Elon 的帝国：风暴后的 SpaceX、Tesla、Neuralink，以及 Anduril 的26亿美元强势出击

Rory O'Driscoll

How freaking awesome the U.S. capitalist system is. I mean, the United States has 4% of the world's people, roughly 23% of the world's GDP, and 67% of the world's market cap. We won. Not just our GDP—our income is higher than our population ratio, but our wealth and our corporate sector are even higher than our GDP.

Some companies that were once worth $1 billion can go to zero. The definition of a great business is when your customers can hate you and still do business with you. The truth is, SpaceX is just such an amazing achievement that even if one of your biggest customers doesn't much like you, they still have to do business with you. That's not quite as true for Tesla.

Guys, I'm so excited for this. I love this time more than any other in the week. I want to start with IPOs. We were talking about it just before, and it is the most important topic for us to discuss.

1. Circle’s IPO: Investors Just Left $BNs on the Table

I want to start with Circle, the strongest IPO since 2020 and, bluntly, much-needed positivity for the ecosystem in terms of the public-market response. How do we analyze the response to Circle being so well received?

Rory O'Driscoll

It's a super interesting transaction in the sense that we've gone from, "Oh, IPOs are hard, nothing's happening," all the way to, "Oh my God, we left so much money on the table," without the intervening period of gratitude for the IPO. We literally went from "the window was shut" 4 weeks ago to "the window was open" to "Oh my God, we underpriced this thing," without the intervening period of gratitude for the IPO.

In the case of Circle, the data says they filed, raised the range, and then it opened at almost 2.5x the IPO price. First of all, great success and an amazing outcome. It was a good company at the IPO price; it's an amazing outcome at the current price. Even if it's not sustainable, I couldn't be happier for everyone involved.

I think the interesting question will then become whether the underpricing issue will be even more acute here than normal. Normally, when you have these IPOs and there's a planned 15% pop but instead there's a 40% pop, everyone's kind of miffed, but the money goes to the company. You have $100 million less than you thought you would. In this case, over half of the IPO was sellers, which means you opted to sell a security—I can't remember, but I think it was $31 a share—and 2 days later it's trading at $80 a share.

That's a lot of money to leave on the table. If you chose to sell in the IPO, you're sitting back and going, "I feel amazing about the outcome, but oh my God, that's a lot of money on the table." I think 20 million shares were secondary, and at $50 or $60 of additional value per share, that's $1 billion that went to the buyers, not the sellers. So that's a fun one.

All the recent IPOs except for SailPoint are up materially. On average, they're up 76.8%. Averages are confusing and misleading, but I thought it was interesting that Mountain Hinge Health and I guess eToro, which I know less about, are not hyped or high-hype stocks. Those are up substantially.

It's just Circle and CoreWeave that are the crazy ones, and they're at the edge of meme stocks, right? CoreWeave is an attempt to invest in AI, which is on fire, and Circle is an attempt to invest in crypto when it's an exciting company, but it has so much interest-rate sensitivity. I think these are meme stocks, and I don't know what Bill Gurley thinks about meme stocks, but I suspect that at the start of a run of IPOs, you have to be careful with the meme stocks, and toward the end of the bull run, you price the hell out of them. Rory and Harry, you may have more thoughts.

I totally agree with what you're saying. I would say, in the interest of being—I don't know, maybe pedantic—they're good companies that have meme value as well. I think something like GameStop was just a meme stock. Both CoreWeave and Circle are exceptionally good companies in big-ass, meaningful industries, but you're right: on top of that, they have meme value, which makes pricing hard.

I think it's a great point, Jason. The other 3 companies are just solid, boring companies doing great. They did exactly what they're meant to do, had that little 20% pop, and are trading nicely. Everybody's happy, right? These guys really ran away from them.

The fun thing is, you can say that, but the interesting question is: What can you do about those kinds of trades? Bill Gurley has been very vocal: "Oh my God, you left all this money on the table." How do you avoid giving up the pop?

The first point is that some pop is necessary, because these stocks haven't been traded. You're asking people to step up and write a check, and there's no prior pricing information, so you have to get paid something for the volatility you're incurring—the risk of a 1-day loss. You inevitably start off in the IPO structure, unlike the direct-listing structure, having to give some kind of pop.

Then the judgment comes down to whether you're overestimating or underestimating the real retail demand, and whether you're underestimating institutional demand. Obviously, in these cases where there's strong retail demand, you've ended up leaving a lot of money on the table.

The fun question, and this is right down in the weeds, is that the bankers will say, "If we didn't take these anchor investors at this lower price, the retail demand wouldn't be there," right? I sat in the room and had those discussions. One party thinks that sounds plausible and correct, and another party says, "But that's a total violation of the efficient-market hypothesis, and I just don't believe it," right?

What happens in the end on these deals—and this is a concrete example of very intelligent investors leaving money on the table—is that the biggest advantage the banker has is that you do this once in your life, or maybe 10 times if you're a VC, and a banker is doing it every week. You have an informational asymmetry there.

They're going to tell you, "Hey, at $31, you get Fidelity and T. Rowe, but at $35, you only get a bunch of hedge funds, and they're going to sell it and flip it." So you opt for $31, and then the really frustrating thing is that it pops to $70, and Fidelity and T. Rowe flip it. You feel like a sucker, but then you go all the way back to your business, and they get on to the next thing the following week.

So it's a very problematic structure, but it's also worth pointing out that the other alternatives don't work that well. At various times, I think SPACs have been a disaster, right? A direct listing only works when you're an amazing company and you're not raising primary capital, because of the regulations—which arguably has not been the case here.

Going back, there were a couple of companies that did those kinds of Dutch auctions, including Google, and they seemed to work, even though Google on the day was a little troubling because it underperformed early on. Obviously, it has done amazingly since.

I'm just curious. It's a niche topic, but I think it's interesting for the future because I think this is going to encourage everyone to go public after this strong performance.

Chime may well be public by the time people listen to this. We'll see the timing, but I saw in the media that they were saying it's going to be a banger IPO because it's 10x oversubscribed. In my limited experience, 10x actually isn't enough, because those are soft commitments and people put in overallocations to make sure they can get it.

2. 50% of Unicorns Are DOOMED. What Happens Now?

You really want to be at 30x oversubscribed to pop hard. I don't know if that's what you've seen with your public companies, but that's tough to get right at the start of an IPO market. How much does that multiple need to be—10x, 20x, 30x? You certainly don't want it to flop, right?

Rory O'Driscoll

Yeah. I mean, yes, you do. 5x is not enough. Five times is not enough. The banks will tell you that 5x isn't enough.

All those 10x, 5x, and 20x oversubscription figures are just so bogus because you've got this game-theory thing where buyers are putting in bigger orders than they actually want because they don't think they're going to get cut back. The demand is entirely theoretical.

The truth is, you have all these relationship bankers. They know all about the business and the story. They've been calling on the company for 3 years and have a relationship with the CEO and CFO, and none of it matters a damn the night before. You're sitting there, and then some person from Equity Capital Markets crawls out of a hole in New York and says, "Here's the big, long list, but this guy's lying. He won't flip. This guy will flip. Sorry, he will flip. This guy won't flip. This guy doesn't really want $10 million; he only wants $2 million. He's pretending to put in $10 million."

The entire decision gets made by someone you spend a little time with, but not as much, on the basis of things you don't quite understand. It's a wildly frustrating process, which is why Bill Gurley is right to be angry about it.

The hard thing is: What do you do better? The very biggest companies, when Stripe finally goes public, can do whatever they want. They can do a direct listing and not raise any primary capital. They could do the Google-type auction, and no matter what, it'll all be fine, right?

Yeah, but the median company—the typical company—is looking to raise primary capital and can't afford to get it wrong.

[Speaker?]

This is a one-time debut, right? You probably aren't so strong that you could power through. If Google had failed or Stripe were to fail, everyone would say it's not about Stripe; it's about the market being weird that week. If your little $2 billion market-cap company doesn't get done, your deal didn't get done. So, as a team, you're in an intrinsically weaker position.

3. Should Stripe and Databricks Finally Go Public?

You end up being discouraged from any innovative process, and it's worth pointing out that most of the innovative processes, like SPACs, failed anyway. So you're back to the time-honored: build a book, raise $200 million in primary capital, deal with the informational asymmetry, do your best, and then, some days—especially on the meme stocks—you're randomly wrong. It's a wildly flawed process for which we can find none better.

The question for me is, does this very positive response across the board with the IPOs that we've seen lead to the window opening more? I think we all agree it does. Does it lead to the window opening to Databricks, to SpaceX, to the biggest companies of today? Then we saw Figma confidentially file for an IPO, and I thought, how do we think about those 2 questions? Does it lead to the juggernauts, and what do we think about Figma?

