Benchmark 对阵 a16z：为什么专注单一阶段的基金会赢

Benchmark did something like 63 Series A's and had a 10% hit rate across 15 years. Dre did 454 Series A's and had a 2% hit rate. Now, in absolute numbers, they had 10 $5 billion hits, and Benchmark had 6. In those numbers is exactly the dilemma of the mega fund versus the focus fund. The focus fund is better at hit rate, has more as a percentage, and lower as an absolute number than the guys cranking through 454 Series A's. That's all you need to know. Ready to go.

Boys, this is the highlight of my week. I love these chats. It's terrible. The team are like, “Harry, why do you enter these ones so much more excited than the others?” And I tell them it's because Rory's going to call me a hypocrite and bash me again, and it's going to be so fun. So thank you both for joining me again today. We're just going to dive straight in.

1. Analysis of $3 Billion Windsurf Acquisition

Okay, Windsurf's potentially confirmed acquisition for $3 billion happened literally half an hour ago. Tweets are going out. How do we feel about this? What changes? What does it mean? What should we take from it? Rory, you probably have better thoughts than me.

Guest

I'll tell you, it's funny. The world is changing so quickly these days, right? I can't keep up. When we talked about this a little while ago, I thought $3 billion was a lot. I'm not being facetious—I thought $3 billion was a lot. When I saw the tweets again, I thought, “It's not that much.” I'm not kidding, and I don't mean I'm not kidding because everyone is just hunting these mega outcomes, right?

Every junior engineer from OpenAI is raising at $10 billion pre, and I'm like, “$3 billion sounds like...” Listen, best product, best co, best everything—it has nothing to do with the company. I'm just—is “anesthetized” the right word?—I'm anesthetized to these numbers now that aren't in the tens of billions. It's like 2021, when you would say no to a $3 billion acquisition for a project management tool, right? That was Kanban.

I just think it's a great company and a great business. This is a place where you can go and, 3 or 4 years ago, do a startup, 2 years ago kind of half-pivot into a Visual Studio kind of fork, crank like crazy, and build something that gets sold for $3 billion bucks. Isn't that amazing? It's why this business is fun, right? All you have to do is just get it right and Bob's your uncle. $3 billion quid. It's a great outcome. Good for them. It makes sense for both sides.

Jason's entire analysis last time is correct. It's 1% of the market cap to play in one of the largest use cases for AI, and one that the core constituency of developers loves. It just makes total sense for OpenAI. So congratulations to the team and for everyone involved.

Do you worry if you're a shareholder in Cursor now? Obviously, you've just paid $10 billion. Distribution is everything. Touchpoints to end consumers are everything. I do not want to stand in front of the OpenAI train. Do you suddenly slightly worry if you're Cursor?

Guest

If you're going to worry now, you might have thought of that before you turned down the offer, if there in fact was such an offer. So it's too late to worry now. You know, the time to have checked your manhood was before you turned down the big number, because whenever you turn down a big number—again, if it was a rumored big number—whenever you do that, whenever you get to that point where you're turning down a big offer, one of the questions you always ask yourself is exactly your one, Harry: How will we feel when it crosses the tape that they bought the number 2 and now we're hanging out there, right?

Whatever regrets they're having or not having now, it would have been more useful a while back. My guess is they're probably using the following logic. I've been through this. When you're the number 2 and you get the heavy squeeze from the adjacent acquirer, it often makes sense to fold, because otherwise they might buy the number 1 and then you're done. When you are the number 1, you might be able to say to yourself, “I'm still the independent winner. I can create value. There are other acquirers.”

So they're probably bravely going forward, saying, “This is the bet we're taking.” But yeah, it takes real courage to turn down whatever was offered and say we're going to compete against these guys instead.

You know what's funny, Rory? It's funny when these deals happen and it's funny to talk about them. They're routine, but what is very interesting is how, especially when it's right after an investment, the different investors react up and down the valuation stack, right?

In my limited experience—you have more experience—it's not always what you'd think, but the advice you get and the feedback and pushback, it's wildly divergent, isn't it?

Guest

It is. It's not always exactly what you think.

In my limited experience—you have more experience—it's not always what you'd think, but the advice you get and the feedback and pushback, it's wildly divergent, isn't it?

Guest

If I just put in at $10 billion and this was a risk factor in my prospectus or my internal diligence, I'm pretty zen about it, right? If I did the $1 billion round at Cursor, I might be thinking, “Oh, this is the worst idea ever—not selling,” right? And then the seed folks that were friends with the CEO, they might be like, “Kumbaya. I only sold half in my secondary anyway. I put in, I sold half, and I love you guys, and you be you,” right?

It's just so different, the feedback you get. It's not that you can't trust the advice; it's just so biased, right?

Totally.

Guest

I think everyone talks their book. You're exactly right. They sometimes don't always talk their book in a way that you think makes logical sense, but we're just human. Everyone comes to these M&A discussions around a boardroom table mentally running their internal cap table and saying, “What does it mean for me?” And that's why one of the things I always tell CEOs is, when you get this kind of offer, understand those numbers for everyone and understand where people are coming from.

Can I ask you a related question, Rory? I'm just interested in what you've seen. There's a vibe check I've seen change as funds have gotten bigger and raised faster, right? For someone that just invested at a mega round, if they just get a 1x in a year or 2, it's okay. You and I met when I sold, and there was so much drama around every exit, right? But when I look at it, I first saw this when Loom sold and Andrea invested at 13, and they got their 1x back in a year. They're like, “That's cool, guys. Go have—let's do the next one together.”

You used to say, “Let's do the next one together” when you did the pre-seed. Now it feels like it's, “Let's do the next one together” at the growth stage. I think it's a good thing if it takes a little pressure off.

Guest

It is. I'll give you a different example of that that actually sticks in my mind more, right? Thrive invested in Instagram, and literally 4 or 5 days later it sold for 2x. In one sense, 2x is not the target return, but if you look at the PR for that, that is the sound bite that leads almost every Wall Street Journal description of them, because it just looks so amazingly savvy. You paid $500 million, everyone thinks you're an idiot, and 2 days later someone sells for $1 billion.

So there is a little bit of—it’s not just about the numbers sometimes; it's about how it's perceived and what it does for you as a firm at that particular point in time. I think that was an amazing entrée for those guys. They look so sharp, and then obviously they built an amazing franchise on it.

I think the Loom example is a different one. I think that was, “Oh my God, we paid a high price in 2021, and I'm going to be saved in 2023. Thank you, God.” Loom was a great exit. I mean, that one I was like, “Kumbaya, guys. Well done getting that out the door.”

I think I messaged Scott Farquhar at Atlassian, being like, “Dude, you could have paid $35 million more. Just give them a billion. You did, like, $975 million. Come on.”

Guest

Some weird story there. Some weird story there. Maybe they were careful not to push too far, Harry. Maybe a billion would have been a no-go.

I also think you asked me, Harry, a while ago, when we did this, about how many 10x deals I've done, right? And I think part of it, when I think a lot about these conversations, is that if you're a seed investor, in the early days it's all good, right? The checks are smaller. But once you have a 10xer, you do become risk-averse, right? Because it's material, especially if you have a double-digit millions fund. That's a big deal, right? And you don't want to lose that. You don't want to lose it.

But as I think some more, that may be true, but the classic thing I look back on and go, “Letting the winners run” is the first golden rule, you know? And so when you—it's probably a mistake, then, being overly risk-averse on my winners. I mean, I look back and I go, when something is working, it's the military doctrine: reinforce success, starve failure.

If you've just had a big 10x markup, if you've got Sequoia late-stage in, provided you don't know something they know about the deal, that should smell like the deal, because, remember, a 2x from here—which is all maybe what a 3x one turns your 10x into, a 30x—your winners are one of the key parts of making the math work, which we're going to talk about later.

Guest

And I agree. I've been both sides of that. I know the fear of, “Oh my God, I remember my first big win.” I'm like, “Oh my God, let nothing go wrong.” You lie awake at night, the night before the IPO, literally thinking everything's going to end. Will this freaking deal get done? And then it does. But you look back and you go—I'm thinking of a specific deal.

