OpenAI与AMD达成数十亿美元交易；Polymarket、Vercel和Supabase完成巨额融资

[Speaker?]

Paul Graeme was right. Sam Alman understands power. He has more power than AMD, so he took 10% of their company for the privilege of selling stuff to him.

Nvidia is making so much money, it's almost incomprehensible. So, I think Jensen knows he's got to give up some of it. The more you do this, the more you just say to yourself, “You just need to do big, exciting deals in trends that are absolutely obvious.” Every time you try to make it harder than that, you lose money.

I am so looking forward to this. We have a lot of news to cover this week. I did laugh when I was putting this together with Jason's brilliant suggestions, thinking that this could also just be called “This Weekend: OpenAI.” That could be a relevant name.

1. OpenAI and AMD's Major Partnership

Obviously, we had Dev Day yesterday. I want to start, though, with the AMD deal. OpenAI announced a major chip supply partnership with AMD. OpenAI will buy AMD's upcoming Instinct chips, up to 6 GW. As part of the deal, they get warrants to purchase up to 10% of AMD. Big news. If we start there, what the fuck does this mean, a week after Nvidia invested $100 billion in them?

Rory O’Driscoll

It means a lot. There's a ton in this. First of all, whoever does corp dev in OpenAI gets a bonus this year. Bizarrely enough, whoever does corp dev in AMD and Nvidia also gets bonuses this year, because so far this has, oddly enough—and I'll come back to that—a win-win.

Let's do each of them in turn. I might start right down in the weeds with the AMD deal, right? Contrast that with Nvidia, because Nvidia is strong when they get equity in OpenAI in return for giving OpenAI chips. OpenAI uses that money to buy those chips.

AMD, because it's weaker, has to give its own equity to OpenAI for the privilege of having OpenAI buy its chips. It speaks to OpenAI clearly, and this is why Paul Graeme was right: Sam Alman understands power. He has more power than AMD, so he took 10% of the company for the privilege of selling stuff to him. He probably has less power than Nvidia, so he let them get equity for the privilege of selling him chips. It speaks to the hierarchy. There's clearly dominance being established. That's the first thing out of the gate.

I just want to understand: he got warrants to purchase. Is that different from him purchasing, and is that different from—

Rory O’Driscoll

It is and it isn't. If you look one level down, what did he get? OpenAI got warrants to purchase 10% of AMD at a penny. In other words, they're free warrants. We'll come to the accounting on that in a second, but only if certain conditions are met: they buy the chips and the AMD stock price goes up.

I can imagine the discussion. OpenAI comes in and says, “Hey, AMD, we're going to buy some shit from you, and that's going to be so good for your stock price that we want warrants to do that.” We may not have mentioned that. They said, “We want warrants to do this deal.” The AMD guys say, “No way. We're selling you chips. We're getting money. What the hell do you mean you're going to get warrants as well?”

OpenAI said, “I bet you your stock will go up just because you're doing business with us, because you're kind of a no-hoper and now we're saving you, and we want to get some of that upside. Therefore, I'll repeat it: we want the warrants.” AMD eventually says, “Here's the warrants.”

Remember, AMD has got to—OpenAI has got to buy the chips, step 1, and step 2, the stock price has got to be high. The interesting thing is that we'll see if that lasts when those chips get shipped. But as of right now, the stock price went up 30-something percent. So, if they were shipping the chips today, they'd be getting the warrants today, right?

OpenAI was correct when they looked AMD in the eye and said, “Dude, your market cap—we're going to get 10% of this company. I think that's worth $30 or $40 billion because it's a $300 billion company.” Your stock price went up $60 billion, so you're up.

I don't know what the goal was, right? It's always interesting, these warrant deals. I've had a few in my portfolio I've been dealing with lately where they're just immaterial sometimes to the company. Sometimes here, $30 or $40 billion is not immaterial to OpenAI. It could flip the stock, right, if it's allowed to. It could be material.

I don't know what the goal is. My initial sense, which I think may be wrong now after what you said, was, listen, we're AMD; we need to diversify away from Nvidia. OpenAI it is, but we're going to do much more for them in the short term than they're going to do for us. Then you're going to turn around and monetize it with our competition. You're going to sell them all that you want, right? That's how I read it. We don't want egg on our face.