T. Rowe

Look, the window was and always has been open for Databricks and Stripe. They just don't want to go through the window, right? There's no—the window's open and shut for the $2 billion market-cap IPO. The window is always open for the $50 billion market-cap IPO. As someone said to me years ago, there's always room at the top, right?

Their decision not to go public is an entirely separate choice about what they think they can do privately versus publicly. Figma's maybe the low end of really amazing, like Stripe, but the high end of more than amazing. I think that's, to me, more of the normal. The window's open for companies, especially—I would say especially if you nearly had a liquidity event 2 years ago, right, in terms of an M&A sale, and obviously you lost that.

All credit to them for regrouping from that and continuing to build. It must have been a very difficult management challenge to pick yourself up having decided to sell to Adobe for $20 billion, not getting the deal done, having to get some capital, and doing a secondary at $10 billion. My guess is it totally makes sense for them to get a great IPO under their belt in a way that—and not do the Stripe thing of staying private for another 3 or 4 years. So, I think that totally makes sense for them.

If you nearly had the win and it was taken away from you right at the last minute, you're like, “This time, I'm getting my freaking win, right? I am posting the IPO. I'm ringing the bell. I'm declaring victory.”

Rory O'Driscoll

The only asterisk and dagger I would add is that these decisions, especially when you get a bunch of VCs and other large shareholders, they're trying to guess all of this. When everyone's caught up in feeling this is the time to IPO, I think people are going to try to go.

Even if they could have 6 months ago—even if the data shows, hey, Mountain Hinge Health, Circle, CoreWeave, eToro, even SailPoint, all could have IPOed last year—the window was sort of open. I just think when everything's trading up, every meeting starts to be about, “Should we go public now?” Add in a good IPO from Chime, and once everyone starts talking about it, you kind of convince yourself and start having those conversations: well, it's time. It'll give us more rigor. We'll get the biggest deal done. When people are in the middle, on the edge, they just go forward in this environment.

Agreed. I also like your comment, Jason, about how the stuff just changes in months. You mentioned CoreWeave, a wildly successful IPO. Interestingly, that was one that had to come down from the filing range as recently as 4 or 5 months ago. In other words, 5 months ago, the bankers tell you you're at X; when it gets closer to the day and we actually find out what the end investors are willing to pay, you're reducing that range. Fast-forward 4 months later, and it's 2x up from the IPO, right?

These high-growth companies are hard to value. Sentiment matters a lot—overall market sentiment and specific sentiment about the deals—and, to some extent, the only way you find out is by putting them out there and seeing what the demand is. I think pretending you have this a priori knowledge of how these things are going to trade, or when the window is going to open, is impossible. You just have to internalize that it's so random and so outside your control.

You have to build the company, and once you're at the stage where you could go public and conceptually think you want to go public, you should do all the preparations to get ready and then just accept the fact that the timing, to some extent, is a bit out of your control.

T. Rowe

Yeah, you know, the other interesting thing when CoreWeave did its IPO and we were talking about it is that not only did it—I mean, CoreWeave, you've got the narrative—not only was it difficult to get done, but they had to reduce the range. We almost made fun of this existential risk they had with the debt, which was this clock ticking down on them, this debt they had to repay.

Because the IPO was so strong, they were able to raise $2 million of additional debt and completely derisk the company. That existential risk is, for all intents and purposes, gone. Maybe you don't need the money like Stripe, but if you can not only IPO in this market and trade up, but then do things to derisk the company even further, like CoreWeave, that's a triple hat trick they got, right? They went from a company teetering on the edge of not being able to repay its debt to being set for the better part of a decade.

Rory O'Driscoll

Agreed on that. Two comments. One is, what it highlights, Jason, is that the public markets in the United States are pretty damn amazing. You can access large amounts of capital in short periods of time, which is why I believe the whole “stay private unless you're cash-flow positive” thing is wrong. If you're the kind of company that needs to raise capital at scale, the dominant, best, and most cost-efficient way to raise capital is in the public markets, by being public and with debt, right? CoreWeave's proven that.

The other thing, just to put it out there, is that it proves how little I know—not to say I think anyone else knows much more. I mean, if we all had great opinions on CoreWeave, there was a 2.5x on the table in 4 months. You could have bought all the CoreWeave your little heart desired, held it for 2 or 3 months, and been 2.5x up. I didn't buy any. What kind of idiot am I, right?

4. US Stock Markets: How They DOMINATE the Global Game

So, again, where we rank on the omniscience factor is probably a 2 or a 3 out of 10, and just internalize that. That's much easier than going from seed to Series A. My God, we can talk about that later.

Rory, I'm going to tee you up on this one. I'm teeing you up so nicely. We've had Deliveroo bought by the Americans and taken off the London Stock Exchange. This week, we had Wise, otherwise known as TransferWise, announce that they're also going to list in the US. Bluntly, a pretty big hammer blow to the London Stock Exchange.

Rory O'Driscoll

Look, it would be fun to make this a ding on the Brits because I'm Irish and we always want an excuse to ding on the Brit, but it's not. The salient point is, it's not about you being bad. It's just about, again, going back to how freaking awesome the US capitalist system is.

The big sound bite on this is that the United States has 4% of the world's people. We have roughly 23% of the world's GDP and, depending on the day, 67% of the world's market cap on the stock exchange. We won. It's so funny. US corporations are efficient, they're highly valued, and they have international businesses. Not just our GDP—our income—is higher than our population ratio, but our wealth, our corporate sector, is even higher than our GDP when you look at those kinds of figures.

A couple of things come from that and probably 2 big conclusions. One is about Wise, but the first one, just to put it out there, is that this is an amazing place to make money. When you look at all the “Oh my God, things are awful,” when you look at those numbers, whatever this system is, whatever this economic order is, it's been pretty damn good for America for 50 years. Let me repeat: 4% of the population, 24% of the income, 67% of the wealth. Yay, us.

Now to the Wise thing. You're exactly right, and you just look at that and go, there's just way more people who want to buy my stock when it's listed in the States. It was interesting that when they announced it, the stock popped 8%. That's basically like saying you can make 8% free money just by listing in the US.

Fundamentally, if you want to list your stock in a market, you want to be trading on the biggest, most liquid market, and that's the United States. Unless you have structural legal reasons, like Chinese companies where you can't be here, or unless you're purely a domestic company—maybe if you're just purely a domestic company. I don't think NatWest, or whichever bank survived the Great Crash over in England, is going to list in the United States. But for big public companies with an international business, why wouldn't you go where 70% of the market cap is and just join the team?

Yeah. The other thing is, I don't know how big some of these will be. Sometimes, Rory, I don't know if you've seen your public companies, but as lay folks, we overestimate how much liquidity there is for a lot of tech stocks. All but the biggest ones are relatively thinly traded, right? Relatively thinly traded.

Especially if you do a smaller IPO like Mountain Hinge Health or Hinge Health, you might be surprised just how close you are to the edge of liquidity. So why wouldn’t you? I wouldn’t want to do anything but the US if I was at the edge of that, right? We overstate the liquidity that’s out there.

Rory O'Driscoll

That’s a good point, because I thought where you were going was, “Oh, it’s not great,” but you’re exactly right: if it’s mediocre—and it is often at the $2 billion and $3 billion level—you just don’t have a chance anywhere else. You’re exactly right that there’s no analyst coverage. The institutional buyers are not there in the single-digit billions.

Right. And it’s just—you better be wherever there is any liquidity. Well, should those companies be public, though? Those $2 billion-to-$5 billion companies where there’s a very thin layer of liquidity, should they even be public?

Rory O'Driscoll

I was with the founder of one of them last week, and he was like, “No, we shouldn’t be. That’s the point. We shouldn’t be public.”

At some point, again, it’s—look, we’ve had this. At some point, you’re going to want to be, because even the thin liquidity of a public market is better than the liquidity in the private markets, which is even worse. Now, yes, if you’re amazing, you can access capital, right? And there are some arguments for founders for staying private.

But, look, as an employee, would you prefer to work for a company where your equity compensation can be earned over time but can only be accessed once or twice a year, to some extent with the approval of management and depending on the specific market on that day? Or would you prefer to work for a company where your equity compensation is freely tradable every day of the year?

I think it’s pretty obvious when you ask that question. So there are trade-offs once you get to some kind of scale, and the trade-offs are different for different kinds of companies depending on how long you can stay private. But in the end, successful big companies are, in the main, going to tend towards an IPO.

