Thinking Machines 联合创始人以35亿美元加入 Meta，Industry Ventures获6.65亿美元收购

In the face of unprecedented wealth, I'm shocked to discover that most people behave badly. At some level, it's not a surprise.

Guest

Something's broken in the way that we're evolving as humans if everything ultimately reduces to what's in it for me.

Would you prefer $2 billion in Thinking Machines unlisted stock, with a chance to be amazing or a chance to go bust, or $3.5 billion of liquid Facebook stock over the next 5 years?

What I'm hearing Rory say is essentially to reduce it all to the big VCs. They're playing the momentum game. If shit happens, shit happens. They can handle it.

Guest

This is life. Everything in life you can price is an option.

1. Pre-Show Chat: Rory Is So Old He Worked with Arthur Rock!!!

Rory, what was it like investing with Arthur Rock?

Guest

I actually invested with Arthur Rock, and I'm totally willing to talk about it.

That shut you up, you little punk.

Guest

Right. Let me tell you what it was like investing with Arthur Rock. He was the scariest dude I ever saw. He was old at that time and just as grumpy as when he was young, and when he said things, you just trembled in fear.

So it was pretty awesome investing with Arthur Rock, actually. You just heard a guy speak who was literally the first VC on the West Coast, who did Intel, and he was pretty direct. It was great, is the answer.

Well, there we go.

Guest

And you thought it was just a rhetorical fucking question.

I did. I did. I mean, Arthur was in the 1960s and 1970s.

Guest

He was still doing deals in 2004. Absolutely. Arthur Rock wrote checks in 2004 and 2005. I was in a board meeting with him—only 1 or 2 of them—but he was pretty damn impressive.

Yeah, Jason.

As I say, you just were scared because no one's going to argue with him.

Yeah. No, the sentence “Mr. Rock, I think you're wrong when you said that” just did not come out of my mouth. I'm pretty punky, and I was even punkier then. But, nope, I'm going with this: this is received wisdom down from the mountain. This is cars and tablets. I'm taking it as read.

By the way, we lost money on the—

No, we didn't, actually. We made money. We made money on that deal.

Yeah, we did. It took a long time. Rory, when did you start in venture?

Guest

1993.

I was born in 1996.

Guest

Oh, that killed the conversation. Just fucking dead.

You know, Harry, you think it's funny, but when you come out here and go to YC Demo Day, you're going to feel old now.

I already feel old. I was in Stockholm for a project.

Guest

You're not young. You're not. When we met you—when the 3 of us met you—you were young. You're not young anymore.

I hope we're just taping this. It's great content, but my partner Andy has this great line. He says, “The real problem with this industry is that there's a huge period of time where everyone says you're a little too young, and then there's this brief shining moment where you're good, and then there is the ‘When is he going to retire?’ moment,” right?

It seems to me it's 10 years each, with 3 years in the middle. And you know what the fuck—

What about the moment when someone comes out of it and decides to do another institutional fund after being wildly successful? That's the craziest one of all, isn't it? Why would anyone do that? Why would anyone cash out at the top of the game, go out and manage their own capital, and then have to deal with LPs? What a headache, man. That's like a sucker bet, isn't it, dude?

My favorite thing with that is LPs. For the first 3 funds, it's “No DPI, no DPI, no DPI,” and then you return a shitload of DPI. Then LPs go, “But are they really hungry anymore?” Because they've made a lot of money.

You're like, “What do you want from me?”

True. That is so true. You can't win, Harry. You can't win.

It's so true. I had LPs, Roger. I don't know if you noticed. They were like, “Oh, but is Roger still hungry?” I'm like, “Have you met Roger?” I just dialed this shit in, man. It's all for the ego.

Guys, before we dive in, there's 1 bit in the last show with Roger I did, and he said, “I plant both feet on the ground every morning and I say, ‘Let's fucking go.’”

Guest 2

And I just love that. It's so good.

That is so much better.

I say that to my AI agents now, too. But they've been working. The problem is they've been working all night. I'm already exhausted by the time I say it to them. They've been working all night. You think I'm kidding, but I'm not. They've been working all night.

Guest 2

The strength of your relationship with your AI agents is just beginning to worry me ever so slightly. I just want to put that out there.

What a world we're living in. Can you believe that when we all started in venture—obviously at different times—here the 4 of us are?

Totally.

It's just like, what is going on, man?

Guest 2

It's a good time in a bad kind of way.

2. Goldman Sachs Acquires Industry Ventures for $665M

Guys, I'm so excited for this. We have a very special guest in Roger, one of my favorite people from the industry. We're going to start with very fresh news: Industry Ventures has been acquired by Goldman Sachs—$665 million as the starting price, with, I think, a $300 million increase depending on performance over the next 5 years, up to 2030. They have $7 billion under management as an asset manager.

I wanted to start with this. How did we analyze and think about this when the news broke last night? Good for Hans, good for the founder. He grafted for 25 years. It felt like it was just after the crash in 2000, and he built that secondary business. It was a great entrepreneurial act, and well done. Let's start with that: just well done to the guy.

Can I ask an ignorant question to Roger and Rory? Maybe I should know this, but I don't know how fund-of-funds economics really work.

At first I read it and thought, “Congratulations for not pushing to $1 billion.” It's $970 million with the earnout. Congratulations on letting the ego walk it back and not putting your fist on the table and saying—in 2025 founder language—it has to be over the top. That was 1 thing.

But then I stepped back and thought for a minute: Hold on. They've got $7 billion under management, right? Here's where the math—I should know this, but I don't. I know it's a fund of funds, but imagine you're taking home a minimum of 2% of that, right? You could do more. That's $140 million. What am I missing in the math?

Guest

You're missing the math. You don't get quite that much. You're implicitly asking 2 questions, maybe. One is: How are asset managers valued? And then, separately, if you're the owner of a business, when should you sell versus keep it and just keep the income stream?

On the first, look, this thing traded at roughly 10% of AUM. I actually went and looked it up, because the mental model for assets under management to enterprise value is very varied, which makes sense because there are different models.

Carlyle and KKR, where they own all the economics, roughly trade at a market cap of 20% of AUM. In other words, if you manage $100 billion, the asset manager entity is worth $20 billion, right? Here, it makes sense as a secondary because the economics aren't typically quite as good as primary investors. It's at 10%.

And then you can go all the way down from there to public asset managers trading at 1 or 2% of AUM. At 10%, it felt about right for this kind of thing. It's, I think, a little bit in line with StepStone and Hamilton Lane and other publicly traded fund-of-funds as well. So it felt around the right price, right?

Guest 2

Much as I hate to agree with Rory—because I don't know, I have so much joy disagreeing with Rory—I actually used to be in this business. I was in financial institutions M&A at the very beginning of my career and sold asset managers. His analysis is spot on.

This is a hybrid, and if you look at it as a percentage, as a multiple of revenue—if you think of, so, Jason, if your numbers, the 20%—if you were to say the expected earnings of that pool is 20% as a fund of funds, I get 10% of that, right? So it's like 2%, so call it $140 million, and then you get some management fees on top.

But of course, if you were to say 20%, it's probably not 20%; it's probably 13 or 14%. You adjust it, and then you're saying it's probably like 10 times revenue for a very, very solid business with a great brand.

I honestly do think that they had started out early in the primary business. Industry was one of the first institutional LPs at IA Capital. They were in IA Capital, they were in IA Capital 1 and 2. Obviously, their secondary business now dwarfs their direct, initial-investing business in funds.

Kudos to Hans. It's been a grind, but they have really ridden the wave beautifully. I think it is a straight-on-market deal.

And for Goldman Sachs, the rationale behind it is they can push a huge amount of their private clients into Industry Ventures moving forward.

Guest 2

Yes, it's a platform for them. Look, all the public asset managers are desperately trying to get into private assets because, at the most basic level, you can get your S&P exposure for less than 10 basis points versus 200 basis points running private capital.

So if you were an asset manager in public stocks, your business is eroding away super fast.

Active management is going away. If you want to be an asset manager, and Goldman is a big, big asset manager, getting a platform like this that you can expand just makes a ton of sense. So, you're right, Raj: they're probably paying 10 times sales, which, on a 50% margin business, is 20 times earnings.

It's a very healthy price, but they're sitting there going, we can jam this through our channel, expand this 10×, and keep our asset business going with some high-expense, high-fee assets at a time when your public assets aren't nearly as profitable.

