OpenAI、SBF与Perplexity：VC知道而你不知道的事

Guys, it is so good to do this. I always say this. This is my favorite show to do.

Jason Lemkin

But you say that to all the girls, Harry.

You know what? Clearly, you don't listen to the show, my friend, because I don't. I actually just thank them cordially and pretend that I'm thrilled.

1. Owner’s New $120M Round at $1BN

I would love to start with some big news, which is Owner's new round. Jason, you led the seed here. It's $90 million at a $900 million valuation, I think it is.

Jason Lemkin

I don't even read the document. Honestly, I don't read the document. I just found out it's $120 million when I read it today. I didn't even know. I didn't even read the document.

All I care about is my ownership. I don't really care about the other numbers; they don't really matter. I'm being a little facetious, but actually, no: $120 million.

The learning is, look, they're at $40 million, growing 10% a month. Sometimes faster, sometimes slower, but the trailing is there, right? They have definitely done a lot in AI-infused marketing, but in some ways, it's also a pre-AI company. I mean, it's really good software. We can talk about that, but it's not OpenAI, right?

What was interesting was that the narrative is right. This is just my learning. With those metrics and that growth, Adam did a great job doing what you're supposed to do: getting to know all the VCs over time and socializing.

But then there is the classic thing, which actually gives me a few hives: open the data room on Monday, get 2 term sheets that afternoon, and get all of the term sheets by Wednesday. That process still makes me nervous to this day, for a variety of reasons.

But the meta-lesson, going into our conversation, is this: triple-triple, double-double is still good enough, right? Owner is growing faster than that, objectively. But if you are, then it feels like there is unlimited—I mean, this is the learning—there is freaking unlimited capital.

Even just the insiders could have filled out $150 million in this round. That's the thing: seed is hard, but if you're in wherever this zone is today—and I think it's harder than ever—if you're in it, there is just unlimited capital. But you have to be in that box. That's my learning in the box, right? That's the meta-learning from it.

Rory, what do you do in those cases? I have this now with the team where they say, “Harry, great founder, great founder. We need to decide by Wednesday, and it's Monday.” And I go, “I can't write a $15 million check by Wednesday when it's Monday. I just can't.”

Rory O'Driscoll

The framing of the question says why it's wrong. If it's Monday and you're hearing it for the first time, and you've got to write by Wednesday, you're just way behind. Even if you crank and write a term sheet, you're also going to be up against someone who's met them before in the last round, is ready to write another term sheet now, and has done some work the prior Monday to Friday. For them, Monday is just confirmatory, and then you've got 2 days.

The speed at which things are moving makes it really hard. But there is no use crying and saying, “Shit, I wish it was 2010.” You've just got to, as you say, play the game on the field.

It's really hard to go from zero to a decision in 2 days. Maybe you can assimilate the data, but it's very hard to assimilate the person. In other words, it's very hard to know enough about that person in 2 interactions spaced 1 or 2 days apart to pull the trigger here. That's the hard part.

We've internalized that you've just got to be tracking them. You've got to put a lot of effort into seeing the ones that you want to see in advance, to know in advance what you want. The consequence of that is the second-order problem: you can't be looking at everything equally, because you can't woo 10 people and you can't woo 200 people. Up front, picking your shots on where you want to do the wooing becomes part of the struggle, right?

I'm criticizing myself. My rule of thumb is that if I don't have a list in our Salesforce of 10 to 20 names that I know I want to see, and that I could imagine investing in over the next 12 months, I'm probably not doing my job. We talk about it internally. Everyone has their hot list.

If you can't name those companies, and you're just wandering around hoping something will turn up on Monday that will make you money on Wednesday, it's not going to be successful.

Totally get you and understand. The question that I'm finding Series A investors and our Series A team asking is: how quickly after the seed can we preempt? It's so freaking competitive that they're saying, “The seed was done a month ago. Can we just bid it up now?” Because when we don't, Lightspeed, General Catalyst, and Sequoia do.

Jason Lemkin

Well, look, if you bid it—as silly as these things sometimes seem when they're bid up in 2 or 3 months, it sounds silly at first, right? But going to Rory's point, let's say you met them and you really liked the deal, but it didn't work out for whatever reason. Then you get 2 more investor updates, and the company has grown 50% in 2 months.

You've already done the diligence. You already met the founder. Why wouldn't you do that deal if you believe the price made sense? You get a second chance, right?

Sequoia seems to be really good at these second, third, and fourth chances. Sequoia did this round at seed, then this round, and then did Clay at $1.5 billion. To me, it seems chaotic, but I don't think it's chaotic, right? It's win when you can win, isn't it, Rory?

Rory O'Driscoll

Yeah, I do struggle with this: should we do it 3 months later? There is a little part of me that says, “Oh my God, that just feels wrong,” right? I admit my biases are that way.

How do you know?

Rory O'Driscoll

They don't need it. You know that this is solving for your problem, not for theirs, fundamentally.

Yes, true.

Rory O'Driscoll

But some of them do and some of them don't. If they don't need it, they're big boys; it's up to them to say they don't want it, right? I don't have to solve everyone else's problem. It's hard enough to solve my own.

I do agree that's an issue, but the bigger issue is whether I can really stomach paying twice as much as someone did 3 months ago. Can you stomach paying twice as much as someone else did 3 months ago? We've never done that. I struggle with that.

But then, as you know, as a theme with me, I always think, “Am I getting that right?” For example, there is no doubt that some of the best companies in high-growth markets have the highest velocity of raising. You look at some companies in the last 2 years because they've just found—I mean, at some zoomed-out level, it's Adam Smith giving a signal to everyone: “These guys have found a place to put capital. Give them more capital, you fools,” right?

OpenAI has been the classic example. We put in $100 million, the model got smart. Let's do $1 billion. It got smarter. Let's do $10 billion. You're going to double down until it doesn't work. That's the signal that says we need to get money into this thing.

I'm wrestling with exactly that: how soon after that should you do it? You don't want to get to the point where you're paying twice the price of someone else for the same company. That feels like an idiot, which is why we haven't done it. But at the same time, as Jason said, 6–9 months later, if it's a good deal, that's when you should be engaging.

2. Why Series A is F****** Today

On the flip side of this, we've just seen—literally just before this, and sorry, Rory, then we'll get to the bits that you have prepped for, because I promise I won't throw in everything—Carta just announced that Series A rounds are down 81%.

Rory O'Driscoll

Yes, I agree with this.

Our Series A team was scratching their heads, asking, “What the fuck has happened to the Series A market?” Are you guys seeing the same thing? How do you reflect on this 81% drop in Series A rounds?

Rory O'Driscoll

I saw the data. The seed and the pre-seed are the “believe in the team” rounds, and the A is the “show me the traction” round. Belief is easy to manufacture, and traction is hard. Once you get to the traction round, you either have it or you don't.

If you don't, I think the smart thing people are doing is not trying to raise. They're just trying to figure it out. We'll talk about a company that did that in a minute and did an amazing job of it.

It's kind of like the law, like a barrister: never ask a question unless you know the answer. Don't go out and try to raise money unless you're pretty certain you've got what it takes. Series A conversions are down.

Jason Lemkin

Who cares? Go make your startup S-tier. The whole point of venture is to invest in S-tier startups. If you built a B-tier startup and it's hard, or you built an A-tier startup that could get funded in 2021 and it can't today, cry me a river.

I mean, anyone that's been a founder—I know there are a few exceptions—but I was there with Owner for years when it couldn't get funded. I couldn't get funded multiple times; I went through sequential years as a founder. It was easy to get funded, then impossible—easy to get funded, impossible, right? Crying about it because it's hard to raise a Series A is a B-tier way to look at it. Be the best of breed; you'll get 5 term sheets. You really will.

Rory O'Driscoll

This is a totally weird analogy, Jason, but I get what you're saying. It's like, years ago, I actually had stunningly bad cancer. You get the numbers and the statistical survival rates, and they're miserable. What my wife and I internalized at one point is that the statistics are actually interesting to doctors because they have lots of patients and want to keep a rough eye on things, right? But to the person—to the patient—it's 0 or 1: you either make it or you don't.

It's the same thing here for the startup. You're exactly right: either you have something that's worth funding, in which case, yay, you, or you don't. The fact that, at the margin, it's interesting that it's slightly easier to get money, right? But fundamentally, I think you are right: you want to get money because you have a good thing, right? Having a good thing is the hard part. And you're right, the money is the easier part.