There are 1,500 unicorns today. That’s what Crunchbase says. Close enough, right? Rich Wong from Accel said this last week: 20% will fail. Okay, so that leaves us with what? 1,200.

How many can easily do tender offers of scale? 1%? 2%? We’ve got 1,200 viable unicorns, many growing at abysmal rates. But if they don’t have a shot at a small IPO, how many of these companies are there? 15 or 20 that can do these tender offers, right? Jason, can I just understand: what does he mean by fail?

Rory O'Driscoll

Fail to be a billion-dollar company.

Rich Wong from Accel said 20% of unicorns this week will just fail. They’re not going to limp along; they’re just going to fail.

Rory O'Driscoll

And that’s true. I think the interesting thing about the SVB report you cited is the other end of the distribution. I mean, this is pretty rough and tough, but my gut would be that roughly 20% will be good enough to get public in the end. Maybe 25% on each side, just for arbitrary sake, and then 50% in the middle are meaningful enough to be valuable but not so meaningful that they’ll go public. They’ll be merged, they’ll combine with someone else, or they’ll be PE. But it’s a distribution. It’s probably a distribution something like that.

Maybe the low-end tail is not 20%; maybe it’s 30%. But I think broadly speaking, it’s correct. Some companies that were once worth a billion dollars can go to zero—easily done—especially if you have debt, especially if you have a high cost structure, and if you’re asleep at the switch. Many companies that were valued north of a billion can flatten out and struggle to get $500 million or $600 million but will realize some value. The percentage that will actually get public, my guess is it’s 20%, plus or minus, maybe even less.

So, if it’s 1,500, I mean, that’s still 300 IPOs. That’s a lot. And it’s only—that’s the key point—and it’s only that last 10% or 20% who have “IPO potential,” who could get a tender done.

And what that says, therefore, is that, for most, the equity compensation for a lot of these companies is notional and not accessible, right? I think there’s also one thing we’re definitely seeing now: as the holding period goes on, there’s not just the ability to do tenders; there’s a need to do tenders.

You can’t tell people, “Sign up at 25 to join the startup. It used to be: work really crazy hours for 4 years, we’ll go public, you’ll ring the bell, and you’ll do really well as an employee. It’s amazing.” Now it’s like, “Sign up for 4 years,” that turns into 12 years, and then at the end we’re still trying to put a tender offer together. That’s just less compelling.

Frankly, people have lives. They need to move on, they need to buy houses, and they want to start families. So the need to just get liquidity to solve employee-compensation problems becomes acute. I think most can’t do tenders. If you can do a tender, I think people are starting to do that.

But at some point, you’re going to say to yourself, “I’m doing tenders as well. I’m not Stripe, where I can do them on demand. Should I just ultimately access the public markets?” I’m increasingly impressed by the liquidity mechanisms and solutions for large-scale private companies today that I see consistently traded.

Unless you’re in the elite, that may fall apart next year. It may not exist next year if they don’t hit the growth targets. There’s always lots of liquidity these days around a financing event, right? And there’s often one more. I don’t know what you’re seeing, Harry. I think if your growth struggles, it instantly evaporates in certain cases.

Absolutely. Or if there’s—I mean, honestly, if I was a top-tier engineer, or almost anyone that wanted to have the comfort of a later-stage company, I might only join someone with a perfected tender-offer program. I wouldn’t even bother with anything else. Why join any other? That might be my first question in the interview: “How does your tender-offer process work?”

Nah, seriously. Look, by definition, if you have more risk, you better have more return. If there are 2 private companies and one of them is among the small number of entities that have this monetizable stock, and you don’t have monetizable stock, there’s a premium for liquidity.

I think you’ll see that in terms of what it takes to attract people, which is yet another reason—I know I’m sounding like a broken record now—why, as you get larger in scale, it’s just going to be more efficient not to be doing these tender offers. Going back to once you hit critical mass, wouldn’t it be a lot easier just to go public?

5. Founders Fund Just Dropped $1B on Anduril. Why?!

While we’re speaking of financings, obviously there’s going public; there are also later-stage rounds and great companies raising later-stage rounds. We’ve discussed before the value of capital concentration, going very, very long in your best companies.

I didn’t realize that, specifically with regard to Anduril raising $2.5 billion in its Series G. I don’t know if you guys knew this: they got $1 billion from Founders Fund in this latest round, again making it their largest-ever check. Founders Fund’s prior largest-ever check was also to Anduril.

Sam [likely Lessin] said it was the only really important company he could think of beyond OpenAI. I found it par for the course, honestly. In all seriousness, after Sam’s insights, I thought, “Hey, dude’s right.” He did point out that it was the only one he said, and Anduril was the one.

I’m sort of joking, but honestly, my learning from that comment he made is that it’s probably an even better company than I realize.

Rory O'Driscoll

Yeah. I mean, it’s the old rule: just because you think someone’s being a jerk doesn’t mean they’re not right, right?

I think Anduril—look, so you ask how I’m surprised? No, I’m not surprised, because all credit to Founders Fund, they told you to do this back when they started the firm. They said, “We’re going to have highly concentrated bets in our best industries and best companies.” Tick.

They’ve also had, again, to be extraordinary—I mean, to give all credit to them—the theme of national-security investing. They co-founded Palantir in, I think, 2003 or 2004. It’s not like they came Johnny-come-lately to this space and started doing it recently and are rushing in to catch up at the late stage. They’ve had this thesis for 20 years.

They founded the company, and they’ve said very clearly, “We’re going to double down and triple down on our biggest bets.” Overlay on top of that, there’s clearly not just a purpose in terms of capital, but there’s clearly a felt purpose in terms of national security that animates the principals in that fund. I respect that. I really do.

I think Anduril is an amazing company, not our focus area. But when you put all that together, and you’re also, let’s be honest, at the Maslow’s hierarchy stage where minimizing risk for your investors is not your 1, 2, 3, or 4 priority, you probably sit there and say, “What do I want to do with my life? I want to give as much money as I can to the company I love the most that’s doing the cause I believe in the most, which is defending the Western world.”

I mean, right back to—and then you cite your Lord of the Rings and you’re done. So, yeah, of course they push it across the table.

Plus, I like that they’re putting—I mean, they’re putting $1 billion in, right? That’s much more ambitious than a growth fund throwing $50 million into the last round of Anthropic, right? Or throwing $20 million into Lovable. It’s just a big bet, right? Again, yeah. Rory, what percentage of capital concentration would be the peak of what you would be comfortable with in a fund?

Rory O'Driscoll

We have a capital-concentration limit of 10%. We typically aim to fund 20 deals, on average at 5%, with a fairly low standard deviation.

So I’d have guessed 7%. I do. Yeah. Do you think that’s what Brian Singerman has said to me before?

Rory O'Driscoll

I’m going to answer exactly that question. We typically haven’t done a whole ton of later-stage follow-on, even in our best investments. Our focus has been early dollars at work, maximizing multiple. I think the more late-stage you go, the bigger the—ironically, this is a really weird comment—the bigger the fund you raise, the more late-stage you’re going by definition.

And oddly enough, the more concentrated you have to make the bets, which is counterintuitive, right? Because there are lots of things valued at $100 million that might go to $1 billion. There are very few things that are $10 billion that might go to $30 billion. So when you find one, you’ve got to put a lot in.

At the stage we play at, it hasn’t been necessary to have that level of concentration. Our limits have been more than fine. We’ve been in that 5% to 10% range. I think to do what Founders Fund said 20 years ago they were going to do—and have done, to be fair to them—you have to have a much higher ability to concentrate, to put that $1 billion or $2 billion to work in late stage.

It’s not been our business, but I think the people who are doing it are doing it right. I mean, if they put $1 billion into Anduril and they really believe it can be worth $100 billion, the partners themselves just make a billion dollars off that one bet. This is an incremental check, but that extra billion, if that goes from $13 billion to $100 billion and they keep 20%, the partners clear a billion dollars personally off a one-day decision.

Honestly, I think the most interesting question to ask managers today is: What would you do if your LPs would let you do anything? And I fundamentally think, if you have a $400 million fund, you put in OpenAI, Anthropic, Cursor, and Anduril.

I totally agree with you that Lockheed Martin has a $150 billion market cap. Doing this at $40 billion, you can see a 3.5x in this being the next generation of Lockheed. I think the truth is, it’s an interesting question: What would you do if you could do anything?

6. What Would Rory Do If LPs Let You Go Wild?

The probable answer would be just like if you told your teenage kids they could do anything. Someone would do amazing, but a lot of them would go off the rails. And the question as a parent is, would you want to sign up for that? Right?

You’re such a smart strategist of venture. What would you do if your LPs would let you do anything?