There's also 1 other point I'd like to raise. You're 100% right, but just a little insight on Goldman Sachs. Goldman has this platform called Apex, and Apex is a platform for their high-net-worth individuals, where they bring these kinds of deals—primary and secondary deals—to their ultra-high-net-worth clients.

This is something where they can actually institutionalize it, create much more product, and it's just something else to give to their ultra-high-net-worth clients. Plus, Industry Ventures itself has institutional clients that can now become Goldman Sachs clients, so it's kind of a win-win, both from a product perspective and from a distribution perspective.

I'll tell you what I like about it. Maybe I'm in the wrong platform, but I get these calls from Morgan Stanley to invest in private equity, and first I'm like, "Have you looked at my exposure? Don't you have access to my account?" The ideas are so dumb. They're so dumb. They're so dumb.

If Hans and the team stay, I think they said they had about an 18% IRR over their history. If that's every day in and out, that's a good baseline for folks to get into, right? If they can productize that, I would take that 100 days out of 100 versus the crazy calls I get from Morgan Stanley. "We'd like to get you a little more private equity exposure." Well, I'm 90% as it is, guys.

Guest 3

But maybe 95% is the right diversification. And I think the other thing to note that's interesting is what kind of GP businesses can, in fact, be 100% sold, right?

The interesting insight is probably not a pure venture firm, because you can't sell 100% of Benchmark. Then you don't have Benchmark, because you don't have the 5 great guys who are doing Benchmark, right? Whereas this is a more productizable business. It's got a lot of secondaries, and it's a lot of fund-of-funds, just like Greenspring, which was a large LP in Scale, got sold to StepStone.

Same kind of thing. These are the kind of businesses that can be sold 100% into a larger institution, and it can work for both sides. Obviously, the seller gets a great capital gain, and the buyer gets something that they can blend in.

You couldn't do that with a venture firm, I would argue, especially a small venture firm, because in the end, all you have is the 3 people. If you cash them out 100%, then you don't have anything, right?

So, I will admit, at times over the last 25 years, you kind of look at the secondary business and go, "That's not nearly as interesting as the business we're in." I love being a primary investor. I love doing my deals. But 1 thing you recognize is you can't sell this business, and Hans could sell his—and he did. So, well done, Hans. Who's laughing now?

It's the fee stream. You're 100% right. Like you mentioned, the Blackstones and the Carlyles and folks like that are asset gatherers. Ultimately, Hans is an asset gatherer, and most of the revenue from it comes from fees. It's not dependent; he's built a machine. I mean, kudos. He built an asset management business. The rest of us here do not run asset management businesses.

Guest 2

Well, give Harry time, but point taken. Give Harry a couple of years.

Guest 3

The more your business is predicated on either a brand or some institutional thing, the more it's like a business and the less it's like just 3 to 5 partners picking great investments, the more monetizable it is.

I could totally believe that your media company with a venture fund attached could be monetizable in a way that Roger's fund or my fund will never be, right? Scale will never be, right? Obviously, Andreessen, who has clearly embarked on the AUM and great-investing journey to bigness and maybe an IPO, believes the same thing, right? There are some businesses that are—

Guest 2

General Catalyst, same thing.

Guest 3

Accel stated the same thing. And, yeah, I would argue even Y Combinator, for example—not saying it's getting old, but it is the definition of a business. It's independent of the greatness, or not, of the current operators. It kind of has heft over and above that, right? Those kinds of things can be sold.

But if your only asset is—I mean, we're going to talk about Roger's new fund in a second, but let's be clear—the only asset in Roger's new fund is Roger's IQ as a stock picker. God help us, right? And without Roger, it does nothing, right? I thought that was good, Roger. I thought it worked nicely, right?

That's just not a monetizable asset, and it's a continuum. But maybe, maybe you can sneak out as a media company. You know, Roger and I are just destined to stay here and be simple, humble stock pickers.

This is so nice with Roger here. You give him shit, not me, and you defend me. This is great. This is like deflection. Roger, nice.

Guest 3

Anything for you, Harry.

3. Thinking Machines Co-Founder Raises $2BN and Then Leaves for Meta

Dude, you're too kind. Now, Jason, this next topic, I felt that you might have a perspective on, given our prior chats. Andrew Tulloch leaves Thinking Machines, the company he co-founded and raised $2 billion for, to join Meta for a reported $3.5 billion. Well done, Andrew Tulloch.

Listen, I hate this term.

I'm triggered.

Should have done computer science.

I'm triggered.

Jason, what did you think? I know you have thoughts.

Well, not only that, literally, I was on LinkedIn just yesterday, and a founder I've known from a distance for a while—I didn't realize he'd left his unicorn and just raised $20 million from Accel to do his next company. I'm like, I guess it's totally cool today to do that. I guess it's totally cool to leave Thinking Machines.

Or did it happen with Ilya's co-founder, too? I get it all confused. Who went to Meta? I guess it's just cool to forget about raising $2 million at a demo day and quitting or keeping the money. Now it's cool to raise a couple billion and check out.

I just don't know, and if it is, I don't know how venture should adapt, if at all. Is it just a risk factor when you invest in Thinking Machines? What was Thinking Machines' pre-money, $10 billion or something like that?

It was $10 billion post.

Good God. And so people are checking out of $10 billion seed companies now.

Honestly, I don't know what—I just feel like a fuddy-duddy, because when I was a founder, good God, there was no way I would leave, no matter how hard it was with my startup. I would just never consider it. And now it seems like it's cool, man. It's cool.

And even your accelerator will take you right back after you quit.

This next sentence is genuinely not meant to be snarky, Jason, even though it's going to come across as it. You absolutely would never quit. You would hang in there, but you also probably never faced the existential dilemma of being offered $3 billion to quit, right?

Most of the time, most people, independent of their startup, aren't worth a multiple of their startup valuation. If they're lucky, they're worth a $500,000-a-year salary. So, in the face of unprecedented wealth, I'm shocked to discover that most people behave badly. The loyalty conversation erodes pretty quickly when you enter the third comma.

You know, that's all fair. You said "fuddy-duddy"; I didn't. But the fact that the notion is, these are people that backed you and believed in you and supported you, and you just peace out and do something else—I do have a bit of an issue with that.

And there are levels to this shit, right? Where, yes, a person sitting at the helm of a $10 billion post-money company, whose shares are worth $2 billion, then goes for $3.5 billion—to me, I look at that and I'm like, "Are you fucking kidding me?" If my kid did that, I would not be happy with my kid. I would be like, "You leave the people that brought you to the dance because you see a prettier girl over here." I don't know, man.

Something's broken in the way that we're evolving as humans if everything ultimately reduces to what's in it for me and there's not another vector involved. I'm super competitive, but I am a wildly competitive guy. I want to win, but it's not at any cost.

Let me say 1 more thing, and then I'll create some oxygen for others. Some of these deals, like Scale AI, were basically—I saw that, and it reminded me of the old-style asset purchase versus the stock purchase. I don't want the liabilities. I don't want all that other stuff. I just want this asset, or in this case, I just want these people.

That has now become de rigueur. I mean, it's hard to call it an acquihire when you're talking about many billions of dollars, right? But essentially, it's an asset purchase, and I think that's something we'll continue to see more and more of.

Guest

But I feel less badly about that than I do what we’re talking about right now. Leaving aside the morality question—which I reserve the right to come back to and take a different perspective—the interesting question Jason asked is: What do you do? What protections do you have? How should investors handle this information, what should they do differently going forward, and what should other founders do?

To state the obvious, you see this interesting thread where founders are realizing extended founder vesting and cliff vesting, and protections for them versus a co-founder leaving, are a legitimate part of the discussion here right now. It may not even have mattered; he may not even have made his cliff. But it does point to being very sure that you and your founders have extended vesting.

Rory O’Driscoll

If, for example, these shares weren’t subject to vesting, then you feel even stupider as both a co-founder and an investor. And I mentioned the co-founder to make sure to make it clear this is not just a VC taking-care-of-themselves perspective, though we’ll come to that in a second.

Purely from a—if you’re a bunch of founders, if you’re 7 people leaving a safe job to go do this startup—you’ve got to run the game theory of: How will I feel if 1 of my 7 co-conspirators bails on me? And what should the economic penalty be to them?

