moat in the world.

People get frustrated when it’s like, “Oh, no one’s gonna give me 8 times ARR, 10 times ARR for this business.” We’re like, “Look at the alternatives—what investors could invest in that’s of a similar factor to your company.”

I think PE’s not coming to save us like everyone thinks it will. I think people also always have lower expectations of what it takes to buy good companies. Like you said there, you need to be a 20% grower and profitable, and there are so many companies where they’re bluntly at 10% and not profitable. That’s a tough spot to be in.

Well, yeah, we look at the data.

How close do you get to the underlying portfolio companies?

Very close. We ask for trending revenue, trending gross profit, and trending free cash flow for the top 10 NAVs of every fund we underwrite.

Does every LP do that? No.

No, definitely not.

So when you look at that, do you think you are able to predictably tell good managers in real time?

Yes.

Really?

Because they’ve got great assets.

Even though you have the ones like Circle, say, where for years it’s like, “Eh, eh, maybe?” Like, okay.

Yeah.

And then it turns into an absolute freaking monster.

There’s always going to be an extreme-distribution, right-tail outcome in these funds that is gonna be impossible for us to underwrite, right? And that’s the beauty of venture capital. So we acknowledge that.

Our valuation of the company, we know it’s probably gonna look quite different 3 to 5 years down the road, but we just wanna know: are these good fundamental businesses that are growing in value and that give us conviction that these are gonna be real, durable businesses one day?

How did you analyze Yale and Harvard selling their venture portfolios?

I think there are a lot of factors that go into it, one being the headwinds we talked about to the endowment model. But you look at Harvard selling $1 billion. It’s a $50 billion endowment. I mean, $1 billion—it’s not like some monumental thing for them. That’s probably just a refresh of the portfolio.

But I think there are real lessons learned. I mean, there was an article out today about Yale and CalPERS. CalPERS was the buyer of a piece of Yale’s portfolio, and I was on the phone with our CIO this morning.

Just take a step back. We had a venture capital fund that we committed to in 2012. This fund was in its tail life, 13 years old. We hadn’t looked at this fund in 3 or 4 years. There was 1 asset left in the fund. It was basically fully realized.

We have a great analytics team and an analytics system that tracks our underlying portfolio companies, but the companies have to be over $1 million in NAV for us to see them in our system. And so this company wasn’t even showing up—Circle, the company we’re talking about—in our system because it was below $1 million.

Fast-forward: the manager was holding it at a 30% discount, plus or minus, to the last-round valuation. I think their last priced round was around $5 billion, so they were holding it at a $3.5 billion-ish valuation. And you look at Circle, and it’s a $50 billion company today.

This is a 13-year fund that is essentially gonna do an extra 3 turns on the fund in its 13th year. Unbelievable. And so the article with Yale and CalPERS was that CalPERS bought a very large piece of their portfolio, part of which was General Catalyst, and Circle was the largest position in that fund.

And you saw it on TV, no? Or you saw—

No, I was reading the S-1.

Yeah.

My son woke me up one morning, and I was up early. I’m reading the S-1 just for fun, and I’m looking through the cap table and I see our GP on there, and I’m thinking, “Oh my gosh.”

And so I go and start looking through the quarterly reports.

And essentially, in a 2-month timeframe, I think it said in the article they bought $500 million, and you get a $100 million write-up from Circle alone. That's the risk of selling secondaries as a long-term venture investor: you're going to have these crazy right-tail outcomes in the fund that could come to fruition at years 8, 9, 10, 11, 12, and 13.

I said to Chuck, I'm not that smart, right? But we would've underwritten Circle if we were looking to sell that fund a year and a half ago, and we probably would've sold. Who would've guessed that Circle was going to trade at 100 times EBITDA in the public markets, and that stablecoins in a year and a half were going to be the hottest sector in crypto? You could not have predicted that. So I'd imagine Yale probably did the same. They probably underwrote that, and they're like, “There's probably not a lot of juice left to squeeze here.”