Rory O'Driscoll

I found this job hard when I started, and it’s hard to be good at one thing. I would say I probably have a conservative bias to keep doing what we’re doing and doing it well, versus trying to do lots of different things.

I think the more you spread, the more you widen your aperture. Yes, the more upside you have, but the more risk you take on. I’m probably—I absolutely am—more risk-averse at the margin and less willing to sign up for trying to do everything, versus sticking to a strategy that you know works.

It’s the “know what’s in the box” thing: know what’s in the box that you can do and understand the limit of the box. We do a really nice job on every dimension of seeing these early-in-revenue enterprise software companies looking to scale and investing in them at As, Bs, and sometimes Cs.

I think trying to go for that, especially as a firm, and now saying, “I’m going to put $100 million or $200 million into something else,” it’s just harder to do. It’s not a constraint from the LP as much as it’s harder to do.

That said, I do believe, as I reflect back, that a little more standard deviation in your bet sizing, probably for the market we’re in now, might be appropriate. I think I’ve said this on the program before. I’m wrestling with that and thinking about that.

You don’t want to go hog-wild. You don’t want to lose your discipline. You don’t want to deliver the product and the consistency you have. But at the same time, if the market has moved, if staying private for longer has consequences, you’ve got to think about what that means for your bet concentration.

So it’s up for grabs; I’m thinking about it. But I don’t feel it’s a pure—I know this is—I don’t feel it’s a pure, “Oh my God, they won’t let me.” I don’t subscribe to the “they won’t let me do something” school of life. I subscribe much more to the “I can do this well and it works” school.

I often say to folks, look, every year you want to build a better firm, right? Every year, when you go to your off-site, you step back and say, “What can we do to be better this year?” We have lots of ambitious, driven people, and they’ll say, “We can do this, this, and this.”

I’m also the person who says, “Yes, and how do we do this and this without losing what we’ve got? Because what we’ve got is pretty good,” right? I think there’s that constant tension. So it’s much less about, “Oh, they won’t let us do this,” and much more about, if you’re doing these things and suddenly say, “I’m also going to do seed, and I’m also going to do tons of late-stage deals,” it’s possible you can be brilliant at everything, but it just gets harder.

The bizarre thing about Founders Fund is they appear to be brilliant at everything. I give them all credit. I just have to say, I’m not sure I could be that brilliant. You have to go back to, “Wow, they’ve executed.” They’ve demonstrated a range of investment acumen across a range of different challenges. That’s just very impressive.

Back to your hypothetical LP question: if everyone was set free to do whatever they want, I think for the median firm it would be value-destructive. And that was a long-winded answer. Sorry, because it was an interesting question that I hadn’t had posed before. It’s one that I think of a lot.

It’s a good one. It’s a really good one. And most actually push the boundaries with a lot more, as I said. Fundamentally, if our job is to make money quickly for LPs and be conscious of that, late stage, if you can get access, bluntly, is—let me tell you what my answer is, for what it’s worth.

If I could do what I wanted and offset the risk into other vehicles, into SPVs, into the ether, then I get that the Anduril–OpenAI thing is the smartest play. But what I personally would do would be to do every dollar up to $1 billion. That would be my version of it in every winner.

Rory O'Driscoll

So what do you mean, every dollar up to $1 billion?

If you do a seed investment and it turns out to be a winner—a top 5% deal—you do every round, A, B, C, D, and E, until $1 billion, and then you stop. You buy, you maximize the ownership, and you maximize the capital in until $1 billion.

Why not, instead of putting $5 million or $7 million in, put $100 million in on the way to $1 billion or $150 million? But still have the benefits of starting with the seed fund.

Rory O'Driscoll

Maybe you have 3 funds, right? And you stack them. I mean, Harry, you have 2. That’s what I would find a way to do if I didn’t have to worry about some of the risk of doing it.

The interesting thing about the hypothesis—“Oh, I’d double down on my winners all the way”—is that it sounds plausible, but it’s interesting when you actually run the math, because there’s not as big an opportunity to stuff money into even most late, quote-unquote, winners as you’d think, right?

If you think about our likely distribution, we do 20 deals. Probably, right, 30% of them fail. You don’t want to put a dime in those. Fifty percent of them are a 1–5x on the money you put in, which means, by definition, the next round at 2x is a 2.5x or less, right? Not compelling.

Only 20% of the deals in any fund, if you’re doing As and Bs, are going to be amazing, right? So now you’re down to only 4 deals out of the 20 where you can, quote-unquote, stuff money into. Remember, “amazing” is a 10x.

What? But still 4. Four is not zero. It’s 4. Four is a 4. That could be $400 million right there. It’s still 4.

Rory O'Driscoll

But yes, it could be. And remember, what was amazing to you at the A and the B might be amazing on that last round. So probably, of those 4, let’s say the 4 of them are 10x-plus-type returns from the A and B prices. By definition, from the C and D prices, they’re going to be 3x and 4x returns.

It’s only if you have the 1 or 2 amazing compounding winners that you can stuff big money in. When you raise that core late-stage opportunity fund, one of a couple of things happens.

Either, A, you right-size it so that it’s a relatively small percentage of your core fund, because there are actually not that many opportunities. The second thing that happens—but people don’t do that—the second thing you do is say, “I don’t have enough money in my portfolio. I’m just now going to do general growth investing. I’m just going to find other deals that do it,” right?

Or the third is—and I think people who are doing it are doing it well—you just ratchet up the N-count, the number of deals at Series A and B, because you’re basically saying, “To make the math on my overall thing work, I need to have not just a really good $1–$5 billion outcome, but a freaking amazing $5-plus-billion outcome.”

The only way to do that—you can say you can do it with great picking, but we’ve discussed that over and over again. You can, but it’s maybe at most 1 per fund. But if you triple the number of at-bats, then you probably roughly 2.5x the chance of being able to move big money at the late stage.

7. What Missing Out on Millions for Docusign Taught Rory

So there’s a whole series of things you’re driven to do once you adopt this: “Oh, I’m going to stuff a load of money into my late-stage deals.” It’s not as simple. The sound bite always looks good in retrospect: “Look, if I knew DocuSign was going to compound to where it did, I would have done the round at $19 a share, too.” I didn’t, right? We did the round at $1 and $2, right? But those outcomes are few and far between, and trying to pull that strategy off in the quote-unquote typical portfolio is really hard.

When you reflect, Rory, on that, could you have known? Is there a lesson that you take from that? “I did it at a buck, I did it at $2, and it’s now at $19.” Is that—“We were at $2, it’s now at $80, and I passed, and I didn’t do a lot of the round at $19”—is there a lesson you can draw?

Rory O'Driscoll

The problem with trying to reflect on that is that the lesson on the good outcome is always that you could have done more. The only way you can be intellectually honest is by asking, “What other deals did I have that looked equally promising that, in retrospect, you shouldn’t have done the round at $19?” Right? Because everyone does this. In retrospect, I wish I’d put more money in my winners. Duh. It’s not an insight; it’s obvious.

The question is, can you put enough money in your winners to move the needle without putting enough money in your losers to drag down your return? That’s the challenge. There are 5 or 7 amazing companies in the last decade. We know the names, right? If you end up at one of those, you probably can stuff to your heart’s content.

Chime is a good example. It looked like something you could stuff to your heart’s content, but it turned out that the price wasn’t $25; it was $12, $15-ish, and stuffing didn’t work, right?

If you look at the DocuSign and even the EchoSign–Adobe Sign lesson, looking back, having been there, I think the venture lesson is: be visionary but relentlessly honest about TAM, because it’s a TAM story at the end of the day. If you have a number-one or number-two player in the space and you see a TAM explosion happening, that’s where you get a big lift, right? It’s hard to know which ones are going to be your best and how much they’re going to run.

I also think it’s probably a little bit of a fake signal in 2021, because even your okay ones got highly valued. So you probably think it’s easier. Late stage looked a lot easier in 2021 than it probably will look across a decade.

Yeah. With my first fund, after 18 months, I predicted my top 5 fund returners. You had your Hopins, your BeReals, your Clubhouses, and none of the 5 outperformers ended up being the outperformers. The 5 outperformers I always had in the middle bucket. Actually, they were always in the mid-tier. It was Roger [likely Ehrenberg] at IIA who said, “That is exactly the same as me.” And I see exactly that in our portfolios. It’s an interesting fact because—

Rory O’Driscoll

Actually, we have a different experience because we’re just slightly later stage. I think at seed, that’s totally true, right? At seed, you almost know nothing. At the A and B stage, what we’ve observed is this: our mental model is a 20% chance of a 5x-plus outcome.