So, if I’m a founder looking at this, I would be thinking about: Is there cliff vesting? Is there 6-year, not 4-year, vesting? Is there a repurchase right? Are there ways to make sure that this doesn’t happen and, if you leave for a competitor, something really bad happens? So, I think that’s got to be on the table because, yeah, even as a founder—

I’m just saying, Rory, is this not just symbolic of the conversation we had before recording, which is the increasingly transactional nature that we’re seeing in rounds, which I moan to you about? I’m a romantic. I like to fall in love with a partner, whether it’s an investment or a romantic partner. It’s super important to have the relationship.

And now it’s like, “Hey, highest-price auction process, zero relationship.” And this is just the embodiment.

Guest 4

Well, I think that’s just been true since we met, Harry. I think it’s just become institutionalized with AI, with deals being done on a Saturday for 9-figure, 10-figure amounts.

What I worry about this—and this is just me—I’m an early-stage investor. We’re mostly early-stage investors here. We’re all relatively early-stage. I don’t believe liquidation preferences matter. I don’t believe they’re a big deal, despite what they say next.

But my liquidation preference has always been—and I put it in quotes; it’s not true—knowing the founder would never quit. That’s my protection as a seed investor. The regular stuff is at the margin, right? Forget the preference stack, or $1 million raised, or $1 trillion. If I know Roger’s never going to quit, that’s the best protection I can get as a seed investor.

But if I’m investing and, no matter how good I think he is, he might quit in 6 months for something better, I guess you can adjust it on a spreadsheet, but it’s a risk I’ve never taken in my history. This has been my downside protection: He won’t quit. She won’t quit.

Rory O’Driscoll

The interesting thing here is the core asset in these investments is a group of 7 engineers, which is pretty unusual compared to most deals you do. Let’s be honest: Most of the time—and you correct me if you’re wrong—at the seed stage, the stage we’re investing at, you spend a lot of time with the CEO. You check in with the VP of engineering once. You just assume it’s a good team. You look at the product, you try and do your due diligence, but you’re not leaning in and saying, “You know what? This seventh, the seventh of 7 co-founders in a list, is pivotal to my investment thesis.”

So, it’s different here because we’re talking not about the motivations of a founder person, but the motivations of an engineer person who was an engineering academic. Then, remember, he spent 14 years or something like that at Meta, went to OpenAI for less than a year, was at Thinking Machines for less than a year, and then went back to Meta.

That’s an unusual career trajectory for an engineer, but it’s not crazy. I was this longtime engineer at place A. I bounced out to this other place, left with them, and then decided, “I just want to go back to the original place I was.” It’s kind of not an unusual pattern of behavior.

What is unusual in this case is, because of the technical nature of these bets, how much reliance we’re putting on the behavior of an engineering and academic talent pool, which probably responds fairly differently than the “I’m a founder, I want to be the CEO” talent pool.

Guest 4

Well, so—but Rory, that last thing to me is the bit, right? This isn’t just job-hopping and then, oh, eventually going back to the place where you kind of earned your stripes. It’s founder responsibility. And I think that’s what’s lacking here: that notion of, if I am taking on this mission with a group of people and a set of capital partners, that conveys a measure of responsibility that I’m discharging.

The minute that I say, “You know what? Screw that responsibility. My responsibility is to me and my not-optimal outcome”—I wouldn’t even say it’s the optimal outcome, the maximal outcome, the near-term max. Who knows whether or not this is better? This may well not be better, but the fact is, the people that they left behind are kind of screwed.

Rory O’Driscoll

Agreed, but I’m going to go back and, first of all, I’m going to do the money because I’ve known you for years and you’re a financially astute person. Question, Roger: Would you prefer $2 billion in Thinking Machines’ unlisted stock, with the chance to be amazing and the chance to go bust, or $3.5 billion of liquid Facebook stock over the next 5 years, just as a pure financial call? How—

Guest 2

Obvious.

Guest 4

Yes, thank you. So, let’s not pretend that they’re equivalent. I mean, you know, it’s such an obvious—

Guest 2

I’m not saying they’re equivalent—10-to-1 better, maybe.

Guest 4

You know, but there’s obviously way more option value in Thinking Machines, right? That could be a $500 billion company.

Guest 2

No, you’re exactly right. You’ve embedded option value versus probably highly liquid $3.5 billion, plus or minus 50%, versus $2 billion could be 0, could be 10. I’d also add the context of who gets it and when they get it.

The dude was at Facebook for 14 years before. I don’t think he was exactly desperate for cash. I said 14—I’m doing that from memory—but it was circa 10 at least. Right?

Guest 4

But it made a difference. The dude’s got $100 million already, being blunt.

True, again. But again, back to—I wonder, and I could be wrong on this next sentence. I don’t love the behavior, but I’m just advocating both sides. You, Jason, made a comment, then Roger made a comment. This person made a commitment.

I wonder, when I look at the due diligence process for that deal, where you have a very charismatic CEO and Mira Murati, how many of the VCs met him? Was there any emotional connection? I’m just wondering here: Did any of them even meet him in person?

Guest 4

As a VC, you should be fired if you write a $100 million-plus check and you don’t meet the co-founders. Period.

I’m with you. But Rory’s point is: How deep do you go on the org chart? And I’m positing, in the context of a transaction that came together where you didn’t even get to know what the product is, I wouldn’t assume a whole bunch of emotional connections on either side. I’m just pushing on the bullshit.

Rory O’Driscoll

Yeah, the big-picture thing is you thrust a bunch of money at people, some of whom you met once or twice, some of whom you maybe didn’t meet at all, and less than 12 months later, one of those folks went back. You know: “Oh, well.”

So, what I’m hearing Rory say is essentially, “Fuck it, the big VCs.” They’re playing the momentum game. If shit happens, shit happens. They can handle it. This is life.

Rory O’Driscoll

Yeah, exactly. Big boys’ rules. People who have $1 billion shouldn’t give a shit about other people who decide to grab it. You might think it’s bad behavior. You might think you wouldn’t back him again, but let me give you clearly: It’s the prisoner’s dilemma.

Once you’re not playing a multi-period game—and when someone’s offering you $3.5 billion, you’re no longer playing a multi-period game—you’re playing a one-and-done, right? You’re going to get bad human behavior.

Frankly, the real thing is, you as a person managing money should be thinking about how to deal with those corner cases. And, you know, I don’t know how you can—that’s the hard thing. If you’re paying $10 billion pre for a raw startup where there’s proven evidence that the asset, which is those 7 minds, will be pursued by someone who’s willing to offer them $1 billion, it makes it real how risky those investments are.

And I’m not sure what the answer to that is. It’s quite terrifying, really.

I mean, the answer is a bigger fund.

Guest 4

What?

A bigger fund.

Guest 4

Bigger fund.

Rory O’Driscoll

That way, you could have a few of these.

Diversification. Got it.

Rory O’Driscoll

You don’t want to be too concentrated with these deals.

But Rory, the way you said that was really astute. You’re right: what used to be a series of multi-turn games, when one looked at their career—if somebody acted badly and burned bridges, that might be their last company. They might not found another company.

Here, if you reduce everything, because of the scale, to a single-turn game, then that wildly increases the volatility of potential outcomes.

Rory O’Driscoll

You’re exactly right. Because, look, I’ll say something here: Roger and I lost money together in a deal, and I would say every one of the management team in that deal behaved well. Every one of them is referenceable by us, and we would give them money again, depending on the deal, and have talked to them about other deals, right?

It felt like a multi-period game. Everyone was stand-up and did the right thing, right? I think you’re right, Roger. These kinds of sums just change the calculus, and you can’t always rely on people doing the right thing. You just have to plan accordingly.

Guest 2

Rule number 1: we don’t admit that Rory’s right, even when he is.

Okay, this is rule number 1. To be fair, that was very astute. No, no, no, no, no. We don’t say this. No.

Guest 2

Harry, I haven’t worked together with Rory in a while, so I forgot that rule. My apologies. It will not happen again.

Rory’s cheating on you with Arthur Rock. Don’t worry, it’s okay.

Guest 2

Yeah. Yeah.

Okay. We said he left OpenAI. SoftBank is reportedly securing a $5 billion margin loan, secured by Arm shares, to invest in OpenAI. How did we analyze this? If they’re getting loans to invest in OpenAI backed by Arm securities, how do you think about that? And what did you think of it?