Guest 2

We took some money off the table early. I look back and go, I probably could have gone from 2x to 10x, which, to state the obvious, is a 20x. And that's a big difference. It's funny: Brian Singerman always taught me the value of the final double, or the next double.

Absolutely. Going from $6 billion to $12 billion in company trajectory actually can be 12 months of work. That is double your returns in a portfolio, often. He's exactly right, and what I say, kind of going back to something I was saying last time, is that almost everything about private equity is better in terms of making money than what we do.

The only thing we have in our favor is that every once in a while, maybe once every 5 years, you find yourself with 10% ownership in a thing that's already worth billions of dollars and is compounding like crazy, right? And you just got to lie back and say, “How far is this going to take me? All the way to, you know, tens or hundreds of billions of dollars?” And that's the outlier that you just don't get in PE that you do so occasionally get in venture.

2. Will Mega Funds Win the Future of Venture Capital

He's exactly right that the double from there is so much easier than grunting it out from $1 million to $5 million in ARR because you did it as a CEO to $10 million in ARR, then the shitty year where you only go to $16 million, then you reaccelerate a little. Oh my God. And now you've taken $10 million and turned it into $25 million of value. Whoop-de-doo. It's let your winners run. I totally agree with that.

We mentioned that, kind of bluntly, the sheer size of returns needed. Since our conversations have started, I have completely changed my mind on who is going to win venture in the next 10 years. Just roll with me, because I want you to react to this. Fundamentally, we see the shift from public markets and IPOs to private markets, with mega-funds like Lightspeed, GC, Thrive—you name it.

Okay, there will not be many more of those. The majority of sovereign wealth funds and mega-players have chosen their provider, their partner, at this stage, and there are 5 to 7 of them. So there are very few places for the OpenAIs, Anthropics, Gleans, and Riplings of the world to go, and outcome scenarios are bigger than we've ever seen, with trillion-dollar companies becoming more normal than ever.

Oh my God, these guys are going to print billions, is my takeaway. On top of that, their cost of capital is so low they can shit on me and you and do 10 on 100 for a seed round, as they've done twice in the last year to me. I think multi-stage wins the next 10 years.

Guest 2

There's a lot in that, and I'm going to have a go. This is where I have the advantage, which our listeners don't, of seeing your agenda. And Harry, I love you starting here, changing your mind at the last minute.

It's good to change your mind when the facts change, so I respect that. But the agenda is intellectually incoherent, then, because you have this one which says the mega-funds are going to win, and then the next thing we're going to talk about is Josh Coppelman's wonderful piece on venture arrogance score, which would kind of argue for the opposite side. Then we're going to talk later about secondaries being the only liquidity, which would also argue against.

So we're oscillating on this question a lot, but it's actually okay because it is actually the biggest question. It's okay that we're wrong at this point and even changing our mind week to week as we think about it. But because I got exactly half an hour's heads-up that you were changing your mind, I think my summary would be, on your thing, “Are they going to win?”

I think I'm going to say 3 things, and then we'll probably end up pulling them apart over the course of the conversation. One, I think they already have won because they have the money. Step 1 in winning is, if you've got $7 billion to invest from 2024 on, forget “will we win?” You have won, right? And you're in an excellent place, right?

The next level of winning first is winning by having the money. Then the second thing is, can they invest it profitably, right? And look, they're all excellent investors. Very basic comment: they already have between 50% and 60% of the capital. Provided they don't under-index the rest of the industry, they're going to have 60% of the wins.

So when someone says, “Oh my God, all these people have all the wins,” I'm like, “Well, dude, they have all the money. If you have all the money and you do all the deals, you get all the wins.” It's just math, right? The real question, for a long time to come, I think you described it exactly correctly, is that because they can do those later rounds, they're going to be able to option-value a seed deal and even option-value an A and a B. That's just the dynamic you're in for a while.

The interesting question over the long term—and it's only over the long term—is: Will that winning be enough? The Coppelman question. In other words, if 6 or 8 people each have a $10 billion pool of capital, can they all make money over the medium term? They will, in the process of figuring that out, trample on a lot of other people's economics.

But the long-term question, when the verdict will go back to the LP, is: Do those funds make enough of a return to warrant re-upping in 3, 5, and 10 years' time? To me, that's not as clear.

Right? So my operating assumption as a mid-tier firm in terms of size is, for the next 3 to 5 years, there are great big walls of capital that will make a lot of investing very difficult. And even if it doesn't work out quite as well for those firms over the medium term—not because they're not great firms; they are, but because there might not just be enough money to go around—even if that's the ultimate outcome, it's going to be, to your point, rocky and, to be very direct, a pain in the ass for the next 5 or 6 years.

You're competing against people who literally look at your Series A and, you know, it's like the mafia guy: “That's a really nice Series A you got there. Be a shame if it got broken.” You know, they're coming in on your Series A business and your seed business saying, “Hey, we can just roll over this thing.”

So, a lot going on in that, right? So, do you think, on winning?

Guest 2

I think in some elements they've won. In some, it's TBD, and it's going to be a long, interesting sorting-out period.

That was a lot. Sorry, Jason.

Guest 2

No, no, you've convinced me. Listen, whenever I get an email from a founder that has what appear to be those metrics, right—that on their seed they're going to get $10 million at $100 million, right? $10 million, $15 million—I just email them back. I'm like, “I can't compete. I don't take the meeting. I don't talk about it. I just say, ‘Amazing, you look amazing. I can't do the deal,’” right?

Once in a while, they'll email me back and say, “Well, what would it take?” And I'll just always offer the maximum that I structurally can. That's worked out a few times, but I've given up instantly. I fold before the first hand because I agree with you.

I don't have the answer to the non-obvious ones, right? Because this money is attracted to the obvious. It has to be attracted to the obvious candidates, right? But that's why—and I think that's why so many people glamorize inception investing, which I ain't going to do. I'm out.

There's one thing I ain't going to do, and more than this, the other thing you have the most respect for is true inception investors that aren't bucket shops, that aren't trying to get 1,000 founders to go through and take 10% of their company. Because picking those inception guys—I mean, dude, dude, they are doing inception investing harder than ever before.

I actually lost an inception investing check this week to the 2 big ones: $10 million on $50 million for some good people out of a good company.

Well, good people.

Guest 2

I don't know that I would—when I think about inception investing, it's not just good people out of a good company with lots of boxes checked. It's not. It is technically inception, but they've already checked several of the boxes.

The real inception is finding the guy down the street from you, Harry, at the carriage house, that didn't go to college, didn't come out of Stripe, didn't go to YC. There's nothing in that guy, and spending months with him, getting to know him, and saying, “Hey, here's $800K.”

That's too much work. No one works that hard these days, do they?

No one works that hard in venture. I've known a few people over the years that work that hard, but most folks are not working that hard. It's easier to just pay 30% more than Harry. That's the easiest way to do venture: get a big fund, wait until you have a term sheet outbidding them.

3. Why Venture Has Become a Bundled Good

Guest 2

Look, it turns out getting a big fund in itself is hard work, to be fair to you. I'm defending those $8 billion. But within the smart-ass comment, the true comment that you're making, Harry, is exactly right.

The easiest way to win, if you've got $8 billion, is not to try and pretend to be anyone's bestie; it's to just be willing to pay a price that gets you the deal. And if you're not making your economics on the going-in round, then you just got more degrees of freedom to do it.

I'm not saying every firm does that, but you're exactly right: the advantage of a wall of money is it means, in those early situations, you can price the thing thinking it's a bundled good. The seed, and even the Series A, is a loss-leading product. It's like milk at the grocery store: come on in, buy your cheap milk, but we're going to upsell you all the strawberries you can buy, baby.

You wait till you see the Series C and D, right? So that's what you're up against. To your point, Harry, there's lots of reasons why those big firms can win.

4. Does Every Fund Have to do Pre-Seed to Win Series A and B Today

It’s also—I saw your podcast earlier this week. There are all the other wonderful things that they bring in terms of platform and all that, but the most wonderful thing an $8 billion firm brings is $8 billion. My question is, can you even do multistage if you don’t do pre-seed today?