If, after 2 years, the company has done what we said it would do, roughly, in terms of performance—in other words, it’s ramped—the probability of it being a 5x-plus outcome goes up to around 60% or 70%. And that’s just because we’re slightly later stage. We’re paying for companies after they have product-market fit and are looking to scale. If, in fact, they scale, your probability of a strong outcome goes up a lot. If they don’t scale, it’s obviously a lot harder. So our data comes back quicker than yours because it’s a little further along, right?

But it’s still what we remain uncertain of. To Jason’s point, the next question is: you’re at seed, and you’re like, “Does it even work at the stage we’re at? Is it a decent business that can grow fast?” Then the later-stage question is, “How big can this be?” which is a TAM question, as Jason pointed out, and how will it ultimately be valued by the public markets, right? That’s the thing that, 2 or 3 years in, we don’t know. We’re all on the same journey at different stages.

Yeah. Be good at the thing you have, but then internalize that at the outer edges, there are things that just aren’t knowable as easily, even when you win the deal. I think an important point that you’ve pushed back on me before, Rory, you said, “Oh, I’d still take these companies, but how fast can you grow, and do financing providers find that rate still attractive?” You’ve said before: the double-double, treble-treble.

But I met a company in vertical SaaS the other day, and I was talking to Jason about it. They’ll scale from $1 million to $7 million over the next 3 years. It’s not attractive. No one’s going to touch that.

Rory O'Driscoll

Well, I would change that word, because you’re beginning to sound like our recent friend. It is attractive for them, and that’s the most important thing. It’s a great entrepreneurial opportunity. It’s just not compelling for our business model.

I think it’s an important thing. I was actually reflecting on the conversation last week: you never want to diss the entrepreneur. Most businesses aren’t venture-fundable. And, to the conversation last week, most venture-fundable deals don’t turn out to be the most important company of the last decade, because by definition only 1 company can be the most important company of the last decade. That doesn’t mean you piss on the entrepreneur on the journey.

I love the fact that this company that’s going from $1 million to $7 million over 3 years—great for him. He probably has a great business. It’s just not our business. That’s all.

So I’m just being a little kind to the entrepreneur. I would respectfully push back and say I’m not pissing on the entrepreneur. I’m saying if you take money with that growth rate, I think there will be an impatience from your vantage that will make your life hard. I want to save you that pain.

Rory O'Driscoll

I think if he said, “That’s not attractive for us,” that would have been a fine statement, right? Again, it’s back to this: “Cloud’s no good” isn’t a helpful thing. It’s like, “That’s a really good business for you, but it would not be a good business for us as venture people.” That way, you’re not downing their journey, but at the same time you’re saying it doesn’t suit our model. It’s not you, it’s me.

But I might do that deal. For what it’s worth, if I thought the founders were incredible, I thought the true TAM was large, and I thought there was upside at growth scale beyond $10 million—and I hate over-discussing price, but in this case, if the price was commensurate with that bet, which was much easier to do a few years ago—I would take that bet.

In the older days, when I did that bet in Pipedrive, I did it at $16 million, right? It’s not that it’s the same, but it’s basically similar metrics at the time. I would do that bet at $16 million pre or $20 million post. If I loved the founders and believed the market was large enough, I might take that risk at $20 million post.

It’s just that they don’t want to do it at that valuation. That world doesn’t exist today. But I might take that, honestly. I might take that risk. Harry’s got his skeptical face on. I share his skepticism. I’m just saying, literally, I would take it. So you could be skeptical. I’m saying honestly, based on what I know of that deal—we talked about it, right?—it’s interesting to me if this is a multibillion-dollar opportunity and it’s just going to take a little longer, and it’s still going to double at that rate, and I love the founders.

Founders matter, and I might take that bet if the valuation gave me time. I’m patient if they gave me time.

Rory O’Driscoll

That’s fair. But implicit in that statement is what you’re basically saying, which is quite an interesting thing in the context of what you’re saying: the near-term traction to $7 million over 3 years may not be predictive of the ultimate potential.

A cited report just did a great report on it. I wrote it up. They did one the other day saying that velocity to $100 million is not the perfect predictor of success at scale. You just need to grow fast enough to $100 million to get there. Isn’t that what the OpenView report said? It said you just have to grow fast enough to $100 million to get there.

Rory O’Driscoll

But super-fast growth, historically, pre-AI, hasn’t fully correlated to success. We’ve looked at every deal we’ve done. This is a really interesting one. We ranked it based on the quartile of growth rate relative to the peers at the time of investment: first, second, third, or fourth quartile.

Now, as you’d expect, literally at the time of investment, we’ve done almost no deals below the second quartile, because obviously we’d be incredibly stupid to do deals that were slow-growing out of the gate, right? But the interesting fact is this: there’s very little correlation between great outcomes and being top quartile or second quartile.

In other words, deals that are growing quickly but not astonishingly quickly have just about the same probability of giving you a great outcome as the hyper-growth companies from day 1, which is, I think, validation of what you’re saying, Jason. For example, Bill.com compounded to a huge multiple, right? Obviously, it was an amazing outcome for us, but the growth rate was always, I think, second quartile. There were companies always growing faster than it back in the day. With René Lacerte, it was slower-growing than that; whenever I would see him, he was always growing a little slower than I was.

Rory O'Driscoll

I knew René quite a bit in the early days, and he was always growing just a little bit more slowly than I was. But it compounded into a force of nature.

Right. That’s exactly right.

Rory O'Driscoll

I really like that conclusion. When we did that work, I was actually really happy because what it shows is that it’s not just a metrics business. I tend, intellectually, to be quantitative, so it was a really good pushback from my own brain to say, “Rory, it’s not about ranking them all and doing the highest-growth thing based on the growth rate at the time of investment.”

It turns out that you want to be in the top half of growth rates, but after that, you have to take into account things other than just the absolute number. You can’t, as I say, just rank them and buy. It turns out there’s nuance within that: capital efficiency matters, time to market, the entrepreneur, growth persistence, and all the other things.

A friend of mine years ago had this wonderful saying about venture. When you’d complain about how hard it is to figure all this shit out, he would say to me, “If it was factory work, they’d pay you factory wages.” In other words, if it was simple, they’d pay you $20 an hour. They don’t, right? It’s not as simple as ranking on growth rate and buying from the top down.

When we did the work, it was pleasing. I think HubSpot was actually, for a couple of quarters right as we invested, only second quartile, and it’s been an amazing outcome.

8. The Shocking Data Behind the SaaS Slowdown

Speaking of growth rates, I do want to get to an important topic, which is a piece of work attributed to [likely Jamin Ball]. We’ve seen this real slowdown in SaaS spending for H1 2025. The question is: Is this an ongoing, more permanent trend? Is this an H1-at-a-time issue, and will we progress through it? How do we think about this in the context of where we are today in terms of this SaaS slowdown in spend?

Rory O'Driscoll

It was sort of worrisome. Talking about HubSpot and Brian Halligan, I remember I asked him, “When was the downturn in SaaS?” We don’t talk about the downturn in SaaS anymore, right? We talked about it last year, and I think the general consensus was that it was sometime in the middle of 2024. Things reaccelerated, and you even saw older companies like Twilio reaccelerate.

But this was tough news to see: The rate of growth has actually declined more this year. It is slower, and there can be different reasons. I wonder why Okta has slowed. It’s still 2 products. Why is Salesforce still in the single digits? It’s so big.

I do worry that the pressure of AI is sucking up all the spend. That wasn’t directly the point of Jamin’s piece, but another person pointed it out on Twitter. I added it to my summary of it: If Cursor did almost half a billion dollars in revenue in the same time, it sucked up an enormous amount of those dollars that would have gone to Okta and Salesforce, right? $500 million is a lot in that period of time.

So AI is sucking up budget. Here’s a real example of it happening, and consolidation is still coming. It’s still coming. I just think this pressure—it’s not even if the 2024 days are behind us—it doesn’t appear to be any easier. I do think the slowdown continues, even though we don’t talk about it as much.

If you zoom out, and we had said this literally 4 or 5 years ago, this is not surprising. The industry is mature. If you think about it, 5 or 6 years ago there was a narrative of, “Oh my gosh, it’s been 20 years of SaaS and cloud, and only 40% of workloads have moved to the cloud.” People would articulate that as if it was good news, and I remember at the time thinking and saying, “You idiot, it’s horrific news. Do you know why? Because you’ve compounded from 1% market share to 40% market share. It took 20 years and an average growth rate of 30%. Do the math. That gets you roughly there.”