I mean, venture is pretty much on top if you have the kind of metrics that you guys got at Owner. Look, I was telling Harry before we started here, whenever there's a good deal announced, I go to our Salesforce, look at the chatter notes, and look at why we passed on that round 4 or 5 years ago and what idiots we are, right? That's just the nature of the thing. At the time, it didn't look obvious. You guys hung in there, you made it work, and all power to you.

The cancer you talked about—why did you come back to venture? Why didn't you throw in the towel and say, “I've had enough of chasing these deals?” Seriously, we've talked about how folks are leaving venture, right? Some certainly aren't leaving electively, right? But we've also seen a lot of folks who made a lot of money step out of venture in the last couple of years. Why did you come back?

Rory O'Driscoll

I did not mean to make this a personal thing. It just came into my head, but I've got no problem talking about it. I internalized, at the time, that I was just in my 40s and I got stage 4 colon cancer. My big aha going through it, when I was doing chemo for about a year and a half, was that I kept on working.

Some people say, “I had this near-death experience. I want to travel around the world, live in a yurt, climb Mount Everest.” What I realized is that I like my work, and I just like to keep doing it. You just do it until you can't. I've processed through the whole existential “What are you doing?” thing and come straight out to: it's a good job, we get paid pretty well, and it's quite interesting. I have no desire to do anything else.

That definitely brought it home for me. It was like, “Hmm, if you're not going to change…” Sometimes it irks my nearest and dearest when she'll say, “Really, you couldn't even change?” And I'm like, “Yeah, that's what you got.”

Jason, you know what I love so much about how cancer changed Rory? I just want to find another enterprise storage company. Give me another company. I don't want to see Avere; I just want another data-storage company with high NRR.

Rory O'Driscoll

Hey, you know, there's a lot to be said for companies. Just give me a good one. I actually used to take calls when I was on chemo. I used to lie on the floor because it hurt. I don't mean to make this a personal thing.

3. Could Tiger Global Be Saved by OpenAI and Scale

As a random comment on the “yay, venture” point, I was one of the first people to get Avastin, which is the Genentech drug that was funded by venture capital way back in the day. Kleiner Perkins funded Genentech, and before that, you'd have been toast. After that, it was literally a week after it was FDA-approved. At the risk of getting political for a second, I don't know what my healthcare provider paid for that drug, but I'm damn glad they did. So, go team.

You said “yay, venture” there. One of the villains, or criminals, of venture, so to speak, over the last few years has been Tiger. People have criticized the deal volume, deal count, and dollars out the door; it's definitely been viewed with some skepticism. But positions with OpenAI and Scale AI are actually proving to make some of that fund performance not look so bad. If we project forward, this could really save them. How do we think about this, chaps?

Jason Lemkin

First of all, it's not a morality play, so there's no right or wrong. They had a very aggressive strategy: they did 300-odd deals in 2021. Obviously, in retrospect, a total mistake. If they had enough time diversification in Tiger 15 or PEP 15, whatever they call it, to get some OpenAI in, and then had the guts to do a lot of it, then, yeah, can you pull it back totally? I don't know, but at the very least, you can salvage something from what looked like a very tough fund.

So, it's entirely plausible that 1 deal has a significant impact. Now, given the size of the fund—$12 billion—given the 350 deals in 2021, this gets to something we've talked about earlier. The only way it works is bet sizing. In other words, if they did 1/350th in OpenAI, it's not going to bail them out, even if it's a 10X, 20X, or 30X. If, on the other hand, they had the guts and the foresight and the courage to put 10% or 15% of the fund in something that could 7X or 8X, then, yeah, maybe they've snatched victory from the jaws of defeat, and more power to them. I don't know the amount they put in, but that's the key fact.

Wait, let's go back in time. Help me, Rory, because you're so good at this. They did 350 Series A deals, basically, in a year, I think. But 2021 was good times, especially in B2B, because everything was working, right? What would it have to work out to in B2B for that fund to do 3X? Help me think through the math, because I think you shouldn't conflate the 2, because the truth of any 2021 exit is a 2018 late-stage deal, right?

Rory O'Driscoll

Yeah. Which is why, by the way, if you look at it, a lot of the late-stage funds are going to have a very excellent 2018 vintage, where, interestingly, the earlier guys won't because they won't have had it all the way through the system by the time the window shut, right? The later-stage funds are going to post a very nice '18 fund.

So, probably Tiger, as the definitive late-stage fund, has a great '18 fund. I haven't checked with PEP 14 or whatever it is, right? Because if you were buying stuff at $2 billion in 2018, you were probably selling stuff at $6 billion in 2021, right? They were the gems at $2 billion there. That was beautiful.

And then, of course, everybody does the same thing. You start thinking you're smart because you're making money, and then you do 315 more deals in 2021, right?

The math of those, Jason—I mean, look, it's a $12 billion fund, so you've got to return—let's just do $24 billion to get a 2X, you know? My assumption is you're not going to have the hit rate.

Jason Lemkin

The calculation you're asking isn't worth doing because this isn't going to work. Twenty percent or 30% of the 315 deals become solid deals, right? There's not going to be enough deals. It's going to work, if it works at all, because they put 20% in OpenAI and 10% in Scale AI, and then they get a couple of drips and drabs. That's a couple of decent '21 deals that become okay, but not much, right? It's just hard to dig out of a $12 billion hole in $100 million increments.

I have to say, I think they are unfairly criticized. I completely agree that the deal volume was off the charts. I don't think the strategy was that wise. That said, if I was an LP in those funds, and they hold on to OpenAI and Scale AI—and I'm actually in some of the back catalog with them, the good and the bad—the bad ones will 1X. There are liquidation preferences. They haven't raised crazy amounts. They will 1X. They won't lose money on them. And there's actually quite a few that are really good.

Rory O'Driscoll

Totally. I think that's fair. People have strategies. They have to work or they don't. The facts come in. The great thing about this business is, in the end, we don't have to say what we think of each other. The numbers tell, right?

I think if they were able to put a lot in those 2 big deals, they win. And they should win because that's a very shrewd move. When you're in trouble and you have a lot of troubled deals in '21, and you say to yourself, objectively speaking, “What's the only strategy that can save this fund?” the correct answer is to shove the remaining 40% of the fund into 1 of the only 2 or 3 things that can go to the moon.

It's a total morphing of the strategy. There's an implicit statement in there that the strategy was wrong, because otherwise you wouldn't be in the hole, right? But at the same time, getting out of the hole counts for points, too. It's a little like Bush in Iraq. It was a mistake to go in, but at least in 2006 he surged his way out.

4. Why SBF is the Greatest Investor of the Last Decade

This could be the Tiger surge, right? You made a big hole, then fixed it.

Rory O'Driscoll

I have to say, I think the thing that's fascinating is actually when you look at FTX: its positions in Anthropic and Cursor would have saved them if they'd had the time to prove that out and they hadn't done the commingling. That would have saved their financial performance with those 2 investments alone.

Jason Lemkin

Well, that's what I tell founders: just do a little fraud. You have to know where the line is between too much fraud, right? And that was too much. That was the problem. You poked the bear too much.

Rory O'Driscoll

You poked the bear.

Jason Lemkin

Yeah, he just—I mean, a little commingling, using the funds to buy a compound in the Bahamas, that's okay, but—

I feel the need to state for the record that Jason, once again, is merely being ironical, and we will not be leading the promotion for this show with a picture of Jason and me with someone saying, “Only a little fraud is fine,” okay?

Jason Lemkin

But I think all of crypto is a little fraud. He just took it too far, right?

No, Dan, Stephen, we couldn't fit that much on a thumbnail. We'd just have “Only a little fraud.”

Jason Lemkin

Yeah, exactly. No mitigation.

Okay, got it.

Rory O'Driscoll

I mean, there's a lot in there, for what it's worth. Firstly, it's just to observe that the fact that Sam invested early in Anthropic and Cursor is astonishing. What a talent. What a willingness to look at new stuff in '21, before the ChatGPT moment, when it was just crazy stuff that people were saying: “This might work. Who knows?” It might have been in '22; I wouldn't swear by it. But it's astonishing to pick 2 of the most important companies in the post-'21 crash and nail it.

Jason Lemkin

Yeah, clearly a very smart man, right?

Rory O’Driscoll

And to your point, the other thing—second thing—is your comment on that, quote-unquote, “have saved him.” It’s worth pointing out that the core business also was a pretty impressive and good business. I mean, the core exchange worked fine, too. It was all the weird shit on top. It was all the commingling. So, yeah, he had an excellent business, blew it with fraud, and did some great venture. That’s quite a polymath.