We have Series A investors here, and they’re like, “For fuck’s sake, Neil Mehta is doing the seed and the pre-seed for Windsurf. Lightspeed and General Catalyst do more pre-seeds than anything. You won’t see the A if you don’t do the pre-seed.” My thinking is about how bundled our goods are today. I don’t think you can actually do the As and the Bs and the best unless you do pre-seed.

Rory O’Driscoll

We’re wrestling with that. I hear you, and it gets to the thing. We’ve become a bundled good at the widest level, where you can go all the way from $100 million to $2 million and that product is on offer, right? If that product, and every other dimension, is just as good as a single-stage investment, you’re right, you’re up against it.

So, you have to think about how you see those seed deals. Do you have to do lots of them in order to see the As and Bs? It’s a legitimate question. How much bundling do you have to do?

Rory, why do you not do it with Scale, in total respect?

Rory O’Driscoll

I have so many LPs message me their love for you after our shows, and we thought about it. Look, put it this way: We’ve wrestled internally, and I would say, actually, we’re thinking about it. I think we have to look across the group.

I just lost a deal a couple of months back where I would say the number-one thing was that the other investors had a relationship from the seed, right? What we won’t do is bullshit and say we’re going to do a whole bunch of these and then not do it. I just don’t think that’s fair to the entrepreneur.

So we’re wrestling with that because if you’re going to say you’re doing it, you’ve got to do it, right? But it’s a legitimate question in this market, where so much is changing and so much of this bundling is taking place. You have to figure out how up and down the stack you have to go.

A zoom-out insight I had is this: I think about it a lot. To the founder, they don’t give a damn about your nuanced, stage-specific strategy. A founder wants 2 things from their venture investor: money—money, lots of it, with the minimum amount of hassle—and perhaps the maximum amount of help. Help, right?

From the founder perspective, they actually don’t care if you’re crap at seed. They don’t care if you’re crap at A, right? What I’ve realized—it’s very obvious when I say it—is that there’s no override from the founder side based on whether you’re executing your investment strategy well.

To a rounding error, if some firm has a “we do every freaking deal” investment strategy, the LPs should be paying attention to that because they’re going to lose money. But from the founder perspective, someone who’s loosey-goosey and drunk with money is their best friend.

So I don’t think this bundling thing is going to stop because the founders don’t dislike it; they’re going to love it. It’s only going to stop if the returns from it are subpar relative to the returns of people who are more specialized. I think that could be true, and we can discuss that, but even if it is true, it’s going to take 5 to 7 years to become obvious.

Rory O’Driscoll

Let me push back on that just briefly. I have a lot of founders in the process of fundraising who ask me, “How many deals do you do per year?” I know full well that they don’t want me to say 12. They want me to say 2 to 3 because they want to feel the love. They want to feel that when they want my attention, they get it.

I think—but that doesn’t reconcile with—I’m pushing again. Fuck, Harry, I do 2 deals a year. Shockingly, I’ve been in this business 30 years. I’ve done 60 deals. I’m a pretty consistent guy. No one gives a fuck.

The odd thing is this: The founder might want you to be focused on him. But the truth is, to your earlier comment, if the firm is doing lots of deals—and many of these big people are—I think, realistically, we do as a firm 8 to 9 deals a year and have done so consistently for the last 15-odd years. There is a big advantage in terms of news flow of doing 20 or 30 deals a year. There’s always something good in the portfolio, right?

Again, it’s all part of the same theme. There is no forcing function between the founder and the investor that worries about investment-return quality. That’s a dynamic between the investor and the LP, and as long as either those funds are working—or people don’t think they’re working, don’t realize they’re not working, I should say—that money is going to be there.

Does that make sense?

Rory O’Driscoll

Mm-hmm.

5. Why Stage Specific Firms Will Win: a16z vs Benchmark

So, we’re going to have people who can do—I’m sure you saw—I don’t know the guy, Rottman at DST, who did an excellent piece on Series As, right? It’s a really great analysis of the volume of As that some of these firms are doing, and then the hit rate and the success rate, which we can talk about later. Firms that are doing well are doing 20 or 30 Series A deals a year. I saw Benchmark had a 33% hit rate on $5 billion companies from 2013 to 2018. That blew me away.

Rory O’Driscoll

Yes, I looked at it, and there are 2 facts in it. One is that Benchmark had a 10% hit rate, and all the other 19 investors listed here—18 investors—had a hit rate between 1% and 3%. Let me repeat that very simply: let’s call it an average of 2%.

So you have a sample set of 20, and I’m going to push, because it gets right back to your comment about the firm that has chosen to be most successful at one stage. It had a hit rate of 10% across 14 years, and the firms that have chosen to do everything have a 2% hit rate. That’s probably not a coincidence. What it said to me is, contrary to what we’ve just been talking about, the focused firm pulled it off better.

Right. Now, it’s super interesting. Compare it to Andreessen, right? Let’s stipulate that both of those firms are amazing and all the people there are wildly smart. Let’s just stipulate that. Therefore, we’re just comparing strategies, right? Then we have to figure out which is the best.

Benchmark did something like 63 Series As and had a 10% hit rate across 15 years. Andreessen did 454 Series As and had a 2% hit rate across the same period of time. Now, in absolute numbers, they had 10 $5 billion hits and Benchmark had 6. Those numbers are exactly the dilemma of the megafund versus the focused fund. It’s all right there.

The focused fund is better at hit rate, has more as a percentage and fewer as an absolute number than the guys cranking through 454 As. That’s all you need to know, right? It just shows clearly that if you scale the thing up, your quality does slip, but the aggregate numbers keep going up.

The only question, therefore, is which of those strategies makes the most money? Does the larger-volume strategy, which 5 or 6 other firms are pursuing as well, pass the return threshold? If it does, that bigger strategy works. That’s the guts of the question.

It was great analysis. I never met this guy, but I printed it off and read it for 2 or 3 hours. I was like, “There’s a ton of information in there.” Those deals also were a while ago, too, right?

Rory O’Driscoll

Yes. I mean, it was such a different time. I’m not saying they won’t reproduce it, but when I look back, when I was investing—that was when I started to invest, right?—it’s nothing like today. There’s nothing in common between 2013 and 2018 and when I started. Nothing.

There are 2 things there. You’re exactly right. There are potentially 2 consequences of that. Let me give you the first: “Hey, it wasn’t as easy as that, right? It’s going to be harder now.”

Rory O’Driscoll

Agreed—for a lot of reasons.

But if it’s harder now—and let’s move from hit rate, which is how many times you got it right, to what percentage of the best outcomes you got, right? In a less competitive time, the best firm, in terms of percentage of total outcomes, got 10% of the good outcomes. Benchmark got 6%, and everyone else bunches in the 2%, 3%, 4%, 5% range.

In other words, when there was less money than there is today, the most aggressive firm—and I say “only” with parentheses because it’s an amazing outcome—got into 10% of the great deals. Now you circle back to the Josh Kopelman point, and he realized it’s really hard to build a fund where the fund model assumes that, in 2026, you’re going to get into 20% of all the good deals.

The most aggressive firm in a less competitive market only got into 10% of the deals, and all the other firms did less than that. So this gets to my point: These big firms are going to find it really hard. It might be possible for one $8 billion fund to be freaking amazing. It’s going to be very damn hard for 6 or 7 $8 billion funds to be freaking amazing at the early-stage level to the same extent required to make the math work.

Well, for me, it’s just about the multitude of trillion-dollar companies that will exist. If there are 2, then you’re right. They’re fucked. If there are 10, there’s a business.

Rory O’Driscoll

I totally agree. Which is why it won’t be on the Series A, though. To the extent that the big strategies work, it won’t be because they get more Series As than 10%. They won’t get better market share in 2025 than they did in 2014.

It will work exactly as you said, Harry, to be clear. It all boils down to this: Are there companies that compound from $100 billion to $400 billion in the private markets before they go public? Right now, there are 2.

There’s SpaceX, there’s OpenAI. If you have 4 or 5 more of those, the math works for everybody, provided you fund them.