The problem is that 3 more years of 30% growth gets you from 40% to about 80%, right? The SaaS slowdown was inevitable once you got to 40% or 50% market share. These are mature, served markets.

The best example of that is Zoom, a company I love and so wish I’d done. Who the hell do you think is left in 2023 who doesn’t have a Zoom account? You’re done. If you didn’t buy one in 2021, you’ve hit TAM. You’ve hit complete TAM. That’s an extreme example, but I think DocuSign wrestled with some of the same things, right? Salesforce—most companies have a CRM. I know they’ll show you some survey that says, on a TAM basis, there are other companies, but the truth is you’ve had 20 years to buy the damn thing. It hasn’t changed. If you haven’t bought it now, you are a trailing-edge adopter.

Independent of AI, the SaaS industry was going to hit the mature stage. You’re right, Jason—it’s all the things that happen at that stage: bundling, consolidation, and grinding out the weak.

Fortunately for us as technology investors, at the same time you got this new thing, which is sucking up all the attention and the dollars. Most importantly, because it’s not quite zero-sum, this is the key hope statement: If AI was just replacing CRM with, let’s call it, AI CRM, then there’s no TAM expansion. It really is a knife fight for limited resources.

If, in fact, AI is taking over some of the work and taking over some of the labor dollars, then to some extent it’s additive, which is obviously what we believe it is. But that’s the crucial statement. I don’t think it’s a fight for table scraps. I think AI is the new, new thing, and to some extent, from an economic perspective, the Cursor dollars in software development—or any other of these things—aren’t directly taking away from the SaaS dollars. I think it’s TAM expansion.

At the practical level of getting shit done, my guess is that a CIO and their people can only take on so many projects. There are no brownie points for taking on a SaaS conversion in 2025. There are a lot of brownie points for doing something in AI. The attention has shifted.

One is that the SaaS business is definitely in the consolidation stage, and the AI business is exploding. I’ve almost overloaded that.

9. AI vs. SaaS: The Great Budget War Begins

Rory O'Driscoll

No, you can’t argue with the 40% math. You can’t argue with that, right? If you don’t dramatically expand the dollars going into core traditional business software, you just can’t compound at the rate the markets want. It’s just not possible.

Exactly. I agree. I think that part was clear. I think on the AI stuff, it’s a question of unlocking new dollars and expressing that clearly.

Rory O'Driscoll

The only thing I want to say is that, Harry, you drive the conversation, but there’s so much groupthink on social media. I want to believe—I want to believe—but I think the evidence that AI will unlock massive budget from the human side, replacing humans from the services budget, is still limited. There’s some evidence of it, but I don’t think it’s a slam dunk today, as we’re doing this.

I don’t think it’s a slam dunk that the overall B2B TAM will go 5× because we’ll attach to other human budgets and replace them with AI. I don’t know that we’ve proven that. It makes sense, but I don’t think we have as much evidence of it as we claim on social media.

You’ve been more apocalyptic in the past, so I’m glad to hear that. I think the evidence is the classic thing: The future is here; it’s just unevenly distributed. In some areas, you are seeing the unlock and you are seeing the automation.

One of the reasons I’m pretty relaxed about all these mass-unemployment stories is that I think it’s a long, secular, 20-year trend that we can invest in. I think you will see replacement of labor by AI on a pretty consistent basis, but not an explosive basis. Thus, it’s a great investment theme, I think.

Rory O'Driscoll

I was actually making a slightly different point—not to interrupt, sorry—but I do think it is happening very quickly in the contact center, at the extreme. I have a lot of exposure to the contact center, and half the folks are being displaced at these companies, but you’re not getting that much more ACV. That’s my worry.

You’re replacing a $40,000 or $50,000-a-year human—$60,000 a year fully burdened with benefits and taxes—with not a $20,000-a-year bill, but a $20-a-month bill. That’s the issue. There’s only—I don’t know that there’s enough TAM appreciation when you trade in a $50,000 human for $240 a year. You can help me do the math.

It is happening. I think I can just tell you by looking at my portfolio. When I look at likely Gorgias, which dominates Shopify for the contact center, their average customer has replaced 40% to 50% of their humans with AI, and their ACV is only up 50%. It’s only up 50% with half their humans replaced with AI.

How much does that really expand the TAM? If it’s only 50%, that’s not enough for your compounding math to be exciting, is it?

If what you’re saying is correct, it wouldn’t be.

Rory O'Driscoll

But I wonder about each of the dimensions. Funny enough, I’ve just been doing some refresh work on the, as you say, call-center and contact-center space. Interestingly, you come up with roughly the same rules of thumb that I see in robotics—a totally different space—which is that people tend to replace labor when there’s a 2-to-1 arbitrage.

In other words, when we’re selling robotics and I’ve seen this over and over again, you go in and say, “You’re spending $100,000 on labor. If, on a robot-as-a-service basis, we can do it for $50,000, they’ll do the deal.” Less than that, it’s not worth the brain damage. But you typically can get that. You can have the market, and you can get that.

Rory O’Driscoll

So, I think in call centers, for example, email and call resolution is a $2-to-$4-a-pop kind of human, email-human-resolution-based process. I think we are seeing companies getting plus or minus $1. So I do think you can get that kind of—you get half the labor you save, too, right? By definition, if you’re only saving a little bit of labor, you’re not going to get enough to uplift. But zooming out, the contact-center software market is $10 billion to $15 billion a year in annual spend, while the contact-center labor market is at least $150 billion.

So now I’m going to hand-wave just for a second. You could look at that and go, “Oh my God, it’s $75 billion if you do the 2-for-1 rule.” So it expands from a $15 billion to a $75 billion market if you can eat the labor. I don’t think you can eat all the labor by any means. But I do believe there is going to be at least a 2x TAM expansion, potentially a 3x TAM expansion, as you take the simple contact-center queries and resolve them on a 2-to-1 basis, for half the price, using AI. The AI company will be able to take that capital, take that money, right?

So I do think there’ll be TAM expansion. I don’t think it’s going to eat the whole contact-center market. There are going to be humans on phones and answering emails for the foreseeable future. But I do think there is both automation to be done and TAM expansion to be gained. If you’re pricing it so low that you’re giving it just at the margin to your existing stuff, then, yeah, that’s going to be hard. But I think the value is there such that you can command more.

Well, I’ll give you 2 other quick thoughts for what it’s worth, and Harry, you can take me off it. A lot of the folks that are exploding in the AI contact-center space with huge numbers have either acquired or indirectly acquired BPOs, so they’re attaching into that $75 billion in a very interesting way. Not all that revenue is necessarily software, SaaS, or AI. I’m not saying it’s sketchy, but some of the leaders might be at the edge of slightly sketchy.

But I’ll give you an example of what’s maybe going to happen.

The one I’m watching—I’m just curious; I don’t have the answers—is AI replacing sales reps, right? AI, we’ve talked about that. Everyone out there in the market selling these products is basically trying to sell a $30,000 to $50,000 to $60,000 price point and up. When we get really good at it, it might be $20 a month. It might be $20 a month.

It wouldn’t surprise me if, when Cursor for sales comes out for real, it’s not a bunch of traditional sales processes trying to charge $50,000. It’s like, “This is just software. This is a really good wrapper, and it’s $30 a month.” I just don’t know that we’re going to be able to sustain all these price points that we hope are sustainable in venture and startups. I’m not sure that, when the underlying COGS approaches zero, it’s going to be as sustainable as we hope. I hope this one is, but that’s my concern for the TAM on the software side.

You’re right. I just know it’s working. It’s just going to take time. Maybe this is coming. It’s just like SaaS at some level. You go to 1999—duh, this is it. Everyone was going to do SaaS. It was obviously the wave of the future, and no one founded another non-SaaS client-server company. That’s a true statement.

At the same time, it took 20 years to get everyone across from non-SaaS to SaaS. The sequencing became really important, and it wasn’t just random. It’s easy to say it was, but it turns out the things that had the highest value from that shift, like sales and CRM, went first, and the things that are lower value, like accounting, took longer.

I think it’ll be exactly the same here. It’s not going to be a cataclysmic change in 1 day. I think you have to pick the spots where it works now and avoid the spots where it’s going to take 5 more years. If you do that, you are seeing value.

I mean, I think you’re seeing it in your contact center. The automation is working. Let’s start with that. You’re saying in likely Gorgias’s case, you’re definitely seeing labor reduction, correct?

Rory O’Driscoll

Massive. Very early last year, there was a massive reduction, just not the ability to command such a large premium that it disrupts the TAM. But my mental model is that there’s a 2-step process. Step 1 is: does the AI work and allow automation? You get over that hurdle, and you’re right. The second question is: can you get paid enough for that?