Do you think he deserves another startup if he gets out on time, if he gets pardoned? Do you think he deserves it? Theranos got another one, right? The Synapse guy. What do you think? Do you think a top fund will give him a couple hundred million to get his next venture going?

Jason Lemkin

That’s a good question. Would you fund him? I don’t think Rory or I would. I’ll defer to Rory, but I think someone will fund him. He’s smart, as Rory said. He showed the upside as well as the downside. I don’t know if there’s a lot of morality in the business. I think someone will give him just, like, $100M to start. I don’t mean this is like an OpenAI spinout. Just a little $100M at a $1B post to hire a team and get things going and see where it goes. It’s not a lot out of the new $5B fund, is it? It’s 2%.

Rory O’Driscoll

I think it’s all cute. And look, there are lots of examples, though. At the risk of sounding like a stick-in-the-mud, I think there’s a big point spread between dodgy, aggressive performance; acute failure—WeWork being the most obvious example of major failure, hubris, grandiosity, and implosion of money on a large scale, but no convicted crime—and, on the other hand, being convicted of a whole bunch of fraud-related offenses and being found guilty of them, right?

I just think that’s a bigger lift in terms of attracting capital. So, no, my gut, at the risk of being serious, is that once you pass the convicted-criminal stage, the bar goes way up. Fairly or unfairly, you can argue your toss; you can argue he’s brilliant or not. But I think once you’ve been sentenced to 25 years by a judge and found guilty by a jury of your peers, there’s a big— it’s a long way back.

He might get pardoned, though. Stranger things happen.

Jason Lemkin

He might get pardoned. I thought you said he might get a partner. I was like, “Andreessen Horowitz, aren’t they?” Oh, wait, that could happen.

Hey, give Andreessen some credit. They didn’t do that deal. They’ll be the first to say it, too. They did not invest in that company, so you can’t tag them with that problem.

Jason Lemkin

That’s totally true. Multicoin did that deal, but you can’t tag them with it.

On the—you said, “Yay, Andreessen, saving venture.” The thing that really stood out to me was essentially Coatue’s marketing of the new fund as being open to anyone with $50K. Philippe Laffont might not all-in, and it really showed this kind of democratization in access to venture, and a new model where you could redeem; there was liquidity built in. How do we think about this?

Jason Lemkin

I watch a lot of YouTube, and they have movies, and Boiler Room came up. It was the story—Vin Diesel. It’s just a story: let’s just rip off the retail investor.

You really are the cynic here.

Jason Lemkin

I’m not a cynic. I just think people are cynical.

I’m not a cynic.

Jason Lemkin

I think this is cynical: ripping people off for $50K when they don’t know what they’re getting themselves into.

Rory O’Driscoll

I’m bringing a couple of things together. One, in defense of the thing, a lot of the big private-equity firms are making the same move. They’re trying to tap additional sources of wealth. I mean, we’ve just discussed it. A lot of the endowments are under pressure for capital. You just have to get money where you can, and it’s one more vehicle.

I think if you look at people like Blackstone and people like that, a significant and expanding portion of their capital is coming from high-net-worth individuals. I don’t know if the bar is as low as $50K, but there’s no doubt that all the wealth managers have these PE-type products where they’ve constructed some element of liquidity similar to what Coatue is doing. So, it’s not like it’s way out there in the blue.

Then I think the other comment I’d make is going back to—I think this is now a bit of a hobbyhorse for me, but just to say it: here we are. This is the complete round trip. Everything used to go public early. Then they stopped going public early, and they stayed private for longer, and all the institutions could do it. And now we’ve recreated this vehicle to allow public investors to invest in these companies.

In fairness to Coatue as well, I want to say at better economic terms than the traditional 2 and 20, right? But when you zoom out a million miles here, this is all madness. These companies should be public, and then Fidelity Growth Fund could do them at 70–50 bips instead of Coatue doing them.

The core problem here is that all these companies want to remain private for longer, and you’ve got to ask yourselves, why is it such a shitty experience being in the public markets? Because it clearly is, right? This is a workaround for a problem that would be better solved some other way.

Is it a workaround that others adopt? Is it a workaround that Lightspeed, GIC, and others go, “Hey, this is something that we should be doing as well?” Or do you need to go to—you think?

Rory O’Driscoll

I think anything that works in finance gets copied immediately. It’s just one of those life rules, right?

Do you think it will work? And respectfully, he’s on All-In because he wants people to invest, and he’s marketing a product. I think they’re very smart people. One thing I didn’t understand, by the way, is the concept of—was it Dell and Bezos as anchor investors? I don’t know if you saw that part of the announcement where they said—I think it was Michael Dell.

I didn’t understand that, because I would have assumed that they can get all the liquidity they can get. They can get whatever asset they want. They passed the $50K threshold. Let’s put it that way: Jeff Bezos and Michael Dell make the $50K sophisticated-investor threshold.

So, I didn’t understand that. But do I think it’ll work and raise capital? Probably it will, because the facts are there. The other PE vehicles have worked. And then let’s get real: at the end, it’ll be packaged as some version of, “This is the only way you can get access to OpenAI, Anthropic, and all the hot new startups that are changing AI. Sign here.” And people will sign, right? Because it’s true. It is the only way you can get access to those assets. So, yeah, I think it’ll sell.

Jason Lemkin

Listen, I’m not a real retail guy, but based on my limited experience, I would always write that if you tell folks they can get access to these hot names, they won’t even understand what the carry economics are. They won’t even process what it is, right?

In fact, I think the way they’re doing the—whatever the carry is—12.5% and 1.6842% fees, it’s almost too insider-baseball-y, right? I don’t think the average retail investor even knows what that means, right? And it’s actually arguably very high, although it’s low compared to traditional venture, right? It depends how you look at it overall. Is this Fidelity, or is this VC?

Rory O’Driscoll

I think, Jason, you’re right. I didn’t take that into account: 50% of the assets are public.

Jason Lemkin

That’s actually how they probably got it.

Rory O’Driscoll

You’re exactly right. That’s probably how they got it.

Jason Lemkin

Half of my assets are public stocks, which should be 50 bips, and half of my assets are venture, which should be 2 and 20, so you blend.

Rory O’Driscoll

You’re exactly right. Good point. So, it’s not cheap. It’s market.

Jason Lemkin

You know, there is a line where this could rip retail investors off, and I just hope it doesn't cross it. There are really only 2 things that go wrong with an investment like this: either you have the wrong manager or it's just the wrong asset class. Maybe the third is the wrong structure. So, let's take it apart.

You're getting an excellent manager, right? Coatue is a top-tier public-private manager. So, if you look at the 3 things that go wrong when a retail investor puts his money in, the dumb-stuff mistake isn't going to happen. You've got a top-tier manager.

I'm sorry. Can I just—if we're going to ask a question—are they a top-tier private manager?

Jason Lemkin

They're at least—put it this way: I would say for the kind of later-stage things they do, they're clearly very sophisticated. They've won some good deals. Are they as good at Series A as, pick a name, Benchmark or KKR? No, but they have established a meaningful-scale franchise and are thoughtful. Compared to a lot of ill-thinking people, they're thoughtful investors.

They're in the flow. They have access. It's not like a million billion coming in where they have no access, because one of the things that goes wrong is when these weird outside vehicles come in and have no pre-existing business, you just don't get the good deals. They're already in the flow. They'll be able to get good deals.

I think the real questions are actually the other 2 issues. One is just the timing of venture: if the next 5 years are tough, no one can save you. Then, obviously, a lot of retail investors will think, “Oh my God, that was dreadful,” right? To me, that's the bigger question.

Lastly, the structure thing is interesting. Can you give people enough liquidity? Will people internalize that there are limits to liquidity? This is not SPY, your friendly local ETF that's fully liquid at 1 minute's notice. There are redemption gaps, redemption blocks—what are gates? That's what they call redemption gates.

If you remember back a couple of years, I think it was either one of the Blackstone or BlackRock funds—I get those 2 mixed up—but they had a real estate fund where they had to put up the gates on liquidity because they just couldn't meet demand. Those, to me, are the questions much more than whether they'll make bad investments. I don't think they'll make bad investments. They're shrewd guys.

5. Why No Individuals Should Invest in Venture Funds

The question is whether the retail investor who thinks they want it today really wants it when it's cyclical and illiquid, and then they discover that. If I step back, I do worry it's being oversold. But any individual I know who wants to LP into venture funds—anybody's, including mine—I tell them, “Don't do it,” because no one understands the illiquidity.