I have friends at Yale who I'm crying for, and I have friends at CalPERS who I'm crying for with happiness.

Well, I'm sure Yale will do well. It's a fantastic team with a great portfolio. But it just goes to show the risk of these fat-tail outcomes in these funds.

Dude, 10% discount, I think, was the reported number. How did you analyze that? Higher than you thought? Lower than you thought?

Much higher than I thought.

You thought it was higher? You thought, “Wow—”

No, sorry. Much lower than I thought.

I think it's a good deal for them.

No, it's a great deal.

Yeah.

That's what I meant. Yeah, sorry.

Yeah, so it's lower than—

Fantastic. Lower, yeah.

Yeah.

Fantastic deal. Not knowing the underlying GPs in that fund and not knowing the mix between buyouts, real estate, or venture, I would've guessed 20%. Yale's got incredible management in its portfolio, right? So they'll have some pricing power. I would've guessed 20%; that would've been probably the number that I would've put on the board.

Do you think we will see many more of these large institutions doing strip sales of their venture portfolios?

I'm not sure. Certain ones with real liquidity needs, I think they'll have to, right? And so that'll be a forcing function. But still, there's not a ton of secondary capital out there that's going to be able to swallow all of that NAV, right? So if every billion-dollar endowment comes out and says, “We're selling 10% of our venture book,” the pricing there—it's a supply-and-demand market, right? There are only so many buyers.

You talked about the liquidity problem and that being a driver in terms of the brutality of the fundraising market. Scale AI, Dream Games, Figma, Revolut, secondaries, Circle, CoreWeave, Hinge Health, which IPO'd, Chime. Are you just drowning in distributions now?

We are thankful to say that we're now self-funding in our venture book this year—

Woo-hoo.

Which is—

I think that's—

Give us a round of applause. I mean, it's the first time since 2021.

So that's a positive. But on the flip side, there's still a lot of liquidity that, sure, has been announced, but the Wiz deal, right? That's going to be a Q1 2026 event, right? That's got to go through FTC approval. Figma hasn't gone public yet. Dream Games, as stated in the article, they've got to get European approval for—

Do you think—

For that deal.

2026 will be a year where that liquidity really hits?

Yes. And what's exciting me is—and this is a crazy statement, right? It's not that I agree with this statement, but—

Come on, it's just me and you.

Yeah, exactly. For the longest period of time, private-market capital was cheaper than public-market capital, which is the most mind-boggling statement as a fundamental investor ever. It's hard to fathom that, right? But that was the case. That is why the best companies in the world didn't go public: because you could get a cheaper cost of capital, you didn't have to do quarterly earnings calls, and you didn't have to go through all the hoops to go public. Why would you go public?

We speak to founders. We understand why they don't want to go public. But the public markets are now pricing risk very differently than they have over the last 3 years. You look at Circle, Nebius, CoreWeave, Palantir, and Cloudflare—these are all businesses trading at extremely healthy multiples. So my message here to all venture capitalists: now is the time. Please take your companies public.

My question to you on the back of that is, Rory O'Driscoll from Scale always laughs at me and goes, “My favorite thing about Harry,” he goes, “Yes, so what about me?” And I specialize in that. So what about me?

Okay.

If we have this liquidity dropping in 2026, does that mean in 2027 you'll have a load of LPs flush with cash coming back to the venture asset class going, “Let's fund some more funds”?

Inherently, it will help, clearly. And particularly as maybe folks take back—

But is the needle moving on that, really?

I mean, it's been such a dearth of liquidity over the past 3 years that 1 year isn't going to solve the industry's problem, right? So we're going to need multiple years of really good liquidity to get back to a normal state. There's still a lot of wood to chop here. But it'll help, right? Undoubtedly, absolutely.

Do you love thematic funds like every other LP does?

We are agnostic. We do not have a mandate or a rule saying, “We're only going to do thematic funds,” or, “We're only going to do generalist funds.” We're a best-athlete portfolio. So when we find really great partners aligned with us for the long term, who we think have an incredible skill set that aligns with what they're trying to do in the fund, whether that's a generalist fund or a sector-focused fund, we'll do it.