It reminded me at first of Shopify and Stripe, and Tobi was so angry in his mind that he put Stripe on the board. Then he went to Klaviyo and said, “Listen, I'm going to do the same thing for email that I did with payments on Shopify, but you've got to give me 10% of your company.” It felt like not getting egg on your face, but maybe they just want the money.

Rory O’Driscoll

Yeah. I mean, it's all together. They came with a kingmaker package and made them an offer they couldn't refuse.

A kingmaker package, right? It's a king package.

As I say, as of today, it would work. But remember, they don't get those penny warrants until they ship the chips and until OpenAI buys them. Whenever you do these vendor deals, you're worried: are they really going to buy the shit?

Two years from now, OpenAI still has to need whatever vast sum of chips they said. If you zoom out a level—and I was genuinely thinking about this last night—what you're seeing here is the Windows–Intel game beginning again.

If you zoom back 30 years, everyone has shifted to Windows. Microsoft—well, actually, I shouldn't say DOS—was the software company that took control of the PC monopoly. Their adjacent partner was Intel. IBM was the old-school company that set them up by doing that famous DOS licensing deal, and AMD was the little player that got dealt in to 10% market share because IBM said, “Dude, we're not going to just rely on Intel anymore. We need a second supplier.” That's 30 years ago.

2. Microsoft Have F***** Up the OpenAI Partnership

The way it unfolded is Intel did well, Microsoft did better, and AMD got a little bit of money, while IBM faded away. Fast-forward to today, the company that's dominating everything is OpenAI. They're the Microsoft of today. They have the consumers. They have the eyeballs, right? They're building this new monopoly.

The equivalent of Intel is now Nvidia. In other words, the only other company that OpenAI needs to make all this shit happen is the chip company. If you've got chips and you've got these guys, you're golden—and obviously, capital.

Nvidia occupies the role of the dominant other part of the duopoly, the two-armed people running the thing. In other words, you've got OpenAI and Nvidia. But bizarrely enough, you still need the second source. In this case, OpenAI is driving that agenda.

Up comes AMD again, 30 years later, with exactly the same shtick: “We're not as good as Intel, we're not as good as Nvidia, but we're here with a second source. Give us some money.” History repeats itself.

And just to say it, the role of IBM has been played by Microsoft. They set this whole thing in motion and they've allowed it to exist. In one sense, for a while, it looked like Microsoft got a good deal because they got access to OpenAI's technology early on and they got a little bit of buzz.

Brutal commentary.

Rory O’Driscoll

Just like IBM got a buzz when they shipped the first PC, because they got a product out the door, Microsoft let this competitor emerge in their midst. I think, going back to Dev Day, it's uncanny how similar it is.

If you're sitting there now and you're Microsoft, you're thinking, “Did we just create a monster?”

Sorry, sorry. Did I own a large chunk of the monster they created in the day?

No. Look, exactly. History, as someone wisely said, doesn't repeat; it rhymes. IBM does not own a big slug of OpenAI. It's a good point.

As we've discussed, you don't get points for venture capital when you're a dominant monopoly. You've got to just stay a monopoly, right? So, yes, better to own 10% or 30% of it than nothing. Great corporate development deal for Microsoft—far better corporate development deal for Microsoft than IBM did 30 years ago, when it didn't take any ownership. So, as we said, the corp dev guy at Microsoft also gets a bonus this year, right?

But from a business perspective, we're going to talk in a second about Dev Day. What you saw is OpenAI basically saying, “Here's the place in which you should run your other apps.” Now, I'm not sure that vision sticks, but if you're Microsoft, you're like, “WTF? We're the place where you should run your other apps. That's what we do. Who the hell are you?”

It's not a complete parallel, but there's a lot going on here that feels uncannily similar, and you kind of have to think about it.

Can we just stay on the deal itself before we move to Dev Day? If you're Nvidia, are you not thinking, “Hang on a minute”?

I thought we had this trusted relationship, this wonderful partnership. We just gave you a ton of money. We just invested. What a strategic relationship—and now you're turning and biting the hand that feeds you. How did it feel?