It's not worth it. It's not worth it. It's not worth it for a large amount of your income because the illiquidity will stretch it out. It's not even worth it to put a little bit of money in: “Oh, I'm going to put $50K into 20VC, and then it does 8X, but it's 17 years later until I get it back.” It's just not worth it for the stress.

I don't know how all the gates will play out, but a lot of folks may get stressed by this investment. I think 99% of folks should only be in liquid investments, including people in tech. You should be as liquid as possible. It's just too stressful. VTI is the perfect product for 99% of people. It's the perfect product. You cannot beat it.

One of the things that I just hate about venture as a founder is when I smell too much greed. Now, a little bit of greed is okay, okay? We're aligned. We're on the cap table together. Greed—there's a good greed, but there's a lot of VC today and in 2021 that's super greedy.

There's a super-greedy element to SPACs. That's why I hate them. You can tell me there are some good ones, but they're super greedy, okay? There are types of SPVs and opportunity funds that are super greedy, and I just don't like the smell of it. I just don't like the smell of it.

When times are good, being super greedy may be the right playbook in venture. Grab the billions. Grab the billions. But there's—I don't know—some of it smells too greedy. There's always that feeling when you go to the retail investor that you've exhausted anywhere else.

I get it. I get the cynicism. But again, I go back to what I said earlier. I think the weird thing that's happening—and this is weird—is that as private becomes more and more of the economy, it just makes sense that the big private houses find they have to access more and more of the public capital to feed the beast, right?

I get you, but a $1.2 billion fund doth butter no parsnips on that extension of private markets. You need $10 billion to be playing in that space.

Jason Lemkin

The way you get $10 billion is you start with $1 billion. So, yeah, look, I hear you. Again, I would say it again: just look at what the PE firms are doing, where private is a significant portion of their total raise.

My guess is, if you're running one of these big firms—and I'm sure there's a PowerPoint on the desk at Andreessen, at General Catalyst, and at Lightspeed on this, right?—if you're running one and you want to make sure that you're matching whatever it is that those guys are doing, that you perceive as your scale peers, small S, not us, then you're going to do what they're going to do.

They're all copying the PE guys. So, the move is inevitable. There's no point getting all moralistic about it. It's just the game they're playing.

Also, credit to you, Jason, for potentially the fastest bet to go south in a long time. I mean, you very confidently last week were like, “You know what? I bet you $100K that AI is going to replace jobs very quickly.” And then this week, Sebastian Siemiatkowski from Klarna, the biggest proponent of replacing people in Klarna with AI, goes, “Yeah, seems I went a bit far, and we're going to be hiring back a load of people to walk back a lot of that AI transition process that we made.” How do you think about that?

Jason Lemkin

Well, look, I'm still going to—listen, there are few things I actually enjoy more than being wrong. I enjoy it, right? I'm happy to admit it. I have no ego in it. I generally enjoy being wrong. Believe it or not, I'm thoughtful when I speak up, because I'm going to win this bet for a bunch of reasons.

This Klarna thing was misunderstood in the beginning, and it was misunderstood today, right? It was yesterday or the other day. It's misunderstood. First of all, there's some drama in what the CEO's trying to do to get attention for whatever reason, right? Drama with the sell.

But here's the point: this is what's happening. In every company that's in AI in B2B, there's a slider, okay? There's either literally or figuratively a slider. If you go into an app like Gorgias, where I'm on the board, where there are 20,000 customers using AI for SMB support, there's literally a slider and you can dial how much AI support you want from 0 to 100, okay?

No matter what anybody says on X and LinkedIn, the average across their SMBs is 20%, okay? Folks that put the energy in get to that 40% number that everyone talks about, but you have to invest time. Out of 20,000, they have a handful of folks that are at 100, like Klarna.

Do you know how many out of 20,000 are at 100? You can literally move the slider. What's your guess, Rory?

Rory O’Driscoll

Probably fewer than 100 people, maybe even 5 or 10.

Jason Lemkin

2. Yeah, 2 went to 100. And these are folks who absolutely knew the trade-offs and knew what the issues were.

This is what Klarna did: they moved the slider to 100, and they did not move it back to 100% humans. He did not say that. What he said is, “I'm bringing back some humans.” So, he just did what 2 out of 20,000 customers did at Gorgias. He moved the whole thing over to 100 to learn, to be dramatic, and it went too far, and he moved the slider back.

But you know what's going to happen every 3 months? That slider's going to move closer to 100. It's going to move closer. I believe in most cases it will never get to 100, but what will happen as we go into next year is folks will be like, “I'll deal with the downside of 100.”

It's only 2 today, but more and more folks will say, “Listen, some orders are going to be wrong, some answers are going to be wrong, but I'd rather have no humans in my new, cool 5-person, billion-dollar startup.” These 5-person, billion-dollar startups are still going to move the slider to 100.

So, I just think it's misunderstood. The fact that he bounced back somewhat—he did not say, “I'm rehiring everybody.” He did not say, “I'm rehiring the 1,000 people that I laid off in support.” He's probably going to rehire 200, and 800 will still be AI'd.

I think I'm going to win the bet, but if I'm wrong, it's cool. It's possible, but I literally see all the data across 20,000 SMBs and this slider. Once I realized it was only 2 that did 100, then I had my aha moment, right? That's too far, right?

But they talked them through it and told them what the downsides were. They walked through it, and they did extra training, and they're like, “We're still going for it. We're still going for it.” Those were products that were simpler, et cetera, et cetera.

First of all, Jason, you're not going to win the bet, because we're calling it now. You've been margin-called; we're closing the book. Okay, I'll send it out now. I'll send you the money.

But actually, having disagreed with you last week, I 100% agree with the way you outlined it this week, because frankly, it was more rational, right? Because you're right.

You know, stylistically, you're an entrepreneur; Seb is an entrepreneur. You have to move big organizations, and one of the ways you move a big organization is you create these big-ass goals, you violently shift the thing one way, and if you have to correct back a little, you do it, right? It's very Elon, right? And we can talk about his automation in the Tesla plant as an example of that. So, I think you're exactly right.

I don't know if the CEO of Klarna believed it and thought, “Maybe we can do it.” If he didn't believe it but thought, “We'll get to 80,” the only way to get to 80 is to try for 100. But it's unfolding exactly the way I would have said last week, and I agree with what you're saying now, which is: pre-LLM, in your customer support, B2B or B2C, you can chip away 20–25% of the number of tickets. Unfortunately, they were the easy tickets because it was, “How do I reset my password?” So, you actually didn't save a whole lot in head count.

Depending on where you are with LLMs, you can get to 50, 60, 70%, right, without any deterioration in service, and, in fact, based on some references we did, an improvement in NPS. For sure improvement in NPS, yeah, for sure, right? It's going to unfold. That's what's going to happen as the base case. Some people are going to try and do 100 and probably back off a little.

My guess is, in a year or 2, you're right. Maybe it won't be Klarna because they're big and they're losing money. But if you were running a small D2C company with a fairly simple product and were just really focused on it, I can see, in a year or 2, saying, “We just don't do support. We answer all the questions.” So, the direction of travel is clear, right? It's just going to take longer than people thought a year ago, but it's not going backward. So, I'm in violent agreement today. This is just a 5-year trend, not a 1-year moment.

6. Why Microsoft Laying 3% of Their Workforce Off is not Enough

I love Seb. I think he's fantastic. I've had him on the show, a friend and investor in Project Europe. So, this is all said with love. I'm surprised by the lack of strategic analysis around the timing of how he bluntly presented this.

Public markets—I had the founder of Duolingo on the show recently—very clearly said, “Hey, public markets take a very binary approach to AI. You're either an AI winner or an AI loser. Very simple. When we started, we were an AI loser. You saw that in the stock price: ChatGPT is coming for language. We changed our positioning around how we use AI. Content creation is powered by AI. Now we're an AI winner,” reflected in pricing. Okay, I'm summarizing, but really, I think it was well articulated there.

When he was going public, he was singing the song of, “We're AI-first, AI, AI, AI.” Now he's no longer going public, he is able to say, “Actually, we're not AI-first. I don't need to project that AI progressiveness right now, and I can save that for the next day when I do want to go public.” I think it's a very clear strategic message from a CEO who was about to go public needing an AI story. Very wise and quite right of him, and it's just the walk-back of that.

Jason Lemkin

You're probably right. You would know best. I think, in general, though, when I see a lot of these statements from public-company CEOs, I wonder if it works, but they're really also telling their team, “It's time to change now.” It's enough. And, honestly, a lot of big-company CEOs I talk to are honestly like, “I'm not sure I need 80% of my team today. They're just the wrong people.”