Rory O’Driscoll

Masa rules. No, nothing new to see here. This is what he does. For better or worse, when he has a feeling, he goes all in. All chips, maximum risk—personal, financial, everything. This is Masa being Masa.

Guest 2

100%. This man is full risk-on all the time and just wants to get the bet on the table. He has been spectacularly right at times and spectacularly wrong at times, but spectacularly willing to play, which, on behalf of the audience, we should be eternally grateful for, right?

And he’s not even that levered. I checked, actually: he owns 90% of Arm still in SoftBank, and Arm is trading at $90-odd billion. He’s got $80 billion of equity there. He can lever up some more. I mean, if he can, he will.

Jason, what did you think?

Guest 4

Yeah, that’s what I thought at first. It seemed to tie into the story of where the hell we’re going to get all this money to fund the tokens that I burn every day as a vibe coder. We still don’t have the answer.

Reflecting on it, though, I thought more about it. It’s actually a smart use of leverage. If you have a $90 billion or $100 billion position, I don’t know—if you’re an individual, you certainly don’t want to pay capital gains on it. A $5 billion margin loan at an acceptable interest rate is probably a smart position, right?

That loan’s probably not going to get called under any scenario, right? But it is kind of weird. It still feels like part of this whole thing where we’re all believing in Sam—which I do believe now—and we’re all believing that $1 trillion in revenue is coming.

Just for the record, let us remind ourselves that in 2002, you did see individual stocks on the Nasdaq go down 90% from the peak. So it is possible the loan will get called. It would just be unlikely.

Yeah. The question is, again, let’s just, for ease of analysis, say that Arm is $100 billion, right? That’s leverable pretty much to $50 billion now. So it very easily could see him back in the news with an incremental $20 billion. He could lever this. He could take $25 billion against the Arm position easily.

And then, to Rory’s point—and we saw this—the thing about Masa, and again, I’m old enough to have seen the Nasdaq run up, the Nasdaq crash, and Masa being Masa, is that he has had so many existential moments where he’s waking up in the middle of the night, sweat pouring down his face, wondering if this is it.

But he’s held tight, and he didn’t go over the line. He went right to the line, and he’s come out. Then his macro theses have been proven out. So, to me, this is a relatively low-octane Masa move.

Rory O’Driscoll

Exactly.

Guest 2

Exactly like that. Yes.

That’s a great visual for the show: Masa surrounded by fires, with low octane just around him.

4. More Data Centres Than Offices: Are We In a Bubble

The thing that I find hard is, when we talk about where the money comes from, the speed of what we’re seeing. We’re now building more data centers than office buildings, and we’re seeing demand for compute become the single biggest constraint. I’m looking at this going, really? The bubble that everyone’s talking about, compared with the fact that we’re building more data centers than office buildings and demand for compute is just off the charts—is this not fundamentally different?

Rory O’Driscoll

I think there’s a lot in that. The counterpoint between building data centers and building offices sounds clever, but it’s, to some extent, trivial, because who the hell is going to be building offices when there’s no one in them, right? The reason we’re building data centers is because we want to put computers in out of the rain. The reason we’re not building office buildings is that people are staying at home. So it’s pretty obvious what you’d build, right?

But I think, stepping back, the wider comment is that we go around this question—“Is it a bubble?” AI capex is going to expand for a while—and over and over again.

I was reading over the weekend this new Stripe Press book, likely Dwarkesh Patel's "Scaling AI," an oral history of scaling. I read it over the weekend. Really good, right? The thing that impressed me was the matter-of-fact way that a number of the people just reiterated it. It was interesting to hear it reiterated: the scaling law has been proven to hold for 6 or 7 years now, at a high degree of accuracy.

And a couple of them blithely said, “It wasn’t a question of when.” They were literally saying, “How long does it take? Of course, we’ll need 1% of GDP to invest in computers, but then we’ll be fine because we’ll have AGI.”

My point is, what I find fantastical—because that’s a shit ton of investment—is that they were like, “Well, that’s just what it’s going to take. And of course we’re going to get to that.” It was the matter-of-fact way in which the smartest people of our generation, thinking about scaling AI, accepted that this was the to-do list, right?

That’s a long-winded way of saying it’s not just, frankly, Sam spouting out of his butt. A whole bunch of these folks are saying, “Yep, this is the task we’ve embarked on ourselves for the next 5 years, and it’s going to take around 1% of GDP to build a compute cluster big enough to get the FLOPs to get the outcome we want.” So they’re going for it, right?

It’s back to what I said: it’s just happening. The only question is, how does the capital get financed, and can it earn a return, right? But if the capital is provided, this is going down. These data centers are going to be built. You can see these guys going, “This is just a slog for 6 or 7 years.” It’s totally predictable. The loss function is predictable.

It’s kind of like, “Why?” “Well, of course we’re going to buy 10,000 computers’ worth, so we’re going to buy 100,000. Then we’re going to buy 1 million, and somewhere along the line we’ll get AGI.” And what’s your point? Why are you even questioning it?

Guest 4

I can tell you just one thing, for what it’s worth. Forget the macro stuff. I’ve been Mr. Vibe Coder in the group. I’ve vibe-coded 8 apps. I haven’t built a piece of software, or been part of building a piece of software, since 2012. I’ve vibe-coded 8 apps in 100 days.

We have 12 AI agents working at SaaStr now. They’ve replaced almost all of our sales team and our whole content team. What I can tell you from that—and folks have been saying this, but I wouldn’t have believed it 90 days ago—is that folks like Amjad Masad and Replit are saying, “You’ve got it backwards. Everyone will consume every available token.”

What I know is that, today, with what we’re doing with 12 agents and 8 apps, I could use 100 times the tokens. I could use 100 times the tokens. Even now, I have to wait 20 minutes to build one feature. Vibe coding is cool, but it doesn’t work at Google speed. Our agents could all do more.

So, if everyone today could use 100 times the tokens—and think how early we are on the journey, right? As we record this, it’s Dreamforce week. Harry’s going to be there this week, I think. As Marc Benioff points out, only 0.1% of Salesforce customers are really using AI. So it’s 100 times 100 times something.

I really do think, and I can see it myself, that we’re not remotely servicing the demand that exists today, right? How it gets paid for is a different question. But I think this is so different from the prior waves, where we just can’t even service this demand.

The only thing I would actually question you on, because I’d love to hear your thoughts on it, is that my mental model of this is, from a technical and demand perspective, it’s all going to happen because the people who are building it want to build it, and Jason, at the margin, wants to use it.

So, if anything's going to constrain this, it's going to be economics. My big-picture model here is: you've got the technical trends, and then you've got the economic trends. The real question is, will it get the return? Will the capital be found to fund it?

I think if it's slowed down, if I'm wrong, it won't be because the technology direction is incorrect. It won't be because the demand isn't insatiable. It will be purely and simply, at the margin, the marginal capital provider says, “Oh my God, even though the scaling law is holding, the economic return from that investment isn't holding.” The scaling law might be log-linear, but every economic phenomenon tends toward diminishing marginal utility. At some point, capitalism is going to say, “I don't know how to tell you this, guys, but you can't have your $1 trillion dream because we just can't afford it,” and we're going to have to slow down a little here.

That's what I'm trying to figure out: are we going to get the economic return quickly enough to warrant the investment? Back in the day, you were a financially astute investor, so what do you think? Put on your trading and thinking hat here.

Guest 2

No, I think it's a—I think you've nailed the dynamics. Yes, economics will dictate that not everything that people want to build will be built, because the capacity won't exist once diminishing marginal returns reach a point where the value of the capital is not met.

I guess one vector that I'm not clear on—and clearly, the people that you're citing, Rory, are a million times smarter than me—is whether there are step-change advances in processing efficiency or simplification of code, such that the amount of processing required per unit declines in a way that people don't expect. Is that going to relieve some of the pressure on the magnitude of infrastructure that's being built?

I don't think so. I think we burn more tokens. Every company is like, “30%, 50% of my company's built with AI. Our engineers—hooray—our engineers came back: 50% built with Cursor.” Does that mean they take the rest of the day off? No. What it should mean is they're shipping more features, right? Because instead of spending an hour on Stack Overflow trying to find a library that was stale or pseudo-open-source, I can do it in 60 seconds. So I just go build another feature.