Jason Lemkin

Well, Sam Alman is—again, we've said it a million times—I learn a lot watching what he says because he is thoughtful and direct. He was very careful to be complimentary to NVIDIA and be clear that they were his number-one vendor. Here's my view; I don't know what you guys think. Listen, Rory's made the point: the only person making any money in AI is NVIDIA. Even Oracle isn't making any money. OpenAI certainly isn't making any money. NVIDIA is making so much money it's almost incomprehensible, right?

So I think Jensen knows he's got to give up some of it, right? I think there's this elaborate dance of losing, of chipping away. He's got to give a little market share. He's got to be a little polite on these deals, as long as at the end of the day he knows he's going to have 90% market share. That's my sense. He's playing it—they're all playing it—very carefully so that he can minimize his price erosion, which he has to deal with, and maximize his market share without creating a huge conflagration.

I think it was very carefully orchestrated. This may end up being very little, right? If AMD isn't fully competitive, no one may end up using these chips except the minimum they need to maintain competitiveness, right? So I don't know. I thought it was very thoughtful about everybody. I thought it was sequenced in the right order. AMD didn't come before NVIDIA, did it? Certainly it wasn't announced first, and everyone showed up to pay homage to Jensen, and he referenced it when he did the NVIDIA-OpenAI deal. I think it was polite. What we can see from Elon is that being impolite has consequences in this space.

I mean, that Elon guy—he hates Sam, doesn't he?

Jason Lemkin

I don't know if xAI would exist if it wasn't for his bone to pick with Sam Alman. He might not have bothered. He might have just gone to Mars faster. But staying with that—and you're right—I think, staying with the dynamics of the deal and what it reveals, you're right on the chip side: it reveals they're a wildly powerful company.

People are going to—no one's going to see a $4.5 trillion market cap, $200 billion in revenue, 50% operating-margin company and do anything other than say, “Get me some of that.” Right? So you're right; he's brilliant at playing out his hand. The interesting thing is not the perception of leverage that NVIDIA has, because the leverage is real and obvious. They allocate the chips.

The interesting thing is the leverage that OpenAI has, even though they're losing a shit ton of money, right? Precisely because they have the users. When you have the users, even if you're burning cash, there's no place NVIDIA can put chips other than to a customer who themselves has enough users to use all those damn chips, right?

And therefore, the stunning thing here is the asset: you can be sitting there losing money hand over fist and still get credit for committing hundreds of billions of dollars you don't have. You can bestow market cap on your vendors, for God's sake, just because you're willing to buy from them, simply because the whole world right now believes, rightly or wrongly, that that $12 billion revenue line is going to get to $200 billion, and it's going to take $100 billion a year in chips to do it. Therefore, selling shit to OpenAI is a business so good that you're willing to give up 10% of your company for free for the privilege.

Here's the weird thing, because I have actually, in my first startup, sold components, and NVIDIA is selling components at the end of the day. It's a crappy place to be on the stack. You know what everyone does when you sell components? Everyone's nice to you because they need you, and they bring you into the conference room, and they bring you coffee. You know what they all say? “Harry, cost plus 20%.”

Now, you can laugh about that. But when you look at NVIDIA with 50% margins, and you're buying from them, you can't help but feel that's the number-one place I'd like to attack if I could. I'm fine if NVIDIA makes $0.20 on a dollar, $0.15, right? But 50 cents? You're just—I mean, F me. If there was—there used to be competition in the GPU market; there just isn't today, right? So it is this weird dynamic where normally you'd be beating up on your vendor. Your vendor would have lower margins than the software provider. Here, it's highly inverted. It's highly inverted.

To play it out, because I thought I was going to disagree with you, but in the end, I'm insane, because what you're saying is this: normally, components businesses are hard because everyone understands the cost structure. You typically only have a few customers. It's like selling telco equipment to the telcos. There are only 20 big telcos. They know they have you over a barrel, and therefore those businesses become pretty tough, right? Because you're right, they just calculate cost and work back in, right?

Rory O’Driscoll

But 2 comments. One is, the only thing that defeats that is architectural lock-in, where you have a monopoly. What you're seeing is a monopoly competing against an oligopoly, with the monopoly provider being NVIDIA. As long as they're a monopoly, the buyer from OpenAI—or Microsoft—can sit there and go, “You bastards, you're only paying $50 a chip to TSMC and you're charging me $300. I hate you. I'd like to do it for less.” And NVIDIA sits there and goes, “Well, we won't.” And you've got no other choices.