Guest 2

Can I ask you a really simple question on this? Harry’s trying to count how many dozen trillion-dollar exits he’s going to have and commit to his LPs. The global economy—the gross world product, what’s the acronym here?—is $100 trillion.

So, I don’t know what revenue multiple we put on everything, including grass and dirt, but out of a $100-trillion world, how many trillion-dollar exits can we have? In a $100-trillion world, including Mars?

I’m going to ground this in fact, because Harry uses “trillion” because that’s his Harry, right? The U.S. GDP is around $30 billion. The U.S. stock market trades at roughly 2 times GDP, so that’s $60–65 billion. Noah Smith wrote a great piece on not conflating income and market cap, but it’s around $60 trillion.

The more important point is, again—and it boils down to this—there’s roughly $1 trillion per year of new value created, plus or minus. That’s been true in the past. The real question is: does that $1 trillion go to $2 trillion? That’s roughly the order of magnitude we’re dealing with here, right?

6. Why AI Will Create Massive Unemployment

What happens if half of the tech labor force—help me do the math—is replaced by AI, which I did not believe 90 days ago? Now I’m 100% convinced. How many trillions does that create for tech companies if half the knowledge workers are turned into AI, which I think is going to start to happen next year—so fast? How many trillion-dollar startups do we get out of that math?

Rory O’Driscoll

I think talking about trillion-dollar startups is not useful. Even though there have been 6—let me make the following sentence. There are 6 companies with a trillion-dollar market cap, and I think all of them but Berkshire were founded by VCs and founded within my lifetime, at least. So, to the extent a trillion dollars is possible, it’s only possible in venture. That’s just a reminder to all our PE friends that, in the end, we are better, right?

But let’s not focus on a trillion because it’s just too much public marketing. I think $100 billion is the mental high end of good, right? So, now to your question on AI: if you’re selling the thing that allows your companies to be more efficient and allows the rest of corporate America to be more efficient, it’s got to be pretty damn good for you. Simple economics says you’re selling the thing that can cut costs and make companies more efficient.

Guest 2

Yeah, but I literally don’t know if it’s in their LP presentation, but these mega-funds—I literally think they’re predicated on this, and likely Vinod Khosla has been saying this for years. I didn’t get it until 90 days ago, until we ran our own AI with 130,000 conversations through it. Now I get it.

How many—half of these knowledge workers are going to be gone in 24 months? Can software capture 10% of that, 5% of that? What is an average knowledge worker worth? $200,000 a year. How many of them are there, times 0.1 for software?

Rory O’Driscoll

I’m going to take the counter on that. This is just the boring part of me: the past is the best predictor of the future. I think AI is exciting. There’ll be lots of savings; it’s the new thing we’re investing in. As yet, in economics, it’s hard to move macro dials.

If it lifts GDP growth from 1.5% to 2%, you’ll barely notice. It’ll be great—it’s going to be like PCs and the internet. It takes a long time to show up in the numbers, so it’s not going to be some step-function change. But that doesn’t mean you can’t make many multibillion-dollar outcomes from it.

7. The $100,000 Bet on the Future of Work

I don’t buy the mass unemployment in 12 to 24 months. I’m not even sure—I’m not smart enough to know what’s going to happen to GDP. In fact, I’m pretty skeptical that most software that increases efficiency really contributes much to growth. Come on, we all use CRM and all this stuff; it hasn’t really contributed much to growth.

Guest 2

But I’ve got to tell you, Rory, we do not need SMB sales reps next year. We do not need marketing managers. We need almost no one in customer success. We need no mediocre QA engineers. We need almost none of the mediocre product managers at startups.

I will bet you $100,000 that more of these people are unemployed in 12 months than you think.

Rory O’Driscoll

$100,000?

Guest 2

And it’s already happening. I’m slow, right? I didn’t get it. We’ve gotten rid of 5 people on our team in the last 90 days due to AI—5 people off our team—and it’s better without them. It’s not just efficiency, Rory; it’s better. They don’t complain about the job. They don’t complain about the job. As soon as AI is even 80% as good as a human, they’ll all be gone.

No one—you can’t get anyone to work at these boring SaaS companies. Literally, I talked with an old marketing manager I worked with. She’s not even that senior. She’s been out of work for 6 months, Rory. She says, “I need to make at least $300,000, and I just want to attend meetings.”

That’s what she said to me. I’ve known her for years. “I want to make at least $300,000 in tech, because that’s what I did in 2021. I went to meetings and made $300,000.” I’m like, “I’ll keep my ears out,” is what I said to her. “I’ll keep my ears out for that $300,000, meetings-only, hands-off-keyboard role.” They’re all going to be gone in a year. They’re going to be gone.

Rory O’Driscoll

You love extremes and, directionally, you’re correct. It’ll take longer, and I predict to you—even your business—you’re saving a bunch. So, you’re saving a bunch in OpEx, and I think you’ll be super-focused on this. You’ll save a bunch more in OpEx.

Guest 2

Let me tell you what, because you’re an entrepreneur. It’s not even OpEx. You know what the problem is? There’s nobody to do the work. That’s the problem, Rory, that VCs are missing. It’s not OpEx, CapEx, or anything else. We’re missing the fact that no one wants to work. No one wants to work.

This is an honesty that people don’t like. You know who says it? Fiverr, Shopify, Duolingo. If you squint at these emails, what they’re really saying is, “No one wants to work, so you’re out of a job.”

You’re the one different, Rory. You’re a better investor than I’ll ever be, but at least Harry and I are managing teams that aren’t just investors. I can’t pay somebody $150,000 to do anything faster, Rory. They will do strategy. They will write a memo that takes 90 days, and they will do something late.

Rory O’Driscoll

Look, no one wants to work. There are a whole bunch of jobs that will be happily automated by AI. I agree. I’m not fighting that trend; we’re investing in that trend.

I think the countervailing trend will mean—I don’t think it’ll result in mass unemployment. I don’t think your profits are going to quadruple. Remember, take Klarna as the public example. They talked boldly, and they probably did save those people, right? But I think in the end you’ll discover it’ll have roughly the same profits as other lenders. It’ll be marginally more efficient. It won’t be a step-function change.

Guest 2

Agreed. I don’t think anything’s getting more profitable. I think this is a misnomer. I think what’s happening is we just can’t hire people who are worth it, so we’re just going to turn the AI on. There’s just no one who wants to work.

My son goes to a school for founder-privileged children. Ten percent of the boys in his class just didn’t want to go to college or work. They were just fine doing nothing. It wasn’t that they had a trust fund; they were just fine doing nothing.

You go on LinkedIn and you know that little circle that says “Open to Work”? Talk to one of those people. They’re unwilling to work. I know this is going to make some people angry, but that circle means, “I’m unwilling to work. I need $300,000 and I’ll do 3 meetings a week.” That’s what that blue circle means: “Open to Work.”

Rory O’Driscoll

I’m not going to argue with grizzled cynicism from the front lines.

Guest 2

You think it’s cynicism? I’m actually bullish on it. I’m excited about the future now because I’m burned out trying to hire people who want 6 figures to do no work. I’m burned out on it.

I actually agree 100% with Jason. Come to London, where it’s even harder, or Europe; it’s even harder. As we know, it has a baguette culture, according to Jason. Or a red-wine culture, whichever one that was. That went down well in a European office. Thanks, Jason.

But you used the analogy of a PC and the internet. There is fundamental installation infrastructure and hardware that goes into that era of technology adoption. We now press Deep Research on a model that we were already using, and we can get rid of 3 researchers. There’s no installation, there’s no fiber, there’s no PC buying, there’s no implementation. It’s completely different.

Rory O’Driscoll

I think the timeline to see it is—first, again, I’m just going to say, as a zoomed-out comment—first of all, I simply don’t think you’re correct. I think GDP growth and productivity growth have been roughly 2% since the dawn of the Industrial Revolution, and I’m willing to lean into the fact that it’ll be 2% for the next 20 years. We all like to think the era that we live in is exceptional, that we’re just 250 years into compounding free-market capitalism.

Thank God, right? And while I think Deep Research is cool, I'm willing to bet that if you compare it to, “Oh my God, we don't have to pump this water out by hand. We've now got an automatic pump, and we can pump out the mine with a steam engine,” or, “Oh my God, we've got electricity. We now don't have to work in the dark,” I think it's probable that, at best, the inventions are equivalent.