It's just—and the better that gets, the more tokens you'll consume. I don't think there's this great efficiency coming. I think we'll just build more and more stuff faster and faster. That's why it's actually so stressful at seed today, because so many of these companies are born almost instantly. They're born almost instantly today. So it's tough doing the A, the B, the C, the D, and the E.

But seed is really hard today, because that company probably didn't exist 7 days ago—or, being less facetious, 30. When we all started, even when Harry started, startups were never good 30 days in. They were terrible. Once in a while, an off-the-charts CTO would build demoware in 30 days that would make your jaw drop, but if you picked at it, it didn't work, right? It's just crazy what you can build so quickly today. It makes it so competitive; it's complicated.

Lovable had their 1st anniversary the other day. I thought that was insane: 1st anniversary, over $170 million in ARR.

Guest

Wow.

It's great. But it also makes that pre-seed, inception phase harder, I think, because how can you intuit differentiation in the way you used to be able to with a little bit of software? “Oh my God, Aaron and Dylan built a folder you could put a file in. I'm in.” Those days are long gone. “How do they do that? You mean it stores on the internet? Get me Rory.”

5. SoftBank Goes for $5BN Leverage Against ARM Stock To Buy More OpenAI

Guest

Yeah. No, it's moving a lot quicker. You're right. And you make a comment here: it makes it hard to be seed. You're right, because you don't know. It makes it also harder to be A and B because you have to pay off.

I mean, the period—and let's be frank, we're all looking for this, really. It's pathetic when you say it from the entrepreneur's perspective. What we're really looking for is that wonderful period where you know, but it's not obvious, and you can invest, right? It turns out that period may have declined to, like, 30 minutes. You have pre-seed Lovable, then you have, like, day 3, it's exploding in revenue, and suddenly you're at $2 billion pre. I mean, it's an exaggeration, but not by a lot. 6 months ago, they were raising at a couple of billion.

So the time period from “we haven't launched yet” to “Oh my God, it's so obvious,” as Jason said, has compressed to the point where it's just—you know, that sweet spot is vanishingly small. Therefore, you're left with the choice of whether to invest into acute uncertainty or—

Guest 2

Acute uncertainty, or businesses that aren't specifically disrupted by this phenomenon, which generally have legal and regulatory challenges that make it not simply, “Do I have better, faster, cleaner code?” There are a bunch of these other issues to address.

And is that your thinking? Because what we didn't say at the start, but Roger's getting back on the field, proving his timing, as is always brilliant. Is that your thinking when you're back on the field or as a seed investor?

Guest 2

I mean, it's part of it. The stuff that we're doing definitely is less resistant to the phenomena that we're talking about on this call. As you well know, Rory in particular—but you've got the rest of you too—acute uncertainty does not trouble me in the least when that acute uncertainty is expressing a deeply held, well-researched thesis that I have. That's just the nature of very early-stage venture.

But I do think that the issues of legal and regulatory complexity—and again, whether it's financial infrastructure, or it's media rights, copyright, patent, IP—make it more nuanced than, “Am I able to develop the next base model or a great platform for developing applications at warp speed?”

I think that's fair. Actually, we had Aaron Levie talking to some of our LPs. I'll maybe come back to that later. But one of the concepts he introduced was something I've been thinking about at the apps layer, and Aaron from Box is always so crisp: he talked about the diffusion rate of this technology across enterprise as a whole.

There's going to be different diffusion rates. The diffusion rate for Lovable will be very different than the diffusion rate of AI for complex medical prognostication. Having a handle on that, having your expectations accordingly, setting your investing thesis accordingly, and varying it by virtue of the diffusion rate, I think will be one of the key skills here.

And assessing, recognizing that in some markets it's going to be done and dusted in 6 months—and you're right, Roger, other markets where there are regulatory constraints, you might be 2 years in before you get your 1st big lighthouse, vertically focused enterprise customers, but then it's boing, ping, and you get the other 5 in 6 months. There are going to be very different adoption patterns by industry.

Guest

I'm not going to say it, but somebody might have said some relatively smart things right there.

I was paraphrasing someone else.

Guest

Aaron Levie—he's a smart guy.

What you don't know is that I prep Rory before the show. I sit down, I share my thoughts, and really—

Guest

I'm just prepping. I'm just—

6. Where is the Alpha in Venture in 2025

I know he's prepping you. Yeah. Yes, yes. You said something about value in regulated markets, maybe where it's more difficult to be disrupted.

Rory, Jason, and I had this great chat last week on the ability to kingmake and how capital can be used as a moat. We discussed, Jason, how you very well and eloquently discussed Polymarket raising $2 billion at $9 billion, and then this week Kalshi, the direct comp, raises from Andreessen Horowitz and Accel at $5 billion, again right after Polymarket's raising at $9 billion. How did you think about this kingmaking possibility? What were the thoughts?

Guest

I mean, let's be honest. What's going on here? This is the purest regulatory-arbitrage play of all time. If you look at the cumulative market cap of regulated sports betting and look at how it has dropped in response to the rise of Polymarket and Kalshi, who are not subject to the same rules and regulations that the legacy companies are, they're literally saying, “We're going to take value here and we're going to place it over here.”

The combination of, at least in the United States, the current administration being extremely predisposed toward prediction-market companies—and now Kalshi has announced that they're going to India as part of their 140-country coverage—if there were a level regulatory playing field, this would not be happening.

But for now, this is one of those circumstances where, when you talk about kingmaking, I think to an extent they are trying to run as quickly as they can to get so big and so powerful that they will not face the parallel regulatory scrutiny that the legacy companies have suffered through since PASPA.

Jason—

That was in my FTX investment memo: just get to that scale where we could push through some of these issues. I feel like we just came up just a little short. If we could have just waited for our buddy David Sacks to get in, then I think we would have really had a fun return on that one.

But it's a good point. Listen, you're obsessed with kingmakers, and I think it's a good topic, Harry.

I think it's right. The only thing that fascinates me about this kingmaking topic is that Polymarket was literally founded by a solo founder in his toilet during lockdown. His picture on Twitter showed him in his bathroom—that was the only place he had to work during the worst lockdown, in March 2020. He founded Polymarket.

It gives me inspiration that new founders will come out of everywhere, right? And so kingmaking works; it is a real issue to talk about. But if you can solo-found Polymarket out of your toilet in March 2020, who knows where the next one is going to come from?

Guest 4

I'm going to push back and bring up a couple of things. One is, I actually think this is an example of kingmaking not mattering, right? I think Roger nailed what's going on here. These are 2 non-sports-betting companies doing so-called prediction markets, where all we talk about is the 10% of the revenue that's political, while 90% of their business is sports betting—but we're not calling it that—and they're just killing it because we all love to sports bet, right?

The number of people who give a fuck about who's going to win the Nobel Prize or whatever else they're betting on that's not sports betting is low, but everyone in America wants to bet on the NFL, and they're cleaning up, right? Good luck to them and Godspeed. That's just what's happening, and Roger is totally correct.

A separate comment on the kingmaking implication, just tracking back to last week, because my short-term retention from memory is actually longer than a week, Harry: we were basically saying that money can pick a king, right? I think this is an example of where it can't. There are 2 good companies. They're both getting a ton of money, they're going to slug it out, and they're going to get relative market share, but I don't think there's kingmaking going on here.

There are 2 reasons kingmaking works. One is if you give one company so much money that they can overwhelm the other, then maybe that's kingmaking. The other is where getting money from brand-name, perceived VCs makes the customers default to you, right? That actually happens in enterprise software. If you're, say, an awesome CEO, and then you get 3 awesome VCs, and you're selling mainly to tech companies in the Valley, you probably have a herding effect.

I think Brett Taylor is an example of someone at the high end. There's a perceived, “Oh, Sierra is amazing. Would you want to take them on?” kind of vibe. I don't think that's true for a second. I don't think anyone betting on Polymarket or Kalshi gives a damn how much money they have, provided they can pay their bet, and doesn't give a damn who that money came from, right?

I think this is an example of non-kingmaking. To be very clear, I think it's just making the bet.

Guest 2

I think it's—but it's definitional. What does kingmaking really mean? To me, kingmaking means something different here. To me, kingmaking doesn't need to be 1 company; call it an oligopoly—a small group of companies that receive an exceptional amount of funding relative to everybody else.