Jason Lemkin

We will, but we're sold out, Rory. We're sold out.

Rory O’Driscoll

So, yeah, maybe in 2031 we could provide you with some of those chips. The fun thing about this semiconductor business is—you know, it was about 20 years ago—venture effectively walked away, with 1 or 2 exceptions, from semiconductors, and they were probably correct, right? Because from a startup perspective, it got really hard around 2003 or 2004-ish. There have been a few since then. One of my colleagues did the deal—not me—but we had one of the last Monolithic Power Systems deals, which was a success, about 2007 or 2008. Since then, there's been almost no venture exits.

In venture land, at the same time, in public land, it's been wildly profitable. You've got NVIDIA, you've got Broadcom, you've got Marvell, you've got a bunch of others. They've basically consolidated, to your point, Jason, so that the remaining providers have significant leverage. If you're going to only have 6 or 8 customers, you better be sure you only have 0 or 1 competitor, because otherwise it's like memory, for example, which is a chip market where there's 3 or 4 competitors. That tends to be wildly cyclical, and prices go to shit in the downturn, right?

3. OpenAI's Developer Day Announcements

If the GPU market ever turned into the memory market—which I'm not saying it will because of the complexity—then that business looks very sad. Just take a look at how Samsung and Micron trade versus how NVIDIA trades. Everyone left them alone for 30 years, and they built a monopoly, and now he's picking up the check from it.

So, if we progress this forward to Dev Day, which we touched on slightly there, why don't we start with one of the major announcements, which was the opening up of apps into ChatGPT? You can essentially use your Figma, your Canva, and your Spotify easily and natively within ChatGPT. I'd love to understand, from your perspective, Jason, why don't we start with you? We touched on it a little bit beforehand, but you left me with a cliffhanger, so we'll save it for the show. Jason, were you impressed by this?

Jason Lemkin

I was underwhelmed. First of all, let me step back. We had Marc Benioff on the show a few weeks back, and I told him this is what I wanted: I wanted to talk to my apps. Remember? Marc, at the time, was—I mean, we love Marc, right?—he was like, “You don't want to. That doesn't make sense, and you don't want to do vibe coding.” Now they're doing vibe coding at Dreamforce. The world changes in the 4 weeks since he's been on the show.

I'm like, I want this. I don't want to log into Salesforce. I want to go to ChatGPT or Claude and say, “Tell me how Harry and Rory are doing this month.” I've literally been a Salesforce customer for 20 years and haven't logged in in a decade. I want my Salesforce in ChatGPT.

But I had 2 thoughts watching this. My point is this: Canva, Spotify—I didn't see an aha moment. These are my 2 thoughts. I didn't see magic. I didn't see something that was so great my jaw dropped and I would copy it.

The 2nd thing I thought, as B2B guys, is this is like Slack 2.0. Slack was our ChatGPT until 20–24 months ago. This is why Marc Benioff bought it for $27 billion. It was our OS, right? We didn't know how to communicate. We didn't know how to work async, and everyone's in Slack all day long. Even today, we still use it; it's just not like it was, right? ChatGPT took a lot of that mindshare.

We'd be sharing in Slack, and every app has a Slack integration. Actually, now that I'm—I don't know, Harry, I don't know if you know, I've been vibe coding lately. Have I told you?

I didn't know that, actually. Please tell me more.

Jason Lemkin

And I will tell you, of all the things that are easy, some stuff is hard to do, and some stuff is easy to do.

OpenAI is really easy. Zapier is really easy. But Slack is super easy. It is still the easiest software to push updates and work bidirectionally. But where are all the apps in Slack? The connectors are there.

How often in Slack are you pulling up or creating a Spotify playlist, creating a Canva image, or even pulling up a CRM record, which you can do? How often do you do it? I bet never. Do you track deals in Slack? You probably could. So, none.