So that's my macro comment, which I can't prove in detail, but I know I'm right on. Now to the specifics: to your point, yeah, you're right. Deep Research is amazing. I just love it for what we do here, and every time you're looking at a deal, if you're not running that out of the gate and doing a whole bunch of really great queries, you're toast.

But we're not going to get rid of all the associates. We're going to make them more efficient. You're going to be able to say, “We can look at more deals, we can know more, we can get some leverage from it,” and maybe you lose 1 or 2. So my point is merely that it's a great trend. It's a wonderful trend, but it's not going to be the step-function change.

I mean, it takes time to diffuse any technology, even AI. Actually, it will be interesting to see the adoption of AI in enterprises over the next 3 to 5 years. If you had a step-function adoption, then you guys would be correct. If everyone went in the space of 12 or 24 months from pre-AI to top-of-the-range, all-it-can-do AI, then maybe you'd be right.

I think humans just don't work like that. There are still people running DOS PC software out there. Maybe they're old, and you're right.

[Speaker?]

And there's no question the overall adoption curve for AI in deep enterprise is going to be slow. It's going to be slow. But the thing is, the early-adopter phase is so large in AI. One, and two, all of tech is becoming an early adopter.

When we started investing, tech was not the largest segment of the economy. Today, it is. So if these old manufacturing guys take 6 years, but all of tech fires half their team—I mean, Marc Benioff said, “Listen, I've got 6,000 people in support. I plan to repurpose them into sales.” I love Marc. Are they really going to be repurposed? How do you repurpose 6,000 people from AI?

First of all, I think that's great. The first part of that comment is really great. The second part is really great and fun. Let's break it up. I think you're exactly right. I will give you that.

What you're right about is the sectoral composition of the US economy in 2024. The early adopters are now a bigger percentage of the total. Obviously, we have an administration that would much prefer us all to be manufacturing toys at home so our kids could have 3 dolls. But given that we're not making dolls at home in America, you're right: it's tech, it's biotech, it's finance. There's a bigger percentage of US GDP that will probably lean into AI more quickly than, say, in the 1980s and 1990s, when a more manufacturing-centric economy leaned into computers.

So, as I think is true, you probably have some accelerated returns to scale from AI that you might have seen with the PC or the internet. It might be boomier or quicker. So I will give you that.

Yeah. I don't know how to draw the curve, but that early-adopter phase plus tech is so large that, in our industry, it's going to lead to massive human disruption. Just ask any founder, just ask any VC. But, man, if you're a hands-off-keyboard middle manager, you're going to be gone any year.

Closing the loop from an economics perspective, because I am a nerd, what you probably will see is massive productivity in those sectors, and then you'll have Baumol's cost disease in health, education, and some of the other stuff. So if overall GDP remains at 2%, you might have massive productivity gains at Salesforce and utterly no productivity gains in healthcare, maybe education, maybe some government sector. So I'll give you that: I think adoption within the tech sector will be super fast.

Now, to your second point: what happens to those 6,000 people? I don't know. I think likely Vinod Khosla has said this so many times. I was at Entrepreneur First, did a demo day out here, and I watched Vinod again after building AI with 130,000. He said half these people are going to be gone. There will be no jobs for them.

“We will build AGI, or whatever you call it. They will have income, and in the end there will be no jobs for them. So taxes will go up. We have to pay for them, right? And we will all be better.”

You can laugh at this, but when I heard him say this a year ago, I thought, “This is the guy who invested in OpenAI.” Today, I'm seeing it in my own AI. There are no jobs for those 6,000 people making $60,000 or $80,000 at Salesforce, with benefits and taxes. Wherever they are, they probably cost six figures. There will be no jobs for them. They may have to work at Subway, and it's terrible. There are no tech jobs for these roles. There is no job, right?

Every CEO I know at growth scale has some version of a hiring freeze going on, no matter what. Unless the growth is insane and it's all you can hire, you've got to get rid of somebody, right? It's all AI-first. So it's even worse for these folks because everyone's got some sort of soft freeze, even if it's just a quality freeze. Who's going to hire these people? They aren't Tier 1. Who's going to hire them?

To what extent is this not just Adam Smith's invisible hand? Some of the biggest shortages in labor markets today are ambulance drivers, fire-engine drivers, truck drivers, plumbers, and roofers. It's terrible.

[Speaker?]

Agreed. I was going to say, again, I'm always willing to change my mind when I hear new data. Thinking about what you said, and seeing that journal article on graduate unemployment creeping up, there's definitely overproduction of some skills and underproduction, as you say, Harry, of some of the more vocational skills. I do think that's a thing.

I think you have to give Peter Thiel credit. 10 or 15 years ago, he said the return on college is pretty good for the good student. It's pretty good for the STEM student, but the marginal return on the marginal entrant to college in the last 10 years is profoundly negative. You've gotten a set of skills that don't have market value, and you owe $150,000.

So I do think you're right there, and those folks are looking for the soft jobs—not soft as in easy, but soft as in marketing or other non-STEM skills—and it's just really hard. Yes, I buy that.

That said, if you've got 3 really smart friends at Stanford and you can crank out a VS Code fork, you too can have $3 billion in 24 months if you can just get it done. There's always going to be room at the top, baby.

[Speaker?]

8. What Does Harvard Losing It’s For Profit Status Mean for Venture

Look, college is already becoming UBI. It has to. A limited element of college is already UBI. Harvard's $200,000 a year or less—you don't pay. At Stanford, I think they raised it from $100,000 to $200,000; you pay nothing.

So great, that's at Harvard. Even though it may become for-profit soon, they can pay for this. But when every college is that way, it's just UBI. You've got to do something with these kids, right? I think that's where we're going.

Can I ask you—you mentioned that Harvard might be a for-profit. I mean, that was absolutely in the news. I put it as one of the number-one stories. When you look at, “We're going to be taking away Harvard's tax-exempt status. It's what they deserve,” which Trump posted to Truth Social, listen, I care about Harvard. I like them very much. They're great to work with.

I also worry intensely that this is going to happen to every endowment fund. And if it does, what happens then? Help me understand: is this the start of a much bigger wave, and how will this impact commitments to venture?

It's the end of civilization. What does it mean for me? He said, “Harry Stebbings,” which, by the way, I actually totally respect, because if we go off into some kind of blather about what we think about Harvard, I'm no more qualified than you or any of us. I'm just a person in the street when it comes to that, right?

Exactly.

What I was going to say about the impact is, again, unfortunately, going right back to the first thing: if it turns out that there is pressure on endowments—and I think even if this doesn't happen, and there are a lot of reasons why it might not, or it gets delayed—there's going to be huge pressure and precautionary cash planning in all the endowments.

Going right back to our discussion at the start, unfortunately, those are the LPs of choice for the small, early, innovative funds. It's another thing that's going to reinforce the big getting bigger, and it'll be harder to be new.

It's not a great trend, because if you're raising $8 billion, you've long since stopped talking to Harvard in a meaningful way. You're actually talking to, pick your sovereign wealth fund. If you're raising $150 million for your first fund, those are the people you'd be going to. So I think it's bad news within our particular world of venture investing.

And it's deliberately not trying to veer over into the political.

I do think we're in the business in venture of funding the things where the US has a massive comparative advantage, and that comparative advantage is typically caused by high intellectual property, high-knowledge-worker industries like biotech, software, and robotics, right? That won't be possible if we don't have a well-funded higher-education sector.

So, we can talk about all the old dumb things Harvard did and did not do over the last 10 years. Particularly with that report that came out on antisemitism, there's a ton they have to be ashamed of. But sticking back on my venture hat and avoiding trying to be Mr. Political for our industry, one of the non-negotiable ingredients is a strong and vibrant technology university system that generates graduates and research that have kick-started the whole thing.

So, I don't want to lose Harvard. They may be arrogant asses. They may do this whole, “I was at school in Boston” thing. Whatever. They turned me down 30 years ago. I'm still grim about that. But I don't want to lose them.