It's not 3%. Microsoft laid off 3% of their company today. It's not enough. It's not enough. The people running the 2018 playbooks are going to become almost useless going forward. So, you always see these comments from public-company CEOs saying, “We have to go harder, guys.” I'm not sure who listens to those statements, but pushing the dial to 100, people are going to listen, right? They're going to react. I think that's what it is, because every—the Fiverr guy is like, “My job is at risk.”

The Fiverr CEO, right? It looked dramatic, but I think he's right. So CEOs are trying to give folks shock therapy, and I don't know that anyone's listening, but hey, at least he said it ahead of time: “Your job is at risk.” The Fiverr CEO said, “Everyone's job at the company is at risk, including mine, with AI.” Right? I think Tobi at Shopify kind of implied that, too, that everyone's job is at risk.

And I believe it with my being. I don't think it's 5 years, as Rory said. I think by the middle of next year, in tech, almost every single person's job will have changed by the middle of next year. It doesn't mean there'll be mass unemployment, but their job will have changed. It's not going to change in 5 years. It's going to change by next year, right? And so CEOs are at least trying to tell people: before you get fired or laid off, you have to step it up.

7. Why Clay Should Raise a Warchest and Go to War

And maybe an interesting distinction here that I'm just internalizing, Jason, is this. We come from slightly different places on this, but I think we're saying the same thing. If you're in charge of driving change—in other words, if you're the CEO of a large organization—you have to make these hyperbolic statements, because otherwise it's just so hard to move 10,000 people. So hard, right? You have to put stakes in the ground. It's a management technique.

Look, I'm sitting back, frankly, as a small VC and analyst-type person. I'm saying, “I'm trying to be very precise about what percentage will be automated in the next 12 months so I can build my 5-year kind of expectations on revenue growth.” It's a very different thing. I'm trying to find the right answer. The CEO is trying to find the right answer for his organization to make progress. And by definition, that's a much more uplifting kind of statement.

I'm looking at my best CEOs as they're driving change here, and they're doing stuff like this. I mean, I'm thinking of one in particular. I got the missive at the start of the year: “By God, we're going to drive this.” And I was like, “Wow, that's what it takes to get through to people.” And I think that's what's going on here.

So, I don't think it's some mystery. Maybe there's some public-market messaging in the AI thing, I don't know. But I think it's just a CEO of a big organization trying to drive change. And the only way you can drive change is just push, push until something breaks and then swallow back a little.

Jason Lemkin

I think it's even harder today. I mean, literally, a CEO of a company that just crossed $100 million asked me, not that I'm any smarter, “How do I create more urgency in this AI age?” I don't even think it's about AI. I think it's about, as a CEO, how the hell you create urgency when we've had this level of urgency for years. I need to double our level of urgency because everything else is urgent today. How the hell do I do that with 10,000 people? How the hell do I create even 5% more urgency?

How do you? It's almost impossible to create more urgency at scale. It's literally—we talk about these issues. It's not that the incumbents don't know about AI. Look at HubSpot. Look at Box. They know about everything, right? But how do you create urgency? That's the problem with big companies: urgency.

What did you say?

Jason Lemkin

The only advice I gave to this particular CEO is: immediately—he was hybrid, so I said to force 100% of people to return to the office in 30 days and let everyone go except your S-tier engineers. Let them all go, because you don't need them.

If you're there from 6:00 in the morning, like 20VC, till midnight, like Harry is, Harry posted on Twitter last night, “My whole team's here.” I get the BST and the PST confused, but there are multiple messages in that tweet that Harry sent out to us. He's like, “Don't work for me if you don't want to be here at night,” right? Or whatever the hell. 100%. We are all in person and we work late, and we believe that the harder you work, the luckier you get. 100%.

But I don't know that you can do it for 10,000. I mean, maybe you can. I mean, Amazon's trying, but this is my only half-decent idea to create urgency: force everyone to come into the office. To be honest, we have a beautiful office in Palo Alto. I only go 2 days a week, so I acknowledge it's hypocritical. I'm not willing to go back, but it's the only idea I have.

8. OpenAI’s New CEO: Non-Technical CEOs Running OpenAI

Speaking of driving change, the single biggest change in 20VC and how we do what we do is simply ChatGPT across everything. This week, a couple of big bits of news: new CEO of Applications, Fiji Simo. I love Fidji. I had her on the show when she was at Instacart. Fantastic operator. How do we evaluate this kind of layered CEO beneath Sam, now CEO of Applications for OpenAI? What do we think?

Jason Lemkin

They weren't going to start being normal now, dude. We are 10 years into the least normal startup on the planet. Why stop now, you know?

Well, listen, is it really that weird? I mean, Salesforce has had CEOs of AI and co-CEOs. Salesforce has had it at their scale, right? $40 billion here. They've had CEOs of divisions for years, right, and co-CEOs. So, I don't know that that is so uncommon.

And there's title inflation—that's the name of the game. Every VP of Sales is a CRO today. So every CCO or even CEO or president wants to be a CEO of something, right? And you remember—obvious to state the obvious, even though it's implicit—is that you are hiring someone who was a CEO of a significant public company, and you're hiring them to not be the single CEO of a much bigger private company. I'm sure there's a dynamic around that as well.

I don't know, but I don't think it's that particularly crazy to give that title in this circumstance.

Jason Lemkin

I'll tell you what I think's weirder. Sorry to interrupt. And listen, I mean, OpenAI—I mean, ChatGPT has like 85% market share, right? And growing. I love using Claude.

We could talk about it, but I can already see how far it’s fallen behind, right? For me, it has some niche advantages. So, it’s won, right? I think the weirdest thing in this is that, for OpenAI, you have a CEO and now another CEO who are both not technical. I think it’s really weird.

Sam is obviously off-the-charts genius-level, no question. He can recruit like no one on planet Earth. Maybe that’s all that matters as a CEO. But my life experience is that nontechnical CEOs—now we have 2 of them—can’t win at companies like OpenAI, yet they are winning, right? My experience is that they almost all fail.

Why? Your co-CEO is now someone with a sales background. Fidji built on that ecosystem at Facebook unlike any other. Why does it need to be technical? I’m naive. Help me understand.

Rory O’Driscoll

Because the rate of change is so fast that I find nontechnical CEOs just can’t understand it. These are my biases in venture and my life experience. They’re really great at sales and marketing, and they’re really great at knobs and dials and spreadsheets and pricing increases, but honestly, Harry, no matter how many times you talk to me about RAG and vectorizing my content, I’m never going to understand how it works.

I could spend 100 hours with the smartest people in the world, and I will never understand how OpenAI works. Never, right? She’s smarter than me, but my point is that all the nontechnical CEOs that I’ve seen, especially those who take over my investments as outside CEOs, never understand the product.

Sam is S-tier—I mean, S-tier, right? And she’s S-tier. But it’s still weird to me that one of the greatest technological innovations of our lifetimes is now run by 2 nontechnical people. I still think it’s weird.

Rory, does it impact how you think about their expansion into the app ecosystem? You mentioned before, very wisely, chat, coding, and customer support. Does it impact how you think about that?

Rory O’Driscoll

A little, not a ton. Going back to that comment I made, one of the things you told me, Harry, is that doing this would make me up my own game on thinking, and you’ve been very right, by the way. I decided after that that the comment was wrong.

I think they will do it. I’ve been saying that they’ve done coding and they’ve done the app. I don’t know if they’ll do customer success because it’s so idiosyncratic, right? I think the app that they’ll do will be very broadly horizontal in a way that coding is and, obviously, in a way that chat is. You can envisage a huge amount more on the consumer side, with shopping in particular.

Absolutely. Totally. Yeah, it’s a thing.

Rory O’Driscoll

It is astonishing that the most compelling technical product in the last 10 or 15 years has been created with a CEO at the helm who was not technical. It speaks to some shrewdness by him and a strong focus on empowering that technical team, feeding them money, and providing them leadership. It’s a stunning achievement. We’ll look back and go, “Wow.”

It would be presumptuous of me to change the weighting on how likely I think it’s going to work or not. But I think it is going to work. At the margin, it doesn’t matter, I suppose, is what I’m saying.

They have won. They are huge, and there’s just so much more to do on the coding and on the consumer side. You’re right, ChatGPT is going to suck in all your brain. It’s going to suck in all your phone calls. It’s going to know everything about you. Remember when, way back when, you thought Evernote would know everything about you? You ain’t seen nothing yet.

It’s just going to take it all in, and you’re just going to be able to defer to it, right? I think there’s a ton to be done. You’re right, web traffic and shopping are just going to be huge things. So, I think there’s a ton to be done there.