So I thought, this is great. It's like Slack, but what do we do in Slack? I was hoping for this aha moment where Sam would show 1 or 2 use cases where my jaw dropped. I'm like, “Holy crap, this is integrating my memory, my data, my learnings. It's combining apps in ways you can't without an API or without a Zap.” I just didn't see the great use case, right?

I'm hoping it comes, because someone's going to figure it out and build a huge business out of it, but I don't know if there'll be 1,000. I'm also not sure we need another app marketplace for the other side of the announcement. I'm not sure we need the 10th app marketplace. Maybe we do. We'll find out. But those were just my thoughts. Nothing—I didn't fall out of my seat for a magic moment. And maybe that's because they built it in 8 weeks.

Rory O’Driscoll

I saw the demos and read the blog posts that they write, which are always so good, by the way. But I think, first of all, when you talk about using apps within ChatGPT, that was one thing, and then AgentKit was more about getting ChatGPT functionality into other apps, as I understand it.

But going first to that, because it's so important to zoom out for folks, the first use case is, I'm in ChatGPT, and the example they use is, I'm doing something and I want to get information from Zillow, right? I can invoke Zillow or invoke some other apps. They had Canva, I think, as another app, and there I'm within ChatGPT and I'm saying, “Create me something,” if it's Canva, or, “Get me 5 houses in this area that have this and this.” You're effectively doing a Zillow search in ChatGPT, right? Cute and clever.

But what you discover very quickly is, boring though it is to say this next sentence, big menus with lots of options are actually a better UI for booking a flight than just being given one. There's a little part of you that wants to say, “Do I want to get the 1:10 out of LA or the 3:20? What time does it get in? And what's the equipment?” A lot of times, all these little “Oh, you can do everything” and fill-in-the-blank claims turn out not to be true.

Now, it's still an amazing freaking business because it has 100% of my mind share when I'm doing any kind of research or thinking. The question is, at what point? My second comment is, I think if I'm doing research on houses, the ability to access some Zillow information is really useful, right, and maybe even to manipulate that information.

But where the cutoff is between, first of all, what you can access from Zillow just on search—in other words, if it's just very data-driven—OpenAI already has a search feature which can get you that information. If it's a little more manipulative, in the sense of, you want to access Zillow and run a search within Zillow, right? Like finding all houses in the Burlingame area under $2 million near a good school district. The answer is 0, but we can pretend. That would be a thing you couldn't do in ChatGPT crawling Zillow externally.

So you do have—I can see a use case there, getting access to Zillow for that. But once you go 2 or 3 steps beyond it, I think very quickly you'll find, “I'm buying a house. The hell, I can just go over to Zillow and do my work,” right? So it's not clear you'll want to do something in this, but it's not clear how much.

And we could be wrong, and this is a better implementation than the last 2. But on the last 2, if you look at the kind of very wide-eyed, excited comments on day 1 about the custom GPTs and the GPT Store, or whatever it was 2 years ago, there were a whole bunch of “the world has changed” comments, and they turned out not to be true. I mean, the world has changed because ChatGPT in and of itself is amazing. It's not clear, and they probably will extract more of your time in ChatGPT by leveraging these apps, but I don't think it's going to be to the death of apps entirely. Does that make sense? You look like you're quizzical there, Harry.

No, I totally get it. I completely understand, and I agree with you. It probably removes 20% of superficial, one-quick-response answers, and then everything else, you actually just go to Spotify to see the playlist because you want to engage with it in the gym and you don't just want to see it in ChatGPT. So I totally get you there.

My question was especially around AgentKit, which, as you mentioned, is the ability for people to build pretty good-quality agents very, very quickly. They did it in 8 minutes in a demo. Does that kill a litany of companies like n8n who promise the same in terms of customer delivery?

Rory O’Driscoll

That's a harder one to assess. I don't know, because it's harder to assess that on the basis of an 8-minute demo. I think I can envisage the kind of agents you're going to be building for enterprises probably are going to require a lot of orchestration and a lot of management. There's a lot of product surface area that a software company just focused on that has to do.

If it's a trivial problem and it's easily integrated with OpenAI, maybe they do kill some of these companies. But my guess is there's more complexity involved, and over the next 2 years, will OpenAI spend the time on that because they've got bigger fish to fry, or will they just make it easy to connect and move on?