Guest

Yeah, okay. You don't want to kill the golden goose.

We have a good thing going here in venture, and a huge amount of it is just the smart, talented young people that come out of these colleges, educated and ready to go. Don't blow it.

During peak lockdown, we briefly took it for granted. Everyone was distributed. We thought maybe it didn't matter, but now we know it matters. It really matters, right? It really matters, right?

9. Why AI is Maiming and Not Killing Growth Companies on the Path to IPO

One piece of news that really struck me was—and it kind of went under the radar—but I called it an “acqui-ouch” because I remember one of the hottest companies at the time, a couple of years ago, was Census. This company was super freaking hot. Everyone wanted to invest. Sequoia did it, they did a next round, and then they got acquired by Fiverr. It was super under the radar. They'd raised $80 million from Sequoia, and I was like, “Wow, that didn't happen how we planned it.”

And I guess my question to you is: How did you guys think about this? Is this the wave of a series of companies that were supposed to be high-flyers just getting bought for cents?

Guest

Well, these are deals, for what it's worth, where, as a seed investor, I'm that guy that you talked about at the beginning. I'm the grouchy guy. Yeah, I'm the grouchy guy—the big, late, last guy, whatever. For me, that was my high-flyer, right? That was my fund-returner, and now I'm getting, like, 8 shares in Fivetran. I'm not so happy, right?

I have one of those deals. I was pretty grouchy about it. Now I have shares in a decacorn that will never IPO. It looked great on the press release, right? Wow. Yeah, I was that grouchy guy for good reasons.

Unlike me, I don't know the details of this deal.

Guest

But yeah, that's where it makes you grouchy. You put all the time in. If you're in there for a year and show up as a board observer, it's not your only hot deal, but as a seed guy, it makes you grouchy.

I mean, Harry, I don't know how to break it to you, but some deals don't work. It sucks. Maybe that hasn't happened yet, but when it does, hold the thought.

You asked what I thought of that. Honestly, I didn't even notice, because it's going to be one of 500 or 600 of these that's going to have to happen. There's somewhere between 800 and 1,000 unicorns, and a couple hundred are going to go public. The rest of them are going to have to be squeezed into other companies, and this is what that's going to look like over and over again.

The craziest one, even though it's not brand-new, was Lacework, right? It was as hot as Wiz, right? Especially me, I'm not a real security expert, so I thought it was, like, number 2. It was just behind Wiz, right?

When I was at re:Invent, it had a 7,000-square-foot booth. It had, like, a $4 million booth. I'm like, “This thing is neck-and-neck with Wiz, right?” And then it sells for nickels, right?

I think, at the risk of playing devil's advocate, yes, for nickels in terms of the enterprise value, but I think there was a significant portion of cash on the table. I think the investors, maybe rightly, maybe wrongly—I’ve heard both sides—looked at each other and said, “If we could get 50, 60, 70 cents on the dollar back from the cash that we put in here, rather than keeping going, maybe that's the right thing to do.”

Now, I don't know. I don't have the specifics, but it wasn't like they took $1.5 billion and burned it all up, right? It's that they just said, “2 years ago, we thought this was awesome. We've reflected. We spent $100 million; we have $800 million left. Let's just call it a day,” which might be a shrewd call. Fast-forward, I'm sure it was objectively the right call.

The stress for different folks in the investor stack can vary. For some folks, it's like, “Whatever. I'm on 20 boards. I don't care.” For someone else, it could be their only winner. The impacts are varied, right? It's never that great for the employees, though.

Speaking of investor exuberance, we saw that with Census. I'm sorry that you didn't notice it, Rory. You're clearly much busier than me. I'm just a humble podcaster. What can I say?

10. Decagon Raises 100x ARR: The Breakdown

My question to you on the back of that is: We also see Decagon raising at 100x in a similar style to 2021. It was $15 million of ARR at $1.5 billion. How did you guys analyze that return to 2021? Was it incredibly strategic?

Guest

I think it's interesting. It's not wholly crazy at all, right? If you run through the logic, if you look at the top 2 or 3 use cases of AI, the number 1 is just personal chat, the number 2 is coding, and the number 3 is customer service. Of all the areas—to Jason's point earlier—it's the one where the ROI, hear me out, is the clearest, right?

We talked to people, and I had an investment in this space pre-GenAI, where the resolution rate was roughly 30–35%. In other words, 1 in 3 calls got solved. We did a bunch of references around this space and around the impact of GenAI, and the conclusion over and over again was that with GenAI, you can get that resolution rate to 60–70%. In other words, you can handle most of your calls without humans, right? Customer support is a massive use case. So it's just a great big market, right?

You start with that, right? Then the only question, therefore, is: Is it going to be a winner-take-most market, or is there going to be more winners? And that's where it gets kind of tricky, right? I think Decagon's done an amazing job. They're right there with Sierra. They've established an interesting lead in some of those early spaces.

I think there could be a lot more competitors than that, but it wasn't crazy. If you could get—you’re leaning into growth—you've got a lot more runway ahead of you than many of the 2021 companies. So, I don't think it was wholly crazy.

I don't understand it. I'm sorry. I know Des Traynor very well at Intercom. Fantastic product guy. Intercom is a bloody amazing story, to not have a huge enterprise value today after 17 years. I know $2 billion is a lot. I know it's a lot, but they're fucking brilliant, and it's taken a lot of money and a lot of time.

And they are one of 10. You mentioned Sierra. You want to go against Brett Taylor? Good luck. And Neil Mehta bankrolling him. Then you want to go against the 50 others coming out of YC, all for individual verticalized solutions. Seriously?

Guest

Yeah, that's the con case. You're exactly right. Look, we've agonized about this market a lot. You have the pre-GenAI people, and I think you're exactly right: Intercom is by far the best of those.

They've done an amazing job of adding AI with Fin. If there's anything that a pre-GenAI company could do to get relevant in GenAI for customer success, I think Intercom have done it. So they get an A++, and they're Irish, so that gives me double votes. I like those guys to win, and no one liked their $2 billion, right?

You're exactly right. I think the problem is whenever you're one of the more mature businesses, you're encumbered by facts, right? You have a scale, you have a growth rate, and you can kind of project off that. If you're doing—I don't know the numbers—$400 million, growing at 20%, whatever, you can value that, and it's kind of bounded.

When you're seeing quadrupling and 5x-ing year on year, from $3 million to $15 million—or is it $5 million to $25 million?—people are just more willing to lean in and say the future's unbounded. You can treble for 3 more years, and suddenly the math works and you're worth $1.5 billion. I know how it happens.

11. Why VCs Are Upside Junkies and What That Means Today

We venture guys love new shit with option value over old shit with intrinsic value. It's as simple as that. We don't do intrinsic value. You know why? There's no upside in intrinsic value. We're junkies—upside junkies, right?

Can you just break that down for those who don't understand? Why is there no option value in intrinsic value?

Guest

Because if something is growing—if something's $400 million, growing at 20%, and it's been doing that for the last 3 years—you’re probably going to grow at 20% for the next 3 years, plus or minus.

Right now, there's a price at which you'd love that asset, but it's not going to trade at that price. It's going to trade at 6x today, it's going to grow 20%, and it's going to trade at 6x when you exit.

[Speaker?]

So, it's pretty bounded. There's no magic pixie-dust upside. If you buy it at 6 times and sell it at 6 times, you can double your money if it compounds for 4 years at 20%, right? With low growth, there's just no way to tell a story where something magic happens.

Conversely, a new deal that's $3 million—or $3 million going to $25 million—well, maybe it'll 5x next year as well. Maybe it'll go to $50 million, followed by $150 million, followed by $300 million, and, oh my God, that's still worth 20 times. It could be the next fill-in-the-blank. That could be worth 20 times $300 million, which is $6 billion. We can pay $1.5 billion now. There you've got your 4x, because you're selling futures and selling upside hope.

Now, people may be massively mispricing that option, which is what you're saying, and you could be right. In other words, the probability of that working might only be 1 in 100, and they're pricing it as if it's 1 in 2. In other words, they're pricing it as if it's certainly going to work. In fact, it just might work, right? And that's where these kinds of bets go wrong.