Here, what I would refer to as the kingmaking is more money to spend on marketing, distribution, and team. At the end of the day, bonusing—that's what makes these companies go around—is the ability to—

Guest 4

Exactly—as long as LTV-to-CAC makes sense, and that's exactly what they're doing. So to me, that's the money. But you're right: customers don't give a fuck. They don't care how much money Kalshi's raised or how much Polymarket has raised.

Guest

Agreed. And look, to that extent, you're right, Roger, in the sense that every time someone gets money to something, that's kind of kingmaking, because you need money to pursue your business. And you're right: oddly enough, these will become capital-intensive businesses, because the spiffs will drive it, right, to get new customers. It's just like Uber, right?

But you're right. At that level, you can define anything as kingmaking, because whenever you pick 2 or 3, when 10 companies could enter a market and only 2 or 3 do, and then they suck up all the capital, that's kingmaking. But I would argue we were defining the term more tightly to say you can pick the winner. I don't think these are picking the winner as much as you're backing the winner, and there's a distinction.

Am I, as the venture person, making the difference? The hell I'm not. That money could come from anyone. As long as Polymarket has enough to pay out their spiffs and meet their CAC, the money's good, right? There's no anointing here. If you think back to kingmaking, there's the anointing process with oil. There's none of that here; you've just got money to play the game.

But you're right, there's no doubt that any of these games that become high-velocity, it's impossible to play. To take the other extreme, it's impossible to contemplate a bootstrapping world in this game right now, any more than it was possible for Uber, any more than it was possible for the LLM game, right?

Many of these high-growth businesses, thank God, lose money and rely on us in venture capital to fund that money, because otherwise we wouldn't have a damn job. If they could all bootstrap, we'd be forced to work for a living.

I'm probably allowed to say this because I'm outside the borders. We're not going to go into a political discussion, but am I the only one to also realize that Eric Trump is on the board of one, another Trump is investing in the other, and Howard Lutnick's son happens to run the fastest-growing investment bank? My word, that seems like an awful lot of coincidences in one go.

I wish I was as good at picking the Trumps. What a great deal, huh, for a regulatory-arbitrage play, Roger?

Guest 2

It was like the old days when you could work at YC and have your own fund on the side. You didn't have to invest through YC. It's a great deal.

100%.

Guest

If you're in crypto, energy, smoke, gaming, or prediction markets, right? There's a handful of things to which this administration has very tight connectivity. If you want help and support and you're in one of those industries, it's extremely clear what the playbook is.

Agreed. Which is why a coherent political philosophy is to say, regulate as few things as possible, because the more things you regulate, the more of this kind of behavior you see, right? And that kind of behavior tends to come from every party, because the minute something is regulated, people have an economic incentive to incentivize the regulators.

I will say the only real objection people have is that the current generation appears to know how to do it at scale. We're not going to do trivial little jobs where, “Oh, I get a nice job when I leave my regulatory position, and I get a nice $500,000-a-year job.” No, we're just going to go wholesale here. Just give me 5% of the company; it's so much quicker, right?

The efficiency of the regulatory arbitrage has definitely gone up. But I think the zoom-out comment is: whenever you have regulation, and there are economic incentives to get close to the regulators, you should have a bias toward regulating as little as possible, especially on economics, if at all possible.

Guest 2

I think this space is particularly interesting because you also have this issue of structural budget deficits in a lot of states—states that regulate gaming—and there are differential tax rates depending on the jurisdiction. Then you have these massive offshore operations of things like Bovada, Crypto.com, and Stake. These companies make billions and billions and billions and billions of dollars.

7. What 90% of Managers Get Wrong About Portfolio Management

The more that Illinois jacks up rates, in-state, the regulated sportsbooks that are subject to these rates reduce investment in the state, handle goes down, and tax revenue goes down. Those customers, who are now getting poorer service, are going to trade offshore in unregulated markets. So Rory's right again, Harry, sorry. In this case, it's such a clear example of how, as levers move, it has these effects in other parts of the market, and generally where it's heading is the unregulated part of the market.

Something that you said: “Don't piss off.” I always think, “Don't piss off Peter Thiel.” Peter Thiel has made a very concerted concentration play in terms of AI bets, very much all-in on OpenAI. It struck me because there was a piece announced this week where they shifted from caution to concentrated AI bets, meaning they were out of the market and now they're obviously very in the market, but with few players.

What struck me, though, was that I've interviewed Hemant at General Catalyst. I've interviewed the team at Lightspeed. I know the team at DST. They've taken the completely opposite approach of, “We don't really know the winners, so let's be in Mistral, let's be in Anthropic, let's be in OpenAI, and let's just index this wave of the best companies,” given the venture brains and Rory's phenomenal wisdom, may I add, that we have thanks to experience.

If you didn't know, with Arthur Rock, guys, Rory does. Poke, poke, poke. I wanted to hear your thoughts. How do you think about these 2 opposing plays in this new world, and where would you sit?

Guest

I'll tell you my guess.

I want to hear what Rory has to say. My guess is that being too diversified in investing in AI today is buying time. It's not knowing, not having the conviction, not knowing. I think it's better to buy time than to completely stay out. There's plenty of reason to do a check into leaders, even if it's not going to 10x the fund, rather than be grouchy, sit it out, or criticize these rounds.

But I think if you're Peter Thiel, sitting on what he has, you want to concentrate. He's, like, 40% of the capital in Founders Fund, plus his own capital. Making little teeny bets, little checks, doesn't get you there, does it? But if you don't know, I would do 100. If you don't know, you might as well do 100. The world is so different from 9 months ago. I think Plan B is to make a lot of bets.

Guest

I think that's actually right. There's a lot in this—there's a lot to unpack—so we'll take a little while. One is, look, we've discussed this: diversification reduces your upside. That's the nature of it; it also reduces your downside. I mean, it just is. It's the central limit theorem. It's not a great insight here, people. You will have a wider variance of returns, positively and negatively, if you have 10 deals in your fund than 30. Literally, the math is clear. So there's nothing there. Logically, the more certain you are that you can call the shots, the more focused you should be.

Founders Fund has both the evidence that they can call the shots, because they've done so, and frankly the confidence to call the shots because they got it right. I totally understand why they're going to try and be more focused. I actually looked at the article, and I was honestly surprised at how diversified they actually were, based on the information shared. Founders Fund I, the growth fund, had 31 investments. Founders Fund II had the mid-to-high teens, and Founders Fund III is aiming to have 10.

To me, I was surprised at how diversified Founders Fund I was. It just didn't feel in sync with what we've seen from these guys in general. If you look at their SpaceX non-diversification, these guys strike me as the most likely to be the most concentrated. So there was nothing surprising to me in that announcement. The only surprising thing was that they weren't there already.

Roger, how are you thinking about concentration with your new fund? You're back in the game. Do you want to do 100 investments out of your new fund, or do you want to just do 5 big ones and go big and go home?

Guest 2

To me, over a 3- to 4-year initial investment period—obviously, funds tend to go a little bit faster, more like 2 to 2.5 years—I tend to be at 20 to 25 portfolio constituents, just to create the farm team, but with significant ownership from each of those checks.

Where I've tended to get very concentrated is on the 2nd and 3rd checks, where we've gotten deep conviction in a team, their execution in the market, and the fact that if they continue to execute with that skill and at that speed, the market opportunity is massive. So, historically—and I'm following a similar playbook—we end up with 3 to 5 companies out of the 20 to 25 companies constituting 75% of the capital deployed.

75%. Roger, is your fund big enough, then, if you're making 20 investments with average $3 million checks today?

Guest 2

They're not average $3 million. Our initial check is way less.

Are you going to get ownership if it's going to be a smaller check size, given that?

Guest 2

It is in the spaces I'm investing in.

And they still exist.

Guest 2

I mean, we just wrote a $1.5 million check at a $10 million post, so 15% ownership, in a really, really cool analytics company that is disrupting a seriously stodgy and screwed-up sector that I think has generalizability outside of that space. So, yes, I do think it's possible to write those kinds of checks, and then, assuming they do a great job, we'd love to write a $3 million to $5 million 2nd check into that company, maybe more.

I'm switching to Roger's fund. I want to find these deals. It's been a few years for me since I've gotten enough of those. I'm switching over.