I think the thing that worries me—I do want to make sure we move away just from OpenAI—is I saw these crazy rounds, and 2 that stood out to me: likely Naveen Rao, who was VP of AI at Databricks, raising $1 billion at a $5 billion pre-money valuation.

Right. Does this just break venture, though? A billion at $5 billion—for me to get a 10x, it needs to be a $100 billion company.

Rory O’Driscoll

No, with dilution. I was worried you were unclear in your 10x math, Harry. That was a bad moment.

No, no, no. I was accounting for dilution, thinking 50%, given the stage and the company style. It'd be 50%. Does this just break venture math?

Rory O’Driscoll

These deals are unusual. If you think about what kind of deals get this, most early-stage startups—definitely consumer, definitely, I believe, apps in particular—are little voyages of discovery. You don't know if the market's there. You're trying to get product-market fit. It doesn't mean that even if the executive, the CEO, is really good, you're automatically going to get the right to win.

Where you do get it is in these hard infrastructure markets. The number of people who can credibly say, “I can solve this level of technical problem that we're now facing,” is much smaller. Right? So you do have that kind of star effect in raising.

I mean, you're seeing it in Thinking Machines Lab, you saw it in Safe Superintelligence, and you're seeing the same kind of thing there. The number of proven people who have the magic in this space is low.

And we should add that Naveen has successfully built 2 companies. One of them, broadly, interestingly, from my recollection, was a hardware company back in the day that was sold to Intel, and then the second one obviously was sold to Databricks—another clever infrastructure company around LLMs.

So this is someone who has twice built a successful deep-tech company in precisely the hottest space on the planet today, knows everyone, and has a proven record of success. I can totally see why he gets that money, because you can stipulate he'll probably pick the right problem and he'll probably get the answer right.

So really, you've kind of compressed a bunch of the venture questions, and then you're right, you're only left with the third one, which is: does it make economic sense? In other words, if he picks the right problem, if he solves that problem, is the market big enough to justify, as you say, at least a $50 billion outcome from your—what is it?—$5 billion, whatever it was, pre-money?

And the answer is clearly, right now, people believe these infrastructure markets are having those kinds of outcomes. We'll see if there's room for everyone. At the margin, maybe skeptical, but I can totally see each step of the logic train that gets you there: proven person, hard problem, hard problems are getting rewarded.

Would I prefer to back him versus 2 computer science graduates out of a really good school who might figure it out a priori? Hell yes, I'd prefer to back him. Now, once the 2 computer science graduates actually get the work done, you can go, “Oh, that might be a far more attractive bet.” And that's most of what we would do, right?

But I can totally see the proven person saying, “Remember, especially for these bigger funds, you can deploy capital at scale with me. I'll probably solve the problem. I'll probably pick the right problem. You'll be in a good deal.” So I can totally see how it happens.

I can totally see why, especially for folks who’ve backed them before, because they know they have the relationship. I think Lux—give them credit—backed them twice. If you’d backed someone twice in a row to build complex technical products, and the first one took a long time and you showed grit, and then the second one took a short time and you made him a ton of money, when he comes into your office the third time, believe me, he gets a nice coffee and a nice seat, and you’re like, “What do you need?”

Totally. I can totally see how this happened. It’s probably quick decisions in all those partner meetings.

Jason Lemkin

You know, I think some of it—I had 2 thoughts. One is, it is a confidence game. Venture has changed, to Harry's point. The question is, if you’re Naveen and you’re at Databricks, you’ve seen $100 billion and more going up, right? So $5 billion to $100 billion seems plausible.

Listen, I’m not as great a founder as any of these guys are, but back in the day when I met Rory, it was really hard to see north of a $1 billion outcome for a lot of these startups. It was just hard to see it. My whole life—and the reason I sold—was probability. It’s like, “Wow, man, I own 30%, but if I get to IPO at a billion, I just can’t even make more money.” You couldn’t see it, right?

And then quickly, you could see $10 billion, right? Now it’s very easy, if you are a Databricks alum, to see $100 billion or more, because you were just there last week. You just saw how—and that round wasn’t hard to close, was it? I mean, everyone and their uncle and aunt wanted to get in at $100 billion.