I can’t speak to the specifics of your company. I do believe—and maybe this is it—I actually think it’s hard. An interesting comment is that it’s harder to command huge value on top of software on the SMB side because typically the amount of labor you’re saving is not a lot.

The wonderful thing about automation for large corporate America is that, when you have 2,000 people in a contact center and you’re paying them each $50,000 a year, the quantum of money gets big enough that the quantum of savings from automation becomes compelling. I do think that, for those kinds of higher-end implementations, you will be able to command value from what you’re delivering with AI.

Next-generation AI for B2B, as you approach true SMB, is more and more going to be included in the base, with a limited upsell opportunity. In the enterprise, we’re going to try to do Agentforce and charge massive amounts of money, and we’ll see where this all leads.

But when an S-tier AI is included for free in an SMB product, and when it’s charged at $20,000 a year to replace 1 human in the enterprise, we’ll see how that works out over the coming years. You’re going to look at—instead of looking at the SMB products that are kind of crummy compared to the enterprise ones—we may again be looking at them in AI and saying, “Wow, they’re better. Look what I get for free included with my SMB CRM. It’s included for free.”

10. SpaceX, Tesla, Neuralink: Elon’s Empire After the Firestorm

Final one, guys, before we do a quick-fire. I was in Sweden at this dinner, and Elon’s tweet came out. We’re not going into politics, so it’s not a political question. I think half the people around the table had over $100 million in SpaceX and probably another $100 million in other Elon companies.

Purely from the business perspective—again, I’m not going into politics. I’m not going into health or drug use, nothing, not doing that—how do we evaluate the Elon companies from here? Is it business as usual? How do we think about that?

Rory O’Driscoll

It was never going to work. So you’ve probably just gotten through that pain point, right? I think it’s pretty clear, if you take a step back, from a company-building perspective, that most investors would have preferred to skip the whole thing, right?

If you’d stayed on the sidelines making nice, kind, supportive noises, but without putting your head above the parapet, you would have probably gotten all the benefits that the market was attributing to that. As a reminder, as late as December or January, there was the “Oh my God, it’s going to be amazing for Tesla” trade going on in the market, right? The stock ran way up, right? If you’d stayed less involved, instead of doing what he did as visibly as he did, you probably could have gotten the benefits without the pain, right?

It’s always worth noting. Again, you go back to the shrewdness of Peter [likely Thiel] at Founders Fund. He got some of the benefits of perceived support for the current administration, probably got some of the halo effect from that, without putting himself in the line of fire, right?

Elon, just because he is that entrepreneur who leads with his heart, did the opposite, and the process of doing it and then withdrawing from it has been painful. That’s why you’re right. I was wrong when I said you’re better off than a month ago. That was an incorrect statement.

Given that it was always going to fail, “Thank God that’s over” is probably their feeling. But it would have been so much better had none of it ever happened, from a pure company perspective.

Is there a company that is more materially impacted than others, do you think, Rory?

Rory O’Driscoll

I think, if any, it’s Tesla, only because the truth is electric vehicles have a whole bunch of specific subsidies, both around purchase and around the ability to resell emissions credits to people like GM, right? All those things can be withdrawn by Congress, and not only is there no obvious direct political cost; it’s a much easier thing to do, and there’s no obvious impact on the government.

I think the wonderful thing about Starlink, and why it’s an amazing business, is that you kind of go through the, “Oh my God, I hate you.” I mean, I can see the government sitting in a room going, “We love you, Elon. Got it. We’ll give you all our business.” Then, “Oh my God, we hate you, Elon.” And then it’s like, “But there’s nothing we can do because we don’t have any other rockets.”

It’s the definition of a great business. The definition of a great business is when your customers can hate you and still do business with you, right? The truth is that SpaceX is such an amazing achievement that what was your largest customer, interestingly, is now no longer your largest customer because Starlink is such a big business. Even if one of your biggest customers doesn’t much like you, they still have to do business with you.

That’s not quite as true for Tesla. So, yeah, I think the impact is slightly worse there. I think, as you point out, Harry, the news cycles are so fast, right? Clearly, Elon’s alienated himself. Tesla is the closest to a consumer product of those, and so he’s alienated a certain segment of his population, which has impacted—you saw it in Europe most extremely, right?—some of the sales.

Rory O'Driscoll

But let's give it a year. I think we'll never forget this episode. Stock prices go up and down. This may sound crazy, but I think in a year we will not have forgotten about it, but maybe no one cares.

I don't even know if Trump cares anymore. I don't know if the rest of the world will care. He may not even care.

Yeah. I'm just glad he's back to doing—because we mentioned Neuralink and the fundraiser—you've just got to go back to putting that episode behind you. What an amazing entrepreneur. Let's remind everyone here—a reminder to the haters: Tesla, wow. SpaceX, wow. And then we forget it. But he won't forget it, by God.

OpenAI founding, wow. And now Neuralink, wow. You've just got to step back and say, please. And Jason, you're also forgetting another one that everyone forgets: The Boring Company, reinventing city infrastructure around the world. Have you done it?

Rory O'Driscoll

No.

It's pretty cool. It just works. It just works. It's pretty cool. At least in Vegas, it doesn't do much, but what it does do is pretty cool.

Rory O'Driscoll

I was just thinking about this this morning. It's such a shame, because that's such an unparalleled record of entrepreneurial success. It's kind of like a management failure of massive proportions, if you think about it, to hire the guy who did that for something that's not like that.

It's something that's political, something that involves making cuts, something that involves making political decisions. It's so good that he's back to doing the thing he's best at.

Everyone should play the position where they can score and win, right? This is the position.

Rory O'Driscoll

So, regardless, that gets back to what I said earlier. It's not that it's better, but everyone's back now in the right place. Elon is back building amazing companies.

Yay, everyone. Oh my God, that was a painful 6 months.

Rory O'Driscoll

I'm sure that's the mental model. One of the beauties of X is that we can see what every billionaire thinks, right? There are so many billionaires in tech on X, and we can see what they think.

The meta question I wonder is: Should you invest once a billionaire becomes unhappy? Chamath seems unhappy. I don't know him. I met him, and he was brilliant for so many years. Elon doesn't seem happy. Can folks still be as innovative and groundbreaking when they reach the unhappy-billionaire phase?

Rory O'Driscoll

I don't know. The one thing I will say on Chamath is that the public persona and the private persona are drastically different. He's actually very humble and kind privately. The public persona is a very different display.

The only meta question is: Are your best days behind you if you become the—let's leave Chamath out of it. I don't really know him, but my very limited interactions are consistent with that.

Rory O'Driscoll

Harry, for sure. I would say at least 50% of the tech billionaires on X are unhappy. Can they still innovate at that stage? Do they still have the same level of drive or passion? Or is their grouchiness an inhibitor to being an innovator?

I think that's the sentence here, and I love what Harry said. The point is this: You don't know if someone's happy until you know them personally, right? The truth is, social media and politics just tend to drive a person into a certain persona. It takes a certain persona to be perceived as winning.

We all understand the heightening effect in social media of the most extreme situations. I think it's one of those environments where, like Gresham's law for money—bad money drives out good—the equivalent of that in social media is that bad, opinionated people drive out good, boring people.

When you're in the political arena, when you're in the social media arena, it just forces a persona that comes across, at least, as very angry and unhappy. I don't have an opinion. I'm willing to suffer the risks of getting a billion dollars and seeing how I do, and that's what's going to happen. I'm willing to run that risk.

11. Kalshi Quick-Fire Round

I love that. One more good deal. I want to do a quickfire. Let's start with number 1: Sundar Pichai leaves Google this year. Yes or no?

Rory O'Driscoll

The only thing I thought about when I saw this was when I was a VP at Adobe and Shantanu [likely Narayen] was the CEO. He wasn't a founder, but he was on the rise there. I didn't directly interact with the Adobe board or others, but when I saw the vibe, it was clear that they were not going to let that guy leave.

This was a company that was no longer founder-run, and he had figured out the transition to the cloud. You could debate the decisions, but his ability to steer that ship in the right direction—everyone at senior Adobe knew it was even riskier to have anybody else. I suspect it's similar at Google. None of the large stakeholders want him to go, no matter what, and they will do almost anything to keep him.

Despite everything on the search side, which is still the majority of revenue, and despite all the threats today, that was my sense. Shantanu was a great CEO, and I'm sure the board and others would have grabbed him by the jacket and never let him leave Adobe when I was there, even for many years.

Yeah, some version of yes. You're right: The question is really abstracting from personal stuff. Random events could cause that. There's some probability that anyone can leave at any point in time; it's called death. I presume that's not what we're saying. What is really a proxy for saying is: Is he doing an amazing job at Google?