Inherently, we've done 1 new sector-focused fund over the past 3 and a half years. So it hasn't been a huge part of our portfolio, but we are absolutely open to it.

What was the best-ever-performing fund you've been a part of?

We had a fund out of China that produced over a 20x net return to LPs.

Wow. I hope you sent them a Christmas card.

We did, yeah.

How do you think about China?

It's a very high bar for us today and a very hard place to invest. There are a couple of really big headwinds. One is the U.S. executive order mandating U.S. dollars can't go into artificial intelligence or semiconductor-related companies or defense companies there, which we completely understand and align with.

But the big problem—what's so unique about the China venture capital market that maybe a lot of founders or LPs who haven't spent time there don't know—is that in China, these GPs raise USD and RMB funds alongside each other, and these RMB funds are from local governments and municipalities. Most of the time for the past 15 years, since the China venture industry has been around, those funds were pari passu. They mostly invested in the same securities.

That isn't the case today, especially now that U.S. dollars cannot go into these AI companies, which I think, the last time I checked, 70% of these deals in the U.S. are AI companies. I mean, it's everything. And so that's a big alignment issue, right? What are we getting exposure to in that fund? That's a big worry.

Totally agree and share that. Super interesting. I'm actually more bullish on China than most people give credit for.

Yeah, we've got incredible partners there that we've had for a long period of time, that are extremely hardworking, extremely smart, and have been great partners to us. It's a hard market today, and frankly, a lot of the best Chinese founders have chosen to raise elsewhere, whether in the U.S., Singapore, or London. It's a tough place.

We mentioned the liquidity. The thing that's also weird and paradoxical to think through is that you mentioned the public-market players just having absolutely ripped. You see Meta throwing out $14.9 billion for Scale. It's like 45 to 50 days of free cash flow. It's really not very much for them. Google's buying Windsurf. Well, it's like a rounding error. They put $3.5 billion into Ray-Ban at the same time, and no one—

Yeah.

—paid any attention. My point being, we have these opposing worlds of liquidity starvation or drought, and then the glut of these public-market players who are playing with market caps that are $2 trillion. How do you think about that?

If Wiz gets approved, every other large Magnificent Seven company is going to see a green light in regard to making big, splashy acquisitions again, which is a good thing. You look at Google, Microsoft, Amazon, and Meta combined—they're doing $600 billion of operating cash flow, just cash coming off the company every single year.

I think they would much rather make very strategic acquisitions than buy back 50 basis points of the company, right? The big worry that I think those companies see today from our purview is that the AI landscape is changing so rapidly that the 12-month period it could take to go through a review and get that acquisition done means that company could be obsolete in 12 months.

Dude, you saw this with Windsurf. It changed a lot in a couple of months.

Yeah. Lots of great, hot AI companies have been very hot, and then they’re not hot. Stability AI—lots of companies have gone through these waves, and there will be many more.

You look at the Wiz deal: there’s a 10% breakup fee there, the largest breakup fee ever for an M&A transaction. Say that someone else wants to do a $30 billion acquisition of Perplexity, and Perplexity says, “We have to wait 12 months.” Their board is going to recommend a 15% breakup fee. Now, will those big companies risk a $4 billion or $5 billion breakup fee and 12 months in which this company might not be what it was 12 months ago?

I think that’s the reason these folks are acting so fast with regard to taking top talent, licensing the IP, licensing the technology, and getting these people building within our company on day 1.

I think it’s the smartest maneuver around it, but it only works when the people and the tech are the assets, not the revenue and the customers. In Wiz’s case—

Yeah.

—the revenue and the customers are the asset. Don’t get me wrong, the team and the technology are too, but—

It also helps that it’s not an AI company.

Sure. But without the revenue and customers, it’s not worth $31 billion.

No.

I do want to ask: you said to me before that OpenAI could still be a zero. When you think about that, what did you mean by that?