So it’s all become—venture’s always been a game, but for founders, man, it’s a super game today, right? It is nothing but. I mean, walk out of YC Demo Day and, as an investor, you feel gamed.

My other thought was Balderton's Revolut seed: 2 million at 8 post, followed by its Series A. Those terms are not in the same genus as this deal.

4. Why VC is the Most Forgiving Asset Class on Price and Valuation

So, I just interviewed Mike Cannon-Brookes from Atlassian, and he said, “Listen, the trouble is there’ll be a load of shit that will lose money, but there will still be some Amazons in this AI wave.”

Yes, but Amazon was priced as Amazon was, and that generated Amazon-level returns.

These are not priced in any universe at Amazon-level returns. Even if you have Amazon-level-plus-plus outcomes, they’re still not venture-style returns.

Well, the IRR could be tolerable if you put enough money to work, right? Put half a million—half a billion—into the round.

If your time to value is much quicker, then sure.

Yeah. Yes, I mean, let’s state the banal: a lot more has to go right. Watch this sentence, so you can laugh at $5 billion pre rather than at $8 billion post. Exactly. Did I really say that and think I’m going to get value from that?

But, yeah, no, totally. Look, give a bunch of examples. I mean, you look at what the Stripe seed round that Aydin and others did was—dirt cheap. The Airbnb round that Sequoia did was compellingly cheap. I look back, and we have very successful 2009 and 2013–14 funds, and you can see it when I’m talking to the younger partners. The thing in their eyes is, “Dude, you were able to buy so cheap, even a shit like you could make money, right? You should try making money today, big guy.” Right?

Entry price has an impact. It’s not the only thing. The wonderful thing about venture is there’s positive convexity. It is the most forgiving equity business when it comes to getting the price wrong.

Let me tell you what I mean by that. In PE, if you get the price wrong, they’re low-variance assets. If you overpay by 50%, you’re toast, because they’re 3x assets and they’re not going to ever be 7x, right? Your degrees of freedom are limited. It’s the same thing in the public markets, right?

The great thing about venture is that it has maximum variance, which means that it is the most forgiving of getting the price wrong, because you have exponential growth on your side. Now, I think we all rely on that, and sometimes we rely on it too much. We find that just because you’re maximally forgiving on overpaying doesn’t mean you’re entirely forgiving on overpaying.

To your point, Harry, you can push a theory to destruction. What you’re saying is, if you pay for everything where you’ve got a 5x return, if you do something wildly amazing and only 1 in 3 of the companies—or 1 in 5 of the companies—does something wildly amazing, because that’s just the way wildly amazing runs in this world, you don’t have much of a return. That’s a fair comment.

I think it’s another clear example, though, of Alex Wang changing venture mindsets on entry-price acceptance, because we all go, “Well, if Alex is worth $14.8 billion, Jesus, Ilya’s worth $30 billion,” and it justifies these prices based on the acquisition of talent. That’s why, whenever anyone uses comps to discuss what we should pay for a deal, I want to bludgeon them to death.

Because the problem with comps is they tell you what company A is worth relative to what companies B and C are worth today, right? In the public markets and the private markets, you’re exactly right: if Scale AI is worth $14 billion, then you’re like, “You’re worth $15 billion, so you can pay $15 billion.” It’s a logical way for a banker to relatively rank things today.

But it turns out, as investors, we’re trying to answer a slightly different question: What are they going to be worth in 7 years? The problem with comps is that if you used them, you would have bought, in 2021, a whole bunch of assets that were only priced at 50 times revenue because the other shit was priced at 80 and you were getting a good deal, right? That turns out to be a very bad way to invest.

You’re right: you can’t rely on nearest-neighbor comps-type analysis to do investing. We all do it to some extent because it’s easy to be a comparison shopper. But you’re exactly right, Harry: you have to have a view that says not only is Scale AI worth $14 billion today, to take your example, but that you believe, on a sustained basis, that companies doing $500 million, $600 million, or $700 million in this kind of business can trade at 20 times revenue over an extended period of time that includes the next decade. That’s a lot harder story to believe.

A slight-variant question here is: let’s assume your LPs will support you—you can do these deals, right? With this deal, you’ve got to deploy the capital. Maybe you’re trying—let’s assume you have access to a lot of capital, like Andreessen or others, right? How fast are people trying to deploy these funds? 18 months or something like that, right?