Rory O'Driscoll

I think we've definitely gone from, “Oh my God, the world is ending at Google,” to, “They're doing good stuff. They're getting good models out there.” It's not obvious that it's all gone to hell in a handbasket.

They have the classic innovator's dilemma. The Google model, the search model, is an awesome cash-spewing machine. But let's be honest: When I get the lineup for the Harry Stebbings podcast, I start on ChatGPT to do my research, not Google anymore, right?

They have that long-term dilemma. It's not clear that putting someone else in the chair will solve that. My guess, by far, is that the most likely outcome is that he does not leave. They will continue to manage what they've got reasonably well without ever solving the existential problem.

I agree with you both. The only thing I do think is interesting is that Sergey Brin is back, and he's back-back. He's speaking publicly about being back, how it's the most exciting time ever, and how they have to win more than ever, having had a hiatus. I'm intrigued to see what that interplay would be.

Rory O'Driscoll

If it were to happen, the founder-comes-back narrative is definitely there. There are precedents, including obviously the most amazing one, and then the Starbucks guy who keeps coming back every 3 years, like it or not.

It would be the narrative that would be easy to sell. My guess is that if the founder decided he wanted to do that, it would be on the table.

[likely Kalshi] agrees with us. They say absolutely: He will stay. Let's go to the next one: The New York Times wins the OpenAI lawsuit. This one's pretty evenly split. Yes or no?

Rory O'Driscoll

If you include “win” or “settle,” then I'd give it 80%. I can't believe that OpenAI is going to want to let it run forever. If you lump settlement into the equation, do I think the New York Times is going to come away with a win of sorts out of this? Yes.

It might be less likely to be a settlement simply because, in the end, close to the trial, people settle. It's less likely that it goes all the way to a jury or a bench trial and then they win. But I think they've got enough of a case to be in the room, and it's one of those problems where money ultimately can help solve it.

Yes, I think they win something from suing. I think they may even make it more clear: They win more from suing and then settling than the other companies that did smaller media deals with OpenAI.

To make it harder for myself, I think their strategy of suing will be validated versus just cutting a $20 million or $30 million deal with OpenAI 2 years ago. I think they will get something from their effort.

Yeah, I agree with you that if they've already determined they're going to pay in the end, then that has to settle, right? In that sense, they'll win, but they may not win the lawsuit because it gets settled.

Rory O'Driscoll

At some level, if they don't, it's going to go to the Supreme Court. The folks in the Supreme Court will decide what fair use means in the age of the internet. That's a big—probably a 48%—gamble, because it's easy to see them coming down on the side of the content providers.

Maybe [likely SaaStr] gets a check. Maybe I deserve a check. I think I deserve a check. ChatGPT scrapes a lot of our content. I get a lot of traffic from it already. Why don't I get a check? Seriously, why don't I get a check? Why does The New York Times get a check and not me?

I don't think it's right or fair under fair use. They're directly taking my content, which is very unique and specific to me. In theory, there's an argument to take it all the way to the Supreme Court and win, and not have to pay anybody, because OpenAI has slurped up the entire internet.

They should settle because it's pretty confusing.

We’ve basically, in the age of AI, decided to surrender a lot of copyrights and a lot of privacy. OpenAI can already record all of our conversations 24/7 in its macOS app. So we’re giving up copyrights, we’re giving up privacy, and we’re going to learn where these new lines are, but they’re not going to be the same as they were 2 years ago. They’re not going to be the same.

The Supreme Court’s conservative. It’s a bunch of Harvard grads. I guess it’s Republican-dominated, but I don’t know which way it’ll go.

I just want to say, I think the most fun thing about that answer is Jason, without blinking or laughing, described Harvard as “conservative.” I think they would be so glad. I think they’re going to put that on their Harvard website: “Look, Harvard grads are conservative. Leave us alone, please.”

Rory O'Driscoll

Damn it. I don’t know whether it’s right to take on the Trump administration, and I know there are a lot of principled reasons, but I do think it’s conservative—not politically, but with a small C. It’s got to be the most conservative organization that I have any affiliation with. Very conservative.

I think one nuance on this that gets to it, which we picked up in a very interesting reference we did on another deal, is really interesting. Obviously, OpenAI and all the models need access to “modern news” to be able to answer real-time questions the way Perplexity initially did and now everyone has copied. You have an LLM plus web search, so they clearly need that. That’s a given: They need access to modern news sources.

A really interesting question, rather, is this: If I have, as OpenAI, say, The Washington Post and The Wall Street Journal, do I need The New York Times? In other words, is a third national news source additive or not? It raises a very interesting question I hadn’t thought about until I talked to this person, who made the point: Yes, you need news to be able to give the full LLM experience, but do you need the third or fourth marginal news source? Maybe not.

That’s the thing that would maybe make me hedge my bet. It may be a separate issue from what the legal rights are on this. You could imagine an LLM saying, “I need to get modern news from the Associated Press and one national newspaper, but I sure as hell don’t need 6.” If you think back, if you reflect even on your experience on a Sunday morning when you’d read 3 or 4 newspapers back when they were papers, by the time you got to the third, you were like, “I know already 90% of the content is repetitive,” right?

The argument for OpenAI gutting it out is that they’re saying, “Hey, I already have it from 2 or 3 people. Maybe I should have paid you for the past, and maybe I’ll lose that part of the case, but I don’t need to license your content on an ongoing basis to be able to deliver the full search-plus-LLM experience, provided I have one provider.”

If that’s the case, there’ll be an interesting game theory process going on around media content pricing. And that’s why Jason with [likely SaaStr] will get paid so handsomely.

Rory O'Driscoll

Absolutely. Unique content. No need. You think I’m kidding, but Harry, how much do you get paid by Twitter a month?

I get, like, $3,000 or $4,000 from Twitter a month. How much do you get? Do you have it turned on?

Rory O'Driscoll

I’m so pissed off about this. I see everyone posting. I don’t know how to do it. Now, I know it’s not the same because Twitter’s opting to pay its creators, but I’m getting paid $50,000 a year for my tweets. I want $500,000 a year from OpenAI for my SaaStr content. It’s more valuable.

I just want, like, $40,000 a month. That’s my— I don’t need millions for tweets. I would get, like, $200,000 a year for tweeting.

You might. Yeah, you might. You might.

All joking aside, I think the fun thing about this is it will drive very interesting conversations about what content is, in fact, valuable.

Speaking of that, one final one for you: Linda Yaccarino—will she leave Twitter this year?

Rory O’Driscoll

I don’t have a clue. I’m going to leave it to Jason.

You know, it’s a fun question.

Rory O'Driscoll

We get glimpses of Twitter’s financials—X’s financials. We don’t see all of them, right? Her job was to bring in the advertisers and create a buffer there. That doesn’t seem to have been wildly successful.

I know this is mean to say. I’m only judging the public persona; I don’t see how the team is managed internally. But objectively, it seems like she’s the weak link on the team. It seems like one you could upgrade. Of his C-level team and his companies, this seems like the CEO that maybe he’ll upgrade this year.

That may be true, but just thinking pragmatically, if I was—and I never thought I’d say this—poor Mr. Elon, coming back from a bruising 6 months in government work and having 5 or 6 amazing companies to work with, plus Twitter, I’d probably say to myself, “Oh, I’m just not going to take on the hard thing. Can I just focus on building cool engineering shit at Twitter, whatever?”

I mean, because remember, the person who changes the CEO is the board—or, in this case, Elon himself. Do you just want the heartache, dude? Just let it run. Don’t be a hero.

Rory O'Driscoll

I hear you. But we forget, like, Elon recruited Ilya Sutskever to OpenAI. The guy’s a good recruiter.

Agreed. It was more a lazy move.

Rory O'Driscoll

I think he knows. Listen, I might be wrong. I don’t work there. The external persona may not be consistent with the internal value, but it seems like she’s the weak link on the team. I think, when he has the moment in time, he will bring in the best media executive in the world that he can get.

For all the folks that he has alienated over the last few months, there are others who probably are bigger fans. Go find the best one. Lean into your super fans if you have them. That’s where the magic is, guys.

Thank you so much for doing this with me, as always. My favorite is always the comments. The comments, honestly, are always my favorites, and I so appreciate you both. This has been wonderful.

Rory O’Driscoll, 6 a.m.—I mean, dude, credit and love.

Rory O’Driscoll

Hey, credit and love.

Jason, you’re the man.

Rory O'Driscoll

Okay, I’m going to stay here for a long time because I’m at 26%. Have fun.