One of the hailed pieces that I always go back to is Bill Gurley's The 10X World piece. Yeah. And you know what? Like you said there about global GDP and 2x: well, you know what? 10x. I look at that and I'm like, getting to $150 million in ARR from $15 million is a real journey. You're paying for it.

It is a journey.

And in a way, look, we had some internal discussions: should we be one of the 100 people pleading to put money into Decagon at $1.5 billion? We had some interesting discussions. That is not a scale deal.

No, it's not.

That was my comment.

But my point is merely, going back to the thing that you said, Harry, the market is changing so much that if every day you're not saying to yourself, “Are we doing it right? Are there things we shouldn't be thinking about that feel unnatural to us?”—and if you're not even asking that question, you're missing the point. Conversely, on the other extreme, if you start drifting off and doing every new thing, you'll probably also screw up because you'll lose what you have.

That's the challenge of being an investing manager in 2025. If you just stick to the same old boring shit, you could be done. If you lose the plot entirely, you could blow all the money, and you've got to thread the needle.

I've just led a deal for a vertical SaaS for dentists. So I'm at the cutting edge of AI. Thank you very much.

You know, the one about Decagon—the meta-question I have is, I've done a lot of investing in support and know a lot about AI in it. It's like this: this is true of Windsurf too, but the defensibility is confusing. I think what they're good at is doing strong enterprise deployments, getting it done, doing the heavy lifting.

Guest 2

I just tried the one on Notion and Substack. It couldn't answer my generic question, but that's not its strength, right? I asked Notion how to embed my AI in Notion, and it said, “The team will get back to you in a day.”

I have a lot of the data. I'm on the board of this company, Gorgias, which is the biggest support company in e-commerce, and I see all the data. They have all the data for all the vendors. The honest truth is that Gorgias's biggest challenge is that they are objectively the best—there are ways to measure it—but the gaps are narrow.

Ripping out a support desk is a big deal. It's not going to happen in the enterprise overnight, right? But I do this moats-versus-momentum thing, even though it's venture nomenclature. I think about moats versus momentum. If Decagon is cool, but a competitor is better next year, I don't know. I just don't know, right?

[Speaker?]

I hear you, but the only argument I'd make is this: there are times when market windows open, and there's a couple of years where you scurry through, and there's no moat at that particular point in time. But momentum begets its own moat, I do believe.

Let's just say, fast-forward 2 or 3 years. I'm going to argue the following: the fate of the customer service market will be like this. Gorgias and 1 or 2 of the old guard will add enough AI and be really relevant—Intercom, Gorgias, a few of those. There'll be 50 new companies trying to do it. But let's say Decagon and Sierra—we have likely Observe.AI on the phone side of it—2 or 3 of them make critical mass and explode, right?

I think at some point, when you become the safe choice, windows shut and the opportunity to walk through them closes. So I don't think those companies will get eroded because 3 years from now Decagon will be at $100 million and new competitors will start taking their stuff away. I think this is a point in time, like Salesforce, where you have the chance to grab a 10- or 15-year market slot.

So I don't—I'll tell you what I worry about. It's not that I don't think you can get the momentum today. What I worry about is just that when there's so much competition, I think everyone's going to be less durable. It's not your 10-year-old SaaS company that's seeing less durability; I think everyone—I think this is new, less durable revenue.

I don't see any reason why the new guys, if you listen to Varun at Windsurf, would be different. He's like, “Our only moat is working harder than everybody else, at speed.” He's not claiming he's building any moat, right? This product didn't even exist 90 days ago.

[Speaker?]

You're exactly right. I do think in enterprises, the truth is, once you're installed, it's hard to take out for exactly the reasons you said. The shit doesn't work so well unless it's trained and tuned. You just have a bias to be there.

And then the other thing is, once you're the perceived leader, you do have all that positive reference value. I mean, you and I would both agree—I don't think Salesforce was winning in 2010 because it was the best CRM. It was winning because it was the default option.

Yeah, but I just worry that with so much competition, it's not just that your revenue is going to go to zero. I worry there's going to be more churn, more downgrades, and harder-to-win deals. Those benefits to hitting scale that we talked about before—I think they're less than they used to be, even if the budgets are exciting.

I just don't know how to predict where the future of anyone will be. There's going to be so many shiny pennies in AI and so much change that these chat apps are great, but when chat voice bots—what's just starting right now—is having real digital people join support, not dumb cartoons or somebody with audio that doesn't match the video. I'm talking about people better than a human joining it.

Now, maybe that's not Sierra or Decagon or Intercom or Gorgias or Zendesk. It may come from another place, right? These spaces aren't seeing a once-a-decade disruption or a once-every-5-years disruption. Now it's once every 5 weeks—literally, every 5 weeks, a disruption.

So this lack of stability is where I think Decagon's revenue growth justifies 100x. It's the stability that I worry about. If it's stable, I'm all in. If it's 1 to 15 in 12 months, I'm all in.

That is fair, in the sense that—we talked about this last week—you have significantly more variance in product-market fit in these AI products than you saw in SaaS. I do believe, and I would still assert, that enterprise-grade, big installs with lots of integration will be way stickier than most. It'll also be slower to build than most.

How are you looking at it? Say it.

Guest 2

Well, AI is the greatest friend for verticalization. We led the Series A for a company called Solve. It's AI for patent lawyers. If you think patent lawyers are switching software tools often, you are high. It's a once-every-10-year switch. It's a difficult thing to do. There is no way the churn is what it is with horizontal developer audiences like it would be with Cursor or likely Codeium.

[Speaker?]

But I would argue—I hear you, and I agree—but I think there's a bit of VC old-school hubris here, which is that it's killed versus maimed. I think once you're embedded in a workflow, once you're core—it could be SMB, it could be mid-market—when you're core, it's hard to rip out. It takes time.

But what's happening with AI is people are looking more often. Deals are more competitive, and there's more pressure on pricing and downgrades. Anyone who says there's not, when some new AI competitor comes in and says, “We will do this at half the price, and it's 10 times better”—even if folks take a look—okay, everyone thinks their sales team is so great at resisting pricing pressures.

You know what happens when they cancel? They'll do the deal for half price. It's just that we're missing the fact that AI can maim leaders, even if it doesn't kill them, and that can take them off the IPO track. That can destroy venture investing if, instead of 50% at $500 million, you're growing 30% at $500 million because you got maimed. You didn't die, but, man, you no longer can IPO. That's terrible.

And that's where I think the people who are hiding their ostriches in the dirt—I think they're the startups—are going to fail, because they're not realizing they're getting these knife cuts, right?

I think it's more that the ostriches are hiding rather than people hiding their ostriches, but I get the metaphor.

So, that Decagon $15 million came from somewhere. It might have come from Intercom. It might have come from Zendesk. I understand the thesis, right? I hear your point, and I do agree that the bar, competition, and churn are significantly greater now because, like I said, I think the world is in flux. It was locked in for 15 years in SaaS land; it’s been in flux for the last 2 years and for the next 2 or 3 years in enterprise land.

This is where I could be wrong, but I’m just going to put it out there. I think the positive view is that a successful set of products in AI starts to gel over the next couple of years, and the people who are in the lead at that point in time get a similar 10-year run to the 10-year runs you and I both benefited from in SaaS, Jason. That risk of churn and that acute competitiveness—that 60 companies shake out to 3 or 4—and the end-market formation evolves in the same way as it did in the enterprise space, which is, typically, any enterprise apps marketplace tends to be a modest oligopoly of 3 or 4 players where you have steady market share. That’s the vision.

If I’m wrong in that vision, then these assets aren’t worth 10 times revenues; they’re only worth 5 times, and everyone is so horribly wrong my head hurts. I at least think it’s much riskier than I thought 100 days ago. Much riskier—not binary. I’m saying it’s much riskier that there isn’t this stable state at the 14th electron, or whatever it is, and that the stable state no longer exists. I don’t believe it exists anymore.

You know, I will say I’m lucky enough to be on one board with an executive—I won’t name him—who’s a very senior technologist in one of the model companies and really understands this. I just shut up and listen to him talk when he talks about model trajectory. His comment, over and over again, is that you just have to internalize what the models are going to do in the next 2 or 3 years, and you might be able to do that because it’s going to be done for you, right?