It takes someone who’s not focused on the pure technical stuff and not focused on raising $500 billion, and it looks like this is that person.

Makes sense. Should we just all chuck our funds into OpenAI at this point? It seems like the easiest way to get a 3x. With a shortened window to liquidity, I’m just so sure that OpenAI is going to be a $1.5 trillion to $2 trillion company right now.

Rory O’Driscoll

Listen, I don’t—I remember, and I might have this a little bit wrong, but I remember I was around when David raised Craft Ventures Fund I, right? I think the next week he put a third of the fund into SpaceX. I had just a little bit of overlap with LPs, and they were like, “David’s crazy.”

I said, “Listen, I only know David as a founder. We were founders together. I don’t know him as an investor. This is one of the smartest guys I know. If he wants to put a third of his fund into SpaceX, I think it’s going to work, right?”

I think he just asked Elon for any extra shares. I don’t know what the story was, and I might have this a little wrong, but I think it’s mostly correct. It obviously worked, right? A third of Fund I seemed like a nutty move, but it was highly concentrated. Can’t lose.

Agreed. We just discussed Tiger half an hour ago, where it may well be that the saving move was doing exactly that.

Rory O’Driscoll

Yeah, I know. It’s not a crazy comment. Look, it’s not what my LPs are paying me to do, but in an open canvas, you say to yourself, that’s a compelling company—one that does have a huge, enormous market opportunity ahead of it. Obviously, it’s got to sort out its entire messy legal structure.

We’ve discussed this a number of times: the kind of single-bet vehicles, making 1 exception without making an exception across the team on everything. I don’t know, right? But you could just put the fund and put 25% into Anduril, 25% into Rippling, 25% into OpenAI, and 25% into—I don’t know, you name your other breakout. It’s not what your LPs pay you to do, but strategically, that is how I think we will have the most value created in a good time window.

I do believe we are in a world of fewer, bigger winners, and that rolls up through the entire venture ecosystem. If you thought your portfolio count had to be 30—a seed fund that had 30 deals—you need 45. We typically targeted 20 deals in our fund. I believe at this stage we need 25 to 27 or 28.

Why? Because instead of exiting at $150 million, we’re going to exit, best case, at $300 million, which is 2 or 3 more years, when 1 in 3 of these companies will fuck up, right? By definition, if you want to end with 3 or 4 winners, you just have to have more at-bats.

So, the moving up of the bar has had consequences that I think ripple up and down the ecosystem. Instead of us having 20 deals and having 4 great outcomes, we might need 27 deals but only have 3 great outcomes. They will be bigger because the good ones will have compounded from $150 million to $300 million, and the bad ones will have failed.

If you take that to the extreme, you’re at your point, Harry: fewer, bigger winners all the way up the stack to 1 huge winner. By OpenAI staying private for longer, it might get to a Facebook-type valuation before it ever goes public—Facebook today, not Facebook when it went public.

So, yes, what you’re saying makes sense at some directional level. The hard thing then to assess is that you can still, at the same time, overpay for growth assets and be wrong by 5 or 6 years, right? History is replete with examples of that.

You look at the Nifty 50 in ’68; they didn’t come back for 10 or 15 years. You look at 2000; the Nasdaq didn’t come back for 14. It is possible to take a good idea and push it to such extremes that you end up wrong. I just don’t know. I haven’t seen the numbers to say, “Look, is 27 times forward revenue for something like OpenAI the right number?”

Let’s take OpenAI: doing $4 billion last year, $12-ish billion this year, $360 billion in that. What’s that? Is that—let me see, $12. I mean, I don’t know. What kind of revenue multiple are you looking at? What kind of compounding are you embedding there?

I looked at it side by side with Google, and it had tracked virtually the same to now. But if you look at their forecast for the next 3 or 4 years, the leaked forecasts are showing higher growth than Google. At some point, you’re overextrapolating into the future. When is that point? I don’t know.

It’s not a stupid idea, but obviously you haven’t done it. Why didn’t you? He’s thinking about it. If you were just about making as much money in the most efficient way possible, would you do it the way you’re doing it?

Rory O’Driscoll

I think the problem with private markets is that you put all your effort into being able to do 1 thing well, and the cost of that is that you don’t focus—you’re not equipped to do something else, right? It’s not been something I’ve focused on, so I don’t think, starting from here, it would be—I don’t know if it would be as easy to be successful at it.

For starters, let’s start with the basic comment. Let’s push on this. OpenAI said in the last round, you either show up with $250 million or don’t show up at all.

Right? Remember that?

Rory O’Driscoll

Yeah, and that’s a good example of the markets. Basically, to play in that space, you have to be equipped to play in it. You have to set out with that as your objective, and you have to equip yourself as a fund with the ability to write 4 $250 million checks. Otherwise, you’re not going to get to see the deals.

So, I just think, sitting here with my $30 million check size, I can make a phone call, but I don’t think they’ll get back to me. I’ll leave a voicemail. Jason’s AI will, at 100%, just look at my customer support ticket and say, “Not worth replying to.”

Jason Lemkin

It always replies.

But you’re right. It replies and says, “Dear Mr. O’Driscoll, you do not meet the accredited investor threshold at $250 million.”

Rory O’Driscoll

You are a dick, Harry.

No, I’m not a dick. I’m not. The question that we brought up there was the nonprofit versus the for-profit, and this was a big point. Now, they’re still a not-for-profit. I thought the whole point was they were moving to a for-profit.

Jason Lemkin

Now it looks like Microsoft got a great deal. I mean, that’s the way venture works, but getting everything they get, plus 10% of all revenue until there’s AGI, plus the returns—I see why Satya was sitting pretty and wasn’t really concerned about all this.

You’ve not only got a situation where he could afford to lose it, but in the end, what looked like a weird backdoor license and M&A deal is an epic investment, right? Not that I think Microsoft cares about it as a financial investment. It can’t care, right? But that’s the irony.

I guess they have to work it out. At some point, it’s like the U.S. and China. At some point, it’s going to be messy, but they’ve just got to work it out. Sometimes things don’t work out.

Rory O’Driscoll

Yeah, but staying with this, I mean, look, I think it’s funny. These things come in—you know, there were probably a bunch of great reasons for doing it at the time—but the original for-profit, not-for-profit structure, and then the Microsoft structure around that, have now left you with this interesting conundrum, right?

I think it’s clear they’re not going back. To be fair, they’re not going back to a not-for-profit. They’re not doing that. What’s happening is they’re not going to just convert the underlying company to a classic for-profit. They’re going to convert it to a public benefit corporation, like Patagonia, where it’s basically just like every other company, but you have, in your articles of incorporation, a specific obligation to look out for interests other than the shareholders.

The good thing about that actually strikes me as a decent solution at that level, because what it says is, “We’re going to be trying to make money, but at the same time, we don’t have this binding obligation just to maximize shareholder value.” So, when you’re public, you won’t get pounded by the Delaware lawyers and told you can’t do this or that because it’s not in the shareholders’ interest.

It’s a useful legal structure. As I say, Patagonia uses it. It’s not something they pulled out of thin air, right? So, the idea is you drop the assets into that entity, and then at the whole-company level, you have the not-for-profit—which I think is a foundation—not-for-profit. You have significant board control and a significant economic stake.

That structure makes sense, and maybe it’ll be a little less contentious in one sense than having the pure subsidiary be for-profit, right? So, it’s not like it’s totally failed. It’s like they’re trying to do this.

But to your point, the hard part is that everyone has a veto. To Jason’s point, the Microsoft deal is so weird and so clever that unwinding it is difficult. I mean, you’re basically asking: How do you value what percentage of a company this represents?

Imagine if Jason did a deal in one of these companies and said, “You’ve got to give me all your profits until I get my $20 million investment back. Then I’ve got to get a rev share, then I’ve got to get something else with a cap.” And now you’ve got to convert that to a percentage ownership in a simple deal, right?

I don’t know how. That’s an interesting one, because you do get into this situation where 2 people have to agree, and I’m not sure where the leverage is. I think Microsoft’s leverage might be pretty strong, because the current deal is pretty interesting and attractive for them.

You’re unwinding something that maybe made sense at the time, but it feels a little clever now. They have a block, and then separately, you have the whole risk of litigation for me, even under this structure: Is it still not-for-profit enough to do it?

So, there’s a lot of wood to chop yet, and a lot of Delaware lawyers, California lawyers, and New York lawyers who are going to put their kids through college on the litigation. A lot of kids. A legion of children are going to go into college on this one.