The way we think about it is that we spend a lot of time on unit economics. From what we see with OpenAI, unit economics are improving rapidly, which is great to see. But still, when you take CapEx into account, why has OpenAI raised 2 of the largest venture capital rounds ever in a span of 12 months? Not because they want interest income from the cash on the balance sheet. It’s because they’re burning $5 billion to $10 billion a year, right?

In my opinion, the music will stop eventually. This would be the ultimate anomaly if a bubble did not pop in AI. You look at past historic, incredible technological moments: the railroad, cars, electricity, steamboats, and the internet. Every single one of those had a bubble that popped. Every single one impacted the equity markets at that time.

Inherently, for the long term, it’s a good thing. It shows that this AI thing is real, and people are going to overinvest. I would find it extremely anomalous if there were not a bubble that popped here.

If folks agree that a bubble will pop eventually, and you do not have control of your own destiny, and you’re sitting at OpenAI with a preference stack—what’s their preference stack today? $70 billion or $80 billion? If stuff hits the fan and no one’s willing to write you a $40 billion equity check anymore because the capital markets have completely gotten smoked, what happens?

Do you think there is a chance that happens, though, honestly? When you look at SpaceX—

SpaceX is self-funding. They don’t need cash. That’s what I mean. You look at Google and Meta: when they went public, Google and Meta had 30% to 40% GAAP operating margins. These were the most profitable companies ever. They had complete control of their own destiny, right?

Whatever happened in the capital markets, it didn’t matter. They could not be killed. SpaceX cannot be killed. Starlink has reached escape velocity. That’s a very high-margin product. They do not need cash. They’re doing secondary tender offers. OpenAI needs cash.

Will OpenAI and Anthropic be independent companies in 5 years’ time?

To say, “Slam dunk, these are going to be trillion-dollar companies 5 years from now,” there is a lot that can happen within that 5-year period. We would say there’s still a good amount of risk in both of those businesses.

Masa has talked about this before, about AI’s impact on global GDP. If it hits 10% GDP productivity growth, then it’s about $10.7 trillion of the $107 trillion labor segment. Do you think AI will have that global impact on GDP within the next 10 years at that scale?

10 years gets closer to that. I thought you were going to maybe say 3 or 5, which I’d say no. I think these technological transitions take a pretty long time historically to bleed into GDP, creating industries.

OpenAI is obviously an incredible company, but they burned up all their GPUs in April because people were making emojis. They were making cartoon figures on the app. That’s not a GDP-boosting product.

Clearly, they’re making inroads, but all these things take time. I think the problem is that time is not your friend. Take the hyperscalers, for example. You look at 2024 to 2027 estimates, and it’s $1 trillion of CapEx they’re putting into the ground. Then you add on venture and industry investing. Say the run rate is $100 billion here in the US, and 80% of that is going into AI companies.

Now, not all of those are going to be CapEx-intensive. Maybe some of those will be application companies. But that’s a lot of money. To invest that sum, if this does not come true for 10 years, there will be a lot of pain.

But we spoke about Nvidia. This is why we pushed back on your thoughts on Nvidia when you said that it was too highly priced. If you believe AI, you buy Nvidia.

I do not think that it’s too highly priced for the business today. One of the benefits of our roles is that we’re generalists, so we get to invest across buyouts, hedge funds, real estate, and public equities. We get to witness some of the best investors in the world, across the world.

A man from your hometown, Chris Hohn, is one of the most incredible investors of all time. He thinks a lot about peak earnings and peak multiple, which is a common theme, obviously, in the public equity industry.

But Nvidia is a cyclical business at the end of the day. You look at their historical financials over the past 20 years. Essentially, every 3 years, they’ve had extremely negative year-over-year revenue growth. Now, it rebounds, but this is a hardware-inventory, cyclical business.

Back to my question: if folks agree that an AI bubble will pop at some point, and the largest buyers of these GPUs are advertising-driven companies—Google, Meta, and Amazon, which now has a very large advertising revenue line—and advertising is also a cyclical business, and you have a global downturn, there is a really plausible scenario in which revenue drops 20%. I think that would be conservative: 20% to 30%.