Rory O’Driscoll

If you're listening to this podcast and you can't see people's eyes, what you're seeing in Harry's eyes is the wide-eyed look, as if, “Can such things even exist? A $10 million post for a company with a product? Can such things exist?” And yes, they can.

And multiple six-figure ACV clients.

Guest 2

Yeah, I believe it.

5 on 50. I'll do it.

Rory O’Driscoll

And, Roger, therein lies the danger to your model for that follow-on check, which is the existence of people like Harry, who'll just snatch it away from you at a high price.

Coming back to the concentration, I actually think, Roger—again, at the risk of being nice—that's exactly the right strategy. It would be easy to say, “Oh, we're going to be concentrated,” but I think what you're saying is correct: you have to start off with a significant element of diversification and then concentrate down.

In fact, this is top of mind for me. We just had our LP meeting, and we would typically be at least a turn later than you, Roger, but my big-picture comment was that we've moved from a world where an exit is $200 million in ARR to a world where an exit is $400 million in ARR at an IPO. So you're just doing your thing here, but way over there at the finish line, the finish line has receded another 2 or 3 years, which means, logically, you've got more risk and more upside. You've just got to hold these things longer.

When you think about that, at the margin, that should have some impact on your portfolio strategy. For us, we've typically been at our stage under 20 deals per fund, and we kind of said you probably need to aim closer to 25, just given this dynamic. Nothing's changing at the stage we're at, but success is further away right now.

So, like you say, trying to concentrate back down, because in the end—I mean, it's no insight, but just to say it again—diversification is the enemy of upside. Concentration gives you more potential, more variance, but you have to do that via aggressive follow-ons. That's a very different strategy, just to call it out, Harry, from what Founders Fund articulated. But it's worth pointing out they're articulating that strategy for a growth fund.

The big aha here is how bifurcated and different different stages of business are. When you're still at the “Will this thing even work?” stage, which is Roger, or “Will it scale?” which is where we are, or maybe Jason somewhere in the middle, you probably need some significant diversification and then concentrate when you're effectively investing in what should be public companies but are just private. The growth fund strategy should be 10, 11, or 12 deals, concentrated. So it doesn't lend itself to a one-dimensional answer on that.

I think this question of how to handle portfolio concentration and what you should be aiming for is just going to be a key part of making the math work.

My challenge here is that I've done the portfolio reviews, and when I look back on Fund I, where there's a meaningful timeline to look back on—the 6 years now—the best performers, your Linears of the world, were not obvious early, and the early outperformers did not signify enterprise value in the long term. Clubhouse, Hopin, BeReal. If you think you can pick your winners early, I think you are wrong. Am I wrong?

Rory O’Driscoll

So, yeah, yes and no.

Guest 2

More yes than no. I think one of the aspects of the strategy we've articulated is this temporal, many-turn approach in order to be able to see progress. Yes, it may affect your ownership if, in fact, you don't have that high degree of confidence in the earliest days and you're leading or writing massive checks into every round.

For instance, I've got these very different ways of getting to multiple fund returners. They did not all look the same. The Trade Desk had multiple near-death experiences, multiple exit opportunities, and was bridged multiple times. It didn't have a product in market for more than a year and a half. That's one all the way over here: once it hit, then it hit.

Then you had something like Wise, which was chugging along, chugging along, chugging along, chugging along. Not that they didn't have hiccups along the way, but fundamentally, that was as close to an up-and-to-the-right company as I've ever been involved with. My first check in Wise was $750,000 at a $5.5 million post.

Okay. Yeah. Oh, yeah, yeah, yeah.

Guest 2

Then a round came in, and we piled in with a round at 20, and then we piled in with a round at 160. We just kept going, and we were 17% of The Trade Desk at IPO and 13% of Wise at IPO, out of a little shitty seed fund.

Datadog's a great example of one that ended up being a fund returner, but we had 2.2%—2%—at IPO because we were in there at pre-seed, and then RTP Global led the seed, and Index Ventures led the Series A. We were not comfortable, to be honest, backing up the truck at either the seed or the A. We wanted to, but it was sharp elbows; we couldn't get what we wanted in there.

We were able to write a check, but not as much as we wanted. We ended up having 2.2%—2%—at IPO, but that still was incredibly valuable because that was a $40 billion company. I'm just saying there are multiple paths, but the thing is, you're playing a multi-turn game, going all the way back to the beginning of this, because this is a multi-turn game.

Rory O’Driscoll

Yeah. In other words, just bringing it back to this point, what you’re saying, Harry, is—I think what Roger’s doing is rejecting the absolutism of your statement, which is that you just don’t know earlier on. He’s giving it more nuance: you don’t always know, but you know more than someone coming in from the outside.

So, at the margin, you can tilt it your way. That’s all you can do. They’re not going to put up a big sign saying, “I swear to God, I promise you this one’s going to be a $10 billion outcome,” right? But as long as you have a differential information advantage and the willingness to use it, what you’re saying, I think, Roger, is that you can tilt the thing slightly in your favor, which is all you can do, which is all concentration is.

I do think, by the way, after you get into revenue, it’s much more—at the stage we invest at, I do so with about a 70% confidence level once you have a year or 2 of revenue. In fact, I can say that exactly because we’ve discovered—we did the math—that if you get the first 2 years that we underwrote in terms of revenue from the moment of our investment, if you get the first 2 years correctly, your probability of getting greater than a 5× goes from 30% to the mid-70s.

In other words, once you’re in revenue and you have product-market fit, you do have a lot of information, and you ought to use that yourself. I do agree. Therefore, at the seed stage, there’s less information by definition, but it’s not zero. It’s not a flip of a coin, is my point. I think that’s Roger’s point, too. If you think back on yours, there probably was signal there if you’re close enough to it to be able to tilt the allocation slightly, and it can make a huge difference.

100%. I’ve realized, as a comment on that, Ro, it’s interesting you talked about it because I’ve realized even at our stage, you’ve got to be trying to do that more and more because the journey from our stage is still 10 years now, whereas before it was 6 or 7. Jason, what do you think at seed?

Guest 4

I think here’s what I—I sort of got this, but now, reflecting on our conversation, I see what I’ve learned. You can get a relatively high degree of certainty at seed. Not the certainty you get at late stage, but you can get it.

What it means is you’re going to end up with a pretty small box. You’re going to have to hunt strange things outside there. You’re going to have to be a founder-attractor magnet. You’re going to have to do something because you’re going to turn away a lot of the deals. You’re going to do the Wises, right? But you’re not going to do the other one that Roger said. Maybe you won’t do The Trade Desk or something. I don’t know. Or whatever. You’re just going to turn away some of them.

You’re not going to do the Clubhouses because it’s wacky, but it might be great, right? But you can still do Wise. I think that’s my error, because I’m concentrated from the first check, right? I have to turn away something where I don’t have certainty, even if it’s cool, because the risk is too high there.

I’m doing 8% of my fund into almost every deal, right? So I have to have such a high hit rate at seed. There are exceptions—I’ll do some small checks—but that’s really where it ends up being. A check and a half is 8%; half of them have to work, right?

It’s a stupid model because you have to turn away Clubhouse and Hopin and maybe even Datadog, but you’ve got to just find the Wises and go all in. It’s a big trade-off, but I do believe that you could just cut those ones out of Fund I and still have a decent fund.

Rory O’Driscoll

That’s so interesting, Jason. For me, 1% to 2% of the fund at entry—that’s what we do. Then it’s that next check that could be for 5% or 7% of the fund.

When we did DigitalOcean, our first check was $3 million. Then, when Andreessen Horowitz came in and led the $37 million Series A, we wrote a $7 million check. We had $10 million in 2 checks. In Datadog, we ended up having $9 million over 4 checks. In Wise, we ended up having $9 million over 4 checks, but that was a much smoother function.

With DigitalOcean, we could just see it taking off, as reflected in Andreessen’s leadership, and we were like, “You know what? This is going to be one of our best companies.” But what are you going to do today when you got in at $8 million post-money and the next round’s at $300 million because the AI kids come in, or $500 million? This happens all the time today. How much of your fund can you really put in? Is it even worth writing that second check? Is it even worth getting out of the office?

Guest 2

Depends. We are hyper, hyper, hyper-disciplined. We look at every check independently of the prior check. That’s the rubric.