I don’t know what the legal bills are probably $20 million a month or $10 million. They could be $10 million a month. $20 million sounds high. It could be $10 million a month in legal bills, right?

In the end, the interesting thing is—and you prompted me to read about it—that the public benefit corporation structure is actually what Anthropic did from day 1. They got it right. OpenAI is basically saying, “We’re going to go for something like the structure Anthropic already has.”

So, it’s obvious that’s the endpoint that just about works. Somehow they get there, but God knows how.

We’re going to speed through a couple of companies where there were notable announcements and news, and then Jason’s got SaaStr. Big week, Rory, for Jason. Go SaaStr. I’m honored he even came. Honestly, what a freaking hero.

9. Why Big Funds are Investing in Perplexity

Perplexity, staying on the theme of chat and LLMs: a rumored $500 million round at a $14 billion valuation, rumored to be led by Excel. Any thoughts?

Rory O’Driscoll

Well, listen, I’m not a total expert, but I do like Perplexity. Perplexity, of the leaders, is the closest to a Google on steroids, right? They announced they were at $100 million ARR recently, right? So, let’s say they’re at $200 million today, which is still generous growth, right?

Does it justify this, or is this a distant number 3 getting a massive premium? I don’t even know if the Anthropic valuation is justified. They may be getting too much of a valuation premium compared to OpenAI. I just don’t know if it deserves it.

In a world where we have a rational M&A strategy—and it’s TBD, if you will, under this administration—the big picture here is that this is an at-bat, a 1-in-3 at-bat, at a trillion-dollar company.

Google’s a trillion-dollar company, and this is a 1-in-3. There are only 3 companies that are going to be relevant here: OpenAI, Anthropic, and Perplexity. The mere fact that you’re in the arena, to use a phrase, and you’ve got users and revenue growth, and you were first with the idea of combining web search and an LLM—you got that done and got it out the door. That gives you the right to play the game, right?

Maybe you can build a standalone thing and get public. In a world where, frankly, IPOs happened like they happened in 1996, 4 search engines went public in 1998. You don’t remember any of them except Yahoo, but it was Lycos and Excite. I can’t even remember the 4th. Poof, right?

Because it was a big, obvious market, it was happening fast, and all of them had a chance to be big. So, going back to that upside-junkie comment I made last week, what Perplexity is selling from an investor perspective is an at-bat—a credible 1-in-3, not equally weighted, to be clear.

OpenAI is clearly going to win, but maybe you can be 3rd, and that’s worth the downside if the prize is $1 trillion. That’s what they’re selling.

I am actually an investor in Perplexity. I think I have to disclose that, otherwise my CFO will kill me. But that was exactly my thinking: How many companies have a credible chance of being a trillion-dollar company, where it is still very, very, very small—very small—but still a credible chance?

And to put the other side of it, in a world where you couldn’t do M&A—which may not be the case today—there’s also a whole bunch of big companies that need to be credible in the space, too. So, plan B shouldn’t be awful.

Now, as I said, the weirdness of M&A in tech for the last 5 years has been head-bangingly depressing, right? But there’s no doubt in my mind there are a couple of trillion-dollar companies thinking, “Maybe I’d like to mess with Google’s head. I need to have this.”

Rory O’Driscoll

100% agree.

Incredible team around them, to be fair. One thing that no one else sees is that their distribution strategy has been very smart in terms of large partnerships. They've done quite a few in Europe with big telecom providers, which gives them default access to consumers' phones. That is very smart. If you look at Google in the same way in the early days, they did exactly the same. It was distribution through partnerships. Very smart.

I realized that the investment memo should be updated. The first paragraph should be the odds of a trillion-dollar outcome. I've just been using an old template. You think I'm being facetious, but I'm using an old template. The first deal I ever did was Pipedrive—we talked about it years ago, right? I remember when I wrote up the investment memo, the odds were of getting to a $100 million exit. That was the old structure of it.

I was forced by the traditional VC firm to fill out the numbers, and I said, “5% chance of a $100 million exit,” right? It did sell for $1.25 billion. But I'm just using my 2013, first-investment-ever lens, and I did 5% now. Instead of $100 million, I mean, then it should have been $1 billion and then $10 billion, but maybe there needs to be another row, which is $1 trillion. What are the odds in the investment memo of $1 trillion? If it's north of 2%, we do the deal. If it's north of 2% odds for $1 trillion, we do the deal. I think some version of that is true.

The opposite is, if you're looking at a deal that has all the risk of a classic private company but doesn't have embedded upside beyond the base case, you probably shouldn't do the deal, right? We're all, as I said, upside junkies. It's the pixie fairy dust that lands on our portfolio every 5 or 6 years that makes the math worthwhile, right? If it doesn't have that, you've got to ask yourself why and whether you should do it. This one has it in spades because it's the biggest pixie dust. It's a trillion-dollar pixie dust, not a $100 billion pixie dust.

In a similar-ish ilk, I was thinking actually about Coinbase. You're going to do more? You're going to do super pro rata in this alleged sell-around? I just don't know. Are you doing super pro rata, pro rata, or none? What are you going to do with your check?

Rory O’Driscoll

My business is not to do a $14 billion priced round.

So, after giving me crap half an hour ago about not listening and just doing what the right thing to do is, you're saying it's not your business. I just want to note that moment. I accept it, but, Rory, before the visuals have been made, do you want to pick this fight, my friend?

Rory O’Driscoll

No, very fair point, but $14 billion is not my game. I'm thrilled that Excel are rumored to be dollar-cost averaging. That's the thing. As an existing investor, you don't have to take the nominal price too seriously. You're dollar-cost averaging it.

The interesting thing is the other one, and then we'll let Jason run, is actually Clay. Clay obviously did this kind of $1.5 billion-priced transaction. I think it was employee secondaries with Sequoia. My thinking there was that people are now believing that Clay is the next credible threat to Salesforce, and that is why people are getting so excited by it. It's actually much more than that. Okay, we're shaking our heads. How do we think about this?

Jason Lemkin

Maybe that's what Sequoia says. What I'll tell you is just 2 things I see in the field. This is pretty close to our core audience, right? It's Clay. First of all, Clay is a great success story. Let's simplify it to that, right? There are a couple of things driving Clay. Certainly, there are super-sophisticated, cutting-edge teams, right? But there's also every struggling marketer today who doesn't understand AI and wants to deploy Clay. There's massive pull.

Every 2021 CMO that's scared they're going to lose their job—nothing's working, search and SEO aren't working, content isn't working, everything's down—they're hiring Clay consultants and Clay engineers, deploying it, and writing checks like Hopin in 2020, okay? It's literally like Hopin in 2020, but for marketers. And I think they earned it, right?

I don't know about Salesforce; maybe that's the vision. I'm telling you what's happening on the ground: just like people needed a digital event solution during the pandemic, marketers need an AI solution now. Their jobs and lives are at risk. This is what I'm seeing. Literally, we were at a SaaStr pre-event last night, and I'll tell you everything I heard about Clay.

That's what everyone's saying. Everyone's saying, “We're going to have a CMO summit on Thursday, and I'll ask everyone.” Clay is the winner. If you're scared of AI, it's unsustainable. We're going to have 400 CMOs on Thursday. If you come, you can sneak into the CMO summit and ask them. You can ask the question for me: How many of you are using Clay? Raise your hand. How many of you are using it because you're scared of AI? I bet if it's a closed-door meeting, they'll say, “Because I'm scared.”

The only question I have for Clay is, I've never seen more great founders than I met last night who have Clay in their sights, right? They're going for it. Everyone wants an easier-to-use Clay. I want Clay with better data sources. I want Clay that can be self-serve. I want Clay for this. I'm not saying Clay won't win, but this is literally a category where the knives are out. It doesn't mean they don't already have scale and won't win, but they can't rest on their laurels.

The 1 thing that's clear is they can't rest. I literally met the CEO last night. He's like, “Hey, are you Anna Lemkin's father?” I'm like, “Yeah.” He said, “Well, I just dropped out of Stanford with your daughter. I have a new competitor to Clay already at $2 million.” Wow. He just dropped out. I'm like, “This kid—he's 19 or 20, with my daughter. What do you mean you're at 2?” He's like, “Well, honestly, we're just much easier to use, and we have better data.” It's the story I've already told you, like, the 3rd time.