Then earnings could drop 40% if they don’t react on their OPEX quickly enough. I looked this morning, and they’re trading at 38 times forward earnings. Maybe it drops to a trough multiple of 24 times, which has been a trough multiple for Nvidia. You just blink, and you’ve had a 70% drawdown.

To think that’s not a possibility in the future, I wouldn’t say that. I’m not going to guarantee that’s going to happen in 1 year, 2 years, or 3 years, but I think it’s a possibility.

One final thing I want to touch on before we do a quick fire is founder-friendly. Everyone loves to say how founder-friendly they are and how founder-friendly their GPs are. How do you think about the founder-friendly tag in venture today?

My background comes from sports. I played football growing up and in college, and I was used to hard coaching. You don’t love it in the moment. You don’t love a coach MF-ing you, screaming at you, telling you that you’re playing terribly, and telling you that you need to do this better and that better.

But it’s better for you, and you know it’s coming from a coach who wants the best for you. They don’t want you to fail. They’re incentivized for you to do the best work possible. I love getting coached hard.

I told our CIO, Chuck Kennedy, when I first joined that he shouldn’t have hired me to begin with. In my mind, there was probably a good chance I wouldn’t make it 6 months, but I was going to try my best. I told Chuck, “I need you to criticize me. I need you to coach me hard.”

He looked at me with crazy eyes, like, “I’ve never heard anyone say this to me in my life.” But I love hard coaching.

No founder is going to be perfect. Founders are going to have weak spots. If you can have people who are, from a loving perspective, close to the business and can supplement certain weak spots and bend the trajectory of a company even a bit, why wouldn’t you? Why wouldn’t you push for that?

Those are going to be tough conversations, but tough conversations aren’t bad things. When we’re sourcing and doing reference work, that’s not something we try to dig out.

We want the most founder-friendly GPs. That's not something we source for.

I'm so glad. Thank God. I'm sure mine would not say I'm the most founder-friendly. Harry says 996. I'm so tired.

I'm so tired.

I'm so tired.

I want to go to sleep.

Dude, I want to do a quick fire with you. I'll say a short statement, and you give me your immediate thoughts. Which venture firm charges 3 and 30 and shouldn't?

Any fund that raises over $4 billion. I think that's a pretty easy answer.

There are firms that do over $4 billion—

No, no, no. Excuse me—on their growth funds, right? So the early-stage funds—

But they do 3 and 30 on growth funds?

No, no. They're charging 2 and 20. I don't think they should charge that. I think a core early-stage fund, if they have produced incredible returns over the past 15 or 20 years, deserves 3 and 30.

What's the biggest lie GPs tell LPs during fundraising?

Oh, that's a great question. I would say, “Miles, this is the perfect fund size for us. We want to be a Union Square or a Benchmark. $300 to $400 million—this is the perfect size. We're never going to raise a bigger fund.” I hear that, I kid you not, at least every other intro meeting I take with a firm.

And it's 99% bullshit?

99.9% bullshit. Yeah.

What's one red flag in a GP that others keep ignoring?

I would go back to alignment. We talked about LPs looking the other way, but alignment is—

GP commitment is one form of alignment. Sorry to interrupt you. How do you guys feel about that?

It is. Yeah. It's a very important data point for us. The nominal number is not as important to us as what that number means to that person. That's very important to us. Frankly, we have 2 quantitative data points, outside of fund size and past returns, that are the best forward-looking indicators for future returns of our funds. One of them is GP commitment. It is an important factor for us.

Who is the most underrated emerging manager today?

He'll probably raise a bigger fund, but I'll say Kevin Hartz at Astaris. I think they've done fabulously well as a partnership.

What do you think makes him so good?

Kevin, please, if you're listening to this, do not use this to raise a billion-dollar fund. What's interesting about that team is, you've got Kevin Hartz, a multiple-time founder who's taken his companies public, been through a lot, and seen a lot. You've got Gotham, who was COO and CFO of Uber. You've got Bennett, who did some incredible deals at Co2. I think it's a very heavyweight, powerful team for a right-sized fund. I don't think there are many of those funds around, frankly. Their ability to have really premier access that traditionally a multistage fund is going to have 99% of, I think, is pretty rare.