That is the argument on your side, Jason: the more the model can do, the more of that software stack gets sucked in. I do agree it’s a countervailing force. I don’t have clarity on it, but until you get a handle on that, you’re right—you are at risk of more disruption than we’ve seen in SaaS in 15 years.

We should be honest: portfolio companies and founders, if they’re seeing a little bit of elevation in churn, more pricing pressure on renewal, Decagon in a couple of deals, whatever it is, they should see this as a canary in a coal mine. Their CRO should not come to the board meeting and say, “Ah, it’s just a little. Yeah, we lost a couple, and we’re seeing a little pressure on downgrades.” This is not a bump. This is an exponential change in terms of risk, and I just think, if nothing else, people—maybe VCs will take the risk, but founders should jump on this. When you see a little bit of this start, you better be all over it.

I talked with Yaman Rangan from HubSpot last week. This is HubSpot. She said that at HubSpot, with Cursor, they are pushing out so many features they can’t put them into production anymore. She wasn’t kidding. They said they’re 50% more productive at HubSpot, which is a big effing deal. The fact that HubSpot is now developing more features than it can push out—think about that when you think you have a stable state in your 50-person startup, or that you can rest at $50 million in ARR.

HubSpot has more features than it can put into production for the first time ever, right? She’s not Dharmesh, but I’m sure it’s 100% accurate. He’s looking at it, right? He’s measuring this by code commits. There are more features than they can put into production, and they just can’t. It’s too much business-process change.

12. Olo Looking to Sell: What Happens When Public Companies Want to Sell

That stable HubSpot was so stable for years. “Oh, we’ll add CRM at $100 million.” I mean, Harry, you guys invested. It was, you know, a generational limit. “I’ll add Sales at $100 million, and I’ll add Service at $300 million, and I’ll just keep layering this beast, and I’ll drive NRR from 85% to 100% to 110%.” It was just this check-the-box approach, right? But if they can build more software than they can push out, what about everybody else?

Can I just touch on one final element before we—? You mentioned “kill or be killed” there, and we’ve talked about the companies that maybe derailed on the way to an IPO. Olo, the public company, is now for sale. The reverse of what we’re talking about is a company that needs to sell. How did we think about and analyze this one?

Guest 2

The thing about this, just so the readers know, is that Harry sends you 20 questions, you cover 18 of them, and then he picks the one that you didn’t cover. I don’t have a developed opinion on all of them.

I think I’d rather be SevenRooms, which DoorDash just bought for $1.2 billion. My advice to folks that are being eclipsed today is: take the offer. Take the effing offer. I’m not an expert in SevenRooms, but I think SevenRooms conceptually has the same challenge Olo does, which is that you’re focused on the enterprise end of an SMB market. Olo has a great founder trying to do big chains of restaurants, but restaurants are a VSB space, not SMB. SevenRooms is the same thing: complex reservation-management software for chains.

They got the $1.2 billion, and Olo didn’t. At some level, it’s true. It may not literally be true, so the only thing I can say is: take those offers. If you’re losing, this is always true in venture, but especially in these moments, take the deal.

The crazy announcement today was that Deliveroo is also getting bought by DoorDash for $2.9 billion. Just to put that in context for you guys—and I don’t mean that rudely, but I’m sure you’re not aware of public markets in the UK—it was valued at between $1.4 billion and $1.5 billion. That is a $1.4 billion delta between how DoorDash valued it and how UK public markets valued it.

It may well be, and I can say this having lived in the UK, that you guys are just crap at valuing tech companies. It’s a genuine comment. DoorDash is the machine. They’re the—I don’t know—$60 billion market-cap company. They’ve done the US. You get out the little map, you start coloring it, and you say, “Oh, Western Europe. We can pick this one up and be done.” You pay a premium, it’s in the noise, and you win.

It’s the beauty, going back to the point that once you have the US domestic tech market as your core starting point, you just end up with the biggest version of everything, except possibly something that’s domestic China. Then you can just pick off Europe, 1 acquisition at a time. Totally makes sense. Good for them.

We’re not going into China. I don’t want to upset anyone. No, we’re not going to China. There are other podcasts that will happily cover politics until we’re blue in the face.

Okay, final, final one. I do just have to ask it: Josh Kopelman’s Venture Arrogance School. What did you guys make of this?

Guest 2

I totally understood it. We run something—we wouldn’t call it the Venture Arrogance School—but it’s the right question everyone should ask: Is there enough market share for me to execute my business model? What do I have to achieve to achieve my business model?

You’re taking away the arrogance, which is just Josh being fun. I think Josh is amazing. Obviously, he’s done really well. Clearly, when you get toward the tail end, you can have a quick sneer at everyone, right? But the analysis itself was spot on. Every single firm should have to say to itself, “Are there enough deals of the size and stage I want to do to make the math work for me?”

Obviously, if you’re a $150 million seed firm, without even doing the math, you’re fine. There are lots of companies out there; you just have to make sure you find them. What he’s implicitly saying is that it gets back to where we started: if you have a $5 billion or $10 billion fund, what percentage of total value do you need to make the model work?

When I was preparing for this, that’s why I got that other analysis I talked about much earlier on the Cory Ratman stuff. No one has got that market share. The summary is: you take Josh’s analysis on what percentage of total value you need to make the math work, and then you take Cory’s historical analysis on what people have done in easier times. The conclusion is that no one has achieved the market share that it would require to make this math work for a venture investor.

You look at that and go, “That’s a sobering statistic.” I’m not saying the model doesn’t work because, as I said, the right answer is you won’t get there doing more Series A’s. The best firm got 10% of the Series A’s; the next best got 6%. Eight firms aren’t each going to get 10%.

The only way all those firms can, quote, make their model work is by stuffing huge amounts of money into late-stage deals. That gets back to the same thing every freaking week: if people stay private for longer, then you can own and compound these assets, probably at a lower return, but probably over the hurdle rate. That’s the bet.

I thought his comment on duration was amazing: 2x in 10 years is relatively similar, in terms of IRR, to 4x in 17. That was—wow—very much a realization.

Guest 2

Yes, time value of money is a bitch.

Yes, the analysis was spot-on. You should run that—you should know it for your firm. We can even go one level down: okay, this is what you need. How many of them do you see? How often do you get the picking right?

Because they’ve got to exist. Then you’ve got to multiply that by how many of them you see, then how many you pick. Are you doing a coverage play?

Guest 2

You know, they mention coverage. Everyone’s doing a coverage play. It’s just a question of what kind of coverage. Some people are trying to cover just the 10 best deals. Some people are trying to cover 100 deals to win them. But everyone, at some level, has to monitor some version of coverage to get there, because that’s what that whole analysis was about.

I think he said one thing—and then we can finish—but he said one thing I did disagree with. He said activity, in terms of deals, drives relevance. Of course he’s right, but it felt relatively binary, and it actually missed the fundamental reason why I do content, which is that content is the most effective way to stay relevant without having to put dollars out the door in deals that you maybe don’t want to do.

Guest 2

Yes. And even though talking with you isn’t fun, Harry, I’d prefer to talk to you than piss away $20 million.

There we go. So you see, Rory, even though you have to do this on a weekly basis, at least it’s better than pissing $20 million out the door on a Decagon at 100x.

Rory O’Driscoll

I’m not going to conflate those 2 things. I think Decagon is genuinely an amazing company, but I agree with you: pissing away? No.

Guest 2

No. And I think Josh is so shrewd. You’ve got to do something to be relevant, and it gets back to the thing: as long as these firms have a lot of dry powder, they can do a lot of deals. The person who does 10 good deals a year struggles to be relevant versus the person who has 100 good deals a year.

Josh is exactly right. We’ve all got to pick our way to win in this market, because it isn’t going away soon. This is the game on the field right now. It might not be the game that is a stable, long-term equilibrium. We might find, 10 years later, that some of this was a horrible mistake and some of this money gets withdrawn, but it’s the game on the field for the next 5 years. So quit bitching and play it. And now I’ve got to go chase a deal.