That's $2 million in a couple of months, right? So, Clay has won this Q1 2025 version of the marketers' platform. I'm not betting against them. But, man, everyone—it's like it took 4 years for folks to figure out why Gong was successful. Anyone in sales adopted Gong and they're like, “I can actually listen to my reps' calls—the dumb things they say to prospects.” But it took everyone else about 4 years to figure out why machine learning on voice calls was disruptive. Now it's happening in days with AI, right? Or weeks—they're figuring it out. Competition's going to be insane in the space, is my summary of my rambling point.

Insane. So, I would raise even more. Forget the secondary; I would keep it as primary because I'd build up my war chest.

So, you raise it as primary?

Jason Lemkin

I'm just saying the competition is so high, right? I would just build. We started this conversation on Owner and why Owner raised $120 million, right? Part of it is because, even though it pulled away from some of its venture-backed competitors, it's such a competitive space, right? I just think I'd love it if you could win in a competitive space and be cash-flow positive with 50 employees, but we have a lot of history of that playbook not working, right? So, I would armor up if I were Clay. I would hire everybody. I would raise another $100 million, and I would just scorch-earth everyone in the space.

Rory O’Driscoll

It's worth pointing out that the original Clay product is a pre-generative-AI product, all around combining multiple data sources and manipulating those sources. It was a horizontal, spreadsheet-type functionality initially, then focused on marketing. The problem it solves is, “Hey, you need to build a 10,000-name list of B2B CFOs between 50 and 500. Do you use ZoomInfo? Do you use People Data Labs? Do you use one of those sources?” No, you use Clay. You import from all of them, and they brilliantly consolidate the 4 or 5 different entities and give you a clean, combined list. It's a RevOps function that was pre-AI, which they solved brilliantly.

That's what the product actually does, right? Now, what they've added is these Claygents, or Clay agents, in the last year or two that take that list and start doing emails, start doing enrichment, and start doing all the agentic work after that. So, they're moving down funnel from list management to enabling AI SDRs, right? It's a totally excellent product. It hit the sweet spot of the need.

It initially was not an “Oh my God, AI GPT” story. It's something totally different, and it's so far from Salesforce CRM that I can't even think about it right now in any useful fashion. So, that's what the product actually does, which should never get in the way of a good story from a VC, obviously, but it does it really well. It fills a need. You're right, too: they have the chance to run fast on top of that and do a bunch more. The roadmap is pretty clear because you've got this AI lift, and now you've got clarity on what the next 5 set of automations could be. But there are lots of people doing that. We have investments in that space. I'm sure you do, too.

There's a whole ton of AI automation coming to the top of funnel sales and marketing. That's what's going on now. But being the vendor to take advantage of AI fear is a great strategy in the next 18 months?

Jason Lemkin

Yes. Everyone's job is going to be at risk.

Tying this conversation together, Jason, is that sustainable?

Jason Lemkin

I don't care if it gets me to 300 million in revenue. The world changes. I want to be right now, if you are the CFO and you're scared.

Hopin was at 200 million in revenue?

Jason Lemkin

Yeah, but we did come out of the pandemic, Harry.

Rory O’Driscoll

No, no, no. I'm with Jason, Harry, because I think that you can use that fear. I'm thinking of some companies where, genuinely, when you did references on them 2 years ago, you talked to the very few customers and they'd be like, "I'm buying this thing because I need to do something in AI." You fast-forward to today, and they would say, "2 years ago, the product's not great." You fast-forward to today, and the same people would say, "The product's improved a lot. I'm now getting real value from it."

But they sold on the fear well ahead of the product. I'm always biting my tongue to think of companies—do not name companies—but I have a couple in mind in the relevant professional services and white-collar spaces. They've gone to finance, gone to legal, gone to folks like that, and it's been very airy-fairy at the start of GPT, but they got in the door, told the big story, sold the fear, and then, over the last 2 or 3 years, built a product.

Jason Lemkin

It's different than Hopin, where the fundamental end demand died.

Rory O’Driscoll

Yeah. If you seize ground but then the ground turns out to be waterlogged, you're toast. But if you seize the ground and it matters, then you win.

Jason Lemkin

I think there's an element of the AI fear story that has been real in the last year and probably for another year or so. I do think people that have budget will spend a lot to not lose their jobs. So if you're that one app for the CRO, the CMO, the CCO, or the CFO where, if they spend six figures on you, the odds that they won't lose their job increase, you will find budget.

You not only will find budget out of the budget cycle, you will find budget this month. You will find six figures of budget: "If I buy you, I might not lose my job." I'm just smelling this with CMOs and Clay. I'm just smelling this fear: "I'm going to lose my job, and all I have to do is write a six-figure check. I'm in." I'm all in. "My budget's 5 million, 10 million in marketing this year. All I have to do is come up with 200K for Clay. Done. Send over the contract tonight."

10. The Impact of AI on Marketing and Sales

In the end, you have to deliver. I mean, Harry, you're right. You've got to deliver the thing roughly that you said you're going to deliver, not something different. But it's a totally legitimate tactic at this point in the cycle, provided the product gets there in the end.

Seriously, I know we've got to wrap, but I can't tell you how many marketers, if they're honest, are looking around when I talk to them and they're like, "I don't know what my team's going to be doing in 6 months. I just don't need this writer. With ChatGPT, with AI, I just don't need my team."

Then everyone's looking for the AI wizard, the AI magician. To the extent this person exists, they're going to work for a super-hot startup. That's the other fear with the CMO, right? They're going to work for Windsurf, right? How is your B2B company, growing 19% this year at 81 million in ARR, going to get the AI wizard, even in marketing, to come to your company? You know what? There's no chance, right?

So you need this Hail Mary to save your job because my infographic person needs a week to make an infographic, and my product marketer takes 30 days to turn around a proof for product marketing—and it's terrible, right? You're looking at your team and you're like, "I don't even know what to do with these people anymore. I don't know what to do with them." So I'm going to buy Clay.

I'll tell you the one thing that is so far off: video. Video and AI capabilities around video editing, in particular for complex video, are so far away today. But Higgsfield is so good, man. My jaw drops with Higgsfield.

Now, listen, I don't think you can produce 20VC with AI. Don't get me wrong, and that's a good thing, right? But I don't even think there's a point in having a static marketing image after Higgsfield. I don't see a point. It's so good. I think within 2 years we'll still be at 20VC level for my video team.

Jason Lemkin

Yeah, you'll need it, right? But I think even when I look at Opus, where I invested, in some ways you can make fun of Opus because you'd be like, "Now everyone can do clips, right?" I mean, other folks do it, but Opus democratized clips, right? The problem is they weren't nearly good enough for us.

But I'll tell you what's interesting. Opus has a feature that's coming out which, for you, is not valuable, but for me, now they'll create a clips video and search all your videos. I have 12 years of video. We're going off to SaaStr in 5 minutes. We're going to produce 300 pieces of content. You think I can make a 1-hour video out of the best of all these speakers?

For 20VC, worthless. Okay, worthless. But for a CMO that's still struggling to get their Zoom webinar working, that's a big change.

But it does not make my content more valuable, sadly, Jason. It actually makes it less valuable because it makes discovery a massive problem. Now we have an infinite supply of content, and discoverability is a problem.

Rory O’Driscoll

Agreed. But that's why you're into that. Make money out of this trend. If you're investing, make money out of the fear. Make money out of fear.

The 2 ways to make money are: one is to sell the tools, so everyone can produce this content at scale. Then, back to your comment, Jason, now I want to do it with my people. I do believe the marketing people who know how to use this stuff not only continue to have their jobs but become even more useful.

Again, it's the same: you've got to be on the side of using this shit and leaning into it. If you're resisting it, you're toast.

But you're right. The honest truth is that it's far fewer people than you can almost even talk about in public. So few people are on top of what AI can do. I'm just going to go back to hiring agencies, and they're going to lose their jobs.

Jason Lemkin

Now you're doing the clown thing again, dude. You slide the scale to 100. I think the scale advances steadily at 5% or 10% a year, and yes, you're going to need fewer people. You're going to get more efficient.

Maybe it's 20%—5% or 10% a month.

Jason Lemkin

But Harry still makes his stuff handcrafted with care in Central London. So there. Trust me, I'm with Rory on that for video and for me.

I love the balance that we have: the beautiful nuance of Rory and the binary U.S. bravado of Jason coming together is just wonderful, and then me just sitting in between. Other than Claude 2, where I just unleashed.

Guys, this was fantastic. Jason, good luck for SaaStr, too. Rory and I are rooting for you. It's going to be a special one.

Jason Lemkin

All right. Well, we'll see you at the one in December in London. Christmas in London. We'll all do it. We'll do it the first week of December together.

Thanks, guys. All right. Talk to you later. Bye.