When you think about a fund investment decision that was a mistake, what did you not see that you wish you had seen?

The key thing we go back to is people—really trying to understand who the people driving the returns at that fund are moving forward.

Do you think you get accurate attribution?

We've gotten much more sophisticated with our reference work. We build our own attribution tables. That's another huge red flag and lie that we get from managers. It's not an outright lie, but they will give us attribution, and then you have one partner leave, retire, or go to another firm, and you're getting this attribution from this new person who clearly was not the partner on this home-run deal.

We understand why they do it—they have to assign somebody to it—but it could be very misleading to a new LP coming into that fund and saying, “These incredible partners who led these incredible deals are all still here.” Through reference work and longevity, we build our own partner attribution.

Would you rather back a 25-year-old first-time manager or a 55-year-old unicorn founder?

Well, if it's Harry, that makes the decision a little bit tougher. I'd say, in general, we would lean toward someone who has been through multiple cycles and has the scar tissue from that. I'd say we'd probably lean toward the 55-year-old, but we're open to everything.

What did you believe about fund investing that you've changed your mind on? For me, in investing, it was people, market, and product. I used to weigh them equally, and I've completely changed my mind around that. Markets change, products change, and this is for seed. I massively over-index on people.

When I first started, from a first-principles perspective, I was drawn to the data, which we laid out here in the beginning. You can't over-index that data too heavily, similar to what we've talked about. I would go back to, at the end of the day, this is a people-driven business.

You can do all the data work you want, which is important, clearly, as we've stated. Really lean on the qualitative reference and people work, and speak to founders. We don't take a lot of founders' time, right? They have a lot of better things to do than speak to measly LPs like us. But really understanding: Why did you choose that partner? Why did that partner choose you? What's that relationship been like? Understanding that dynamic is critical for us.

What fund are you not in that you wish you were in?

Union Square.

Easy one.

Yeah.

What's the wildest GP behavior you've seen in a fundraising process?

I've got one good one and one bad one.

Oh, go on.

Okay. What do you want first?

Start with the good.

The good. Long Journey Mentors, an incredible partnership between Lee Jacobs, Cyan Banister, and Ariel Zuckerberg. We had a celebratory dinner in San Francisco. We got toward the end of the dinner, and somehow we started talking about ping-pong. I'm a pretty good ping-pong player, and I brought up that when I was in college, I won the Pennsylvania State Ping-Pong Championship, which is true. I did.

Lee immediately said, “There's no way you're a better ping-pong player than me. I'm a really good ping-pong player.” So Cyan said, “Well, we need to settle this.” It was around 9:00 p.m., and we had just finished dinner. I said, “Yeah, I mean, I don't know how we do this.” She said, “I'll find a ping-pong bar.” Cyan got on her phone, found a ping-pong bar, and we all went to a ping-pong bar at 9:30 p.m. in San Francisco. Lee and I played ping-pong for about an hour.

Who won?

I won. Yeah.

And then you wrote the check.

Yeah, exactly.

If he beat you, the check was canceled.

Yeah, and we got a discount on management fees.

That is unbelievable. I also love the Americans. You're like, “9:00 p.m., dinner was finished.”

Yeah.

So in Europe, that's when you start drinks.

Just starting.

That's so funny.

Yeah. The bad one—this one forever stands out—was that we underwrote a manager in 2023 that was holding OpenSea at $13 billion. That one was—

And you questioned them on it?

Yeah, obviously.

And they came back?

They said, “We're going to revise our valuation policy, and we're going to revise that mark.” That was a crazy one. Yeah.

Dude, that is absolutely wild. Listen, I so appreciate having you in the studio. I so appreciate the friendship. This has been so much fun to do, so thank you so much for joining me, man.

Thanks for having me. This has been a blast. We could talk about it all day.