If I look at that and say, “The information I’ve got about this being a $300 million or $3 billion company is that I think this could be a $100 billion company,” then I will write a very meaningful check into that company, up to about 10% of the fund.

That’s exactly what we did with The Trade Desk. By the time we had written 4 checks—pre-seed, bridge, bridge, and then a check into the Series A, which was barely a Series A—it was at a $16 million post-money. We just crawled to that point. Then it was whoosh.

Then there was an air gap, and there was nothing until the $20 million Series B, which was $15 million primary and $5 million secondary, at a $280 million post-money. We wrote a $3 million check out of a $50 million fund into that at a $280 million post-money, after having cumulatively written a little over $2 million over those first 4 checks.

That $3 million ended up turning into $40 million, and that was a great investment. Sometimes I wonder today if that math works as well, right? Let’s say you did the seed at $8 million and you own 15%. How big is your fund today—$150 million or something like that? Maybe—I think it was disclosed—$100 million, okay?

Guest 4

It’s about ownership. It’s a cash-on-cash business.

Fair enough, but it won’t change your carry or your personal economics all that much to go from 15% to 15.1%. That’s just not the issue. The issue is if it goes from $300 million to $10 billion and that check’s a 30×.

Rory O’Driscoll

I think there’s an embedded concern in what Jason’s saying that he’s not articulating, but when you articulate it, I think it’s okay. What you’re saying is: if all the following rounds are priced incorrectly relative to the ultimate exit value, does your strategy work?

It’s a fair question, and it’s the question you would ask if you’re living in Silicon Valley in 2025.

Guest 2

Don’t invest in those rounds.

Rory O’Driscoll

I was going to say, to save Roger the trouble of making the point: if the worst thing that happens is my initial check gets marked up and, on the follow-on, I don’t need to chase the money, then I’m money-good on the check I’ve written. Every check I’ve written is money-good, and provided every check you write is money-good in the end, you’ll die rich. It’s just one of those things, right?

I actually feel the need at this stage to remember that I was on a board with Roger for years. For the longest time—because he’d worked in finance before this, and this is a compliment, really, Roger—I actually thought he’d been an options trader, because no one I know understands options value better than Roger.

Halfway through, you explained to me that you’d actually worked in risk management, right, at, I think, Bloomberg or Goldman or somewhere like that. I do think you’ve got a very good understanding of option value here, and it’s just showing through, right?

Guest 2

Yeah, 100%. The odd thing, and the reason I’m so focused on this, is that as the exit bar has gone from $200 million to $400 million to $900 million or whatever it is, more of those dynamics are going to pervade the business we’re in. Because if you look at it—if you back into it from exit—this is what you’re dealing with.

For me, I think of the next check as the opportunity cost of cash. That cash can be deployed elsewhere in a new option, so to speak. As you see the price inflate—and I get what you’re saying, that the exit potential inflates too—but as you see the price inflate, the multiple does compress to some extent.

Guest 2

What is the risk-adjusted value of that capital? That’s the question every time, whether you’re writing a check at $10 million or at $1 billion. That’s it. You need to take all of those variables into account in order to make a decision.

Literally, everything in life can be priced as an option. Rory’s heard the spiel. I walk through life and everything looks like the Greeks. Everything looks like options theory, because that’s life.

Rory O’Driscoll

What you’re saying is that you have information there, right?

Guest 2

Yeah. Again, it’s also worth the Peter Thiel quote at this point. The big learning he had—and it is true—is that when you do a deal and then a big, reputable outside investor does a follow-on round at what feels like a high price, do everything you can in it because there’s a lot of signal in that.

Guest 4

It’s not always true, and I could cite examples where it’s not true. But risk-adjusted, information-adjusted, that next round in the deal that you’ve been in that’s performing well, provided the follow-on price is contemplatable, the advice would be: adjust your scales upward. Don’t be guilty of anchoring on what you did. You’ve got to find a way to take into account the information since then, both operational and from the outside round.

One nerdy thought—and maybe, Roger, you can educate me offline sometime. When I started investing, I learned from Founders Fund. I came up with that 10% threshold, too: put 10% of the fund into your winners. I’ve done it since I was able to do it myself. The mathematical problem I’ve had is, imagine you have 3 or 4 potential winners in your fund, right? You do exhaust a lot of your capital relatively early.

Listen, I’m not as good an investor as you, Roger, and I never will be, but I have come to regret some of my third checks because I just wish I had more flexibility in the midlife. I wish—because, yeah, you can back—you can get up to that 10% limit pretty damn fast if you have more than 1 breakout. You can be super disciplined and say it has to be OpenAI or better, but if your fund is $100 million, you could exhaust $30 or $40 million of it. But if you have 4 breakouts in today’s world, you could even do it in a year, potentially, right? Maybe it’s okay, but you run out of reserves. You run out of capital.

Rory O’Driscoll

You’re right. And I think what Roger is saying is he wants that to happen.

Guest 4

What?

Rory O’Driscoll

No, I know. I just think if you’re on the board and you’re not checking out—if you’re going to be on the board for 10 years, you have to be there. Even though the founders are now allowed to quit whenever they want, they can check out any day they want. I think if you’re going to own double digits, my view ethically is you’ve got to be there till the end. Otherwise, you’re dead weight on the cap table. You don’t get to check out after 24 months and say, “Great job, guys,” and show up once a year as an observer.

You’ve got to show up, and sometimes that means writing more checks, right? But that’s one of your core investments. The other thing is, you can always write small support checks after that. Just like, “I’m participating in the round, but I’m obviously not driving it,” because I’ve already invested $10 million out of my $100 million fund. That’s one thing. But obviously, in terms of stewardship, it’s one of your core investments, and you will be with it until the day you’re done.

Guest 4

Sure.

Rory O’Driscoll

The second thing—and this is a whole other set of conversations that we could potentially have another time—is that one of the ways I very intentionally structured IA as we moved through time was with parallel LPs, to be able to do cross-fund investing. That’s a very hard thing to do, but if you can do it, do it, because it creates that ability.

For people who don’t understand that, you’re saying: have the same LPs across funds, so you’re able to have cross-fund investing without conflicts?

Guest 2

That’s exactly right, so it does not become this existential issue. By the way, we dealt with this with a portfolio company in IA I, where IA I and IA II did not exactly have matching LPs. The LPAC was like, “Yes, we showed them the analysis. Whatever. If you want to do it, go do it. Just understand, if this doesn’t work, you’ve got some explaining to do.”

The risk of that check was not simply financial; it was reputational. Ultimately, we ended up deciding not to write that check. In the case with later companies—Fund II and Fund III—when we had parallel LPs, all of a sudden we weren’t investing out of a $100 million fund. We were investing out of a $260 million fund plus $16 million.

Rory O’Driscoll

I think that’s true, and we do cross-fund, too. The only comment I’ll make is, actually, it doesn’t directly address Jason’s comment, to be really direct. Any cross-fund you do is going to be a good deal; otherwise, you’re an idiot, and you’re not an idiot, right?

So I think good deals—you can find follow-on checks where you can cross-fund, even if the LPs aren’t fully aligned between funds, provided you run a process because it’s a good deal. And in the end, what I think the separate comment Jason was making is, bluntly put, how much do you keep back for your marginal deals if you’re stuck in them for 10 years and you want to be supportive, versus playing early in your best deals at the risk of not being able to make follow-on checks later, right?

I heard your answer, Roger. You’re basically cold-bloodedly allocating to the very best deals, and you’re willing to have $100,000—“Hey, I’ll try my best”—but $100,000 means I don’t have $1 million left in my pocket.

Guest 2

Correct. And obviously, at some point, we also try to get to between 110% and 120% invested through recycling. Recycling dollars can be used for those purposes, but I’m not going to optimize my asset allocation because of the potential for uncomfortable conversations down the road.

Guest 4

Yes, I remember one of those uncomfortable conversations, and you didn’t optimize your allocation for it. He said, “Just keeping score, Roger. Not that I forget—it’s only been 10 years.” Yes.

Rory O’Driscoll

Wow.

Guest 2

And I remember what our other larger investor said to you at that point in time vividly. I might remember that as well. This is great. This is fun. Harry, thank you for inviting me.

I do. I told you, Arthur was busy doing the board work that Rory should have done. Guys, this has been so much fun to do. Thank you so much for joining me. I’ve loved having you all.

Guest 2

Thanks so much, guys.