How many candidates are there like these? How many of these generational founders are there, even if you have to hold your nose when you make the investment, even if you have to hope and pray? Even if you see Rory’s math, it’s like, God, 80% of these have to work out for the math to work. You do have to deploy. There always is pressure for 90% of VCs. There’s pressure to deploy it, right?

Let’s actually take that drill-down because I think that’s really important. Let’s look at Andreessen’s new fund: $6 billion, now expanded to $7.5 billion, reportedly. Say you’re putting in $300 million or $400 million of this $1 billion raise. How many of these? Gosh, you’ve got to find 15 of these in, let’s say, a 2-year period. But that’s okay, depending—yeah, $300 million times 20 is $6 billion. Keep rolling.

You’ve got to find them, right?

You’ve got fees, my friend.

And reserves. Okay, I agree: 15 to 20.

How many are there? Not to keep going here, but how many of these candidates are there out there?

Is there 8 a year? Because that’s kind of the 7.5 a year, too. If I’m being really precise, Harry, we can cut Naveen in half and you have 7.5.

Yeah.

Yes, but you’ve also got to say, look, we’re all Pavlovian. We do the things that feel good. Once they start feeling good, you keep doing them until something hurts, right?

Based on the news of the leaked Andreessen numbers, you’ve got to say they’re excellent. They’ve earned the right to throw $7 billion on the table. The truth is, we’re human. We’ll just keep doing this until Mr. Market delivers a sad lesson that says you’ve overreached, and so far that hasn’t happened.

But I don’t think it will. I don’t think it will happen. They’re about to get the payday of paydays in venture land, with a minimum of $40 billion back from Databricks.

Yes, I mean, totally. Look, at some point, what it takes to change is externally driven.

We’re jumping around here, but the people who determine whether it’s allowed to continue or not are the LPs—the people who put us in business. I thought it was a really interesting one, Chris, that you highlighted: Brown and Northwestern selling VC stakes.

We saw Yale and Harvard sell VC stakes earlier this year. Will this be the new normal? Is this a new wave of LP liquidity that we’ll continue to see more and more of?

Jason Lemkin

We can talk about illiquid assets and whether they’re getting an adequate premium over liquid assets. That’s one discussion, and that’s a good discussion. On top of that, you have the extra phenomenon of university endowments under particular pressure because of the political pressure and the push on universities for a whole bunch of changes, which means they feel the need to be more liquid.

So, I think for those guys, it’s not going to be the new normal in the sense that you’re not going to be doing this because it’s not going to persist for a long time. By definition, if you’re selling a lot of venture assets, you’re probably not going to buy a whole load more, right? So, it will tail off. I think it’s a readjustment. What you’re probably seeing is some reassessment of the Yale endowment model and exactly how much liquidity you want to have in your portfolio.

Wouldn’t it be—? It’s funny: I just went through, hopefully for the first and last time, having an LP sell a position. I only have so many LPs. I don’t have as much money as either of you do. It was interesting to watch this firm, Evercore, manage it across a lot of fund managers.

Even though the process was handled very poorly and was very friction-filled, what I thought at the end of it was that there should be a lot more of this. What I didn’t know—I never read the LPA. My LPs have no rights to sell at all. Literally nothing. There are no exceptions if you’re struggling or anything like that. It’s just shocking, because a lot of this stuff is LP-friendly. This one is absolute: you have no rights to sell.

I thought that, in today’s world, where it could be 20 years until even good funds wind down, and where founders want to go longer and longer and longer, some LPs don’t care. But if the Harvards of the world care, and everyone says to me that it would be nice if there were more LP liquidity in a way that wasn’t bad for GPs, I think it would be a positive outcome.

What I saw—and this is only an N of 1 for me—is that it’s a pretty broken, friction-filled, weird corner of the market, right? They need my permission, so they’re trying to manipulate me. These guys claimed they had the rights to do it, but they were lying because they didn’t have the balls to ask for my consent. It was very interesting from a games perspective.

Believe it or not, I try to be a nice guy. I said, “If you have the rights, do whatever you want.” Then they said, “We didn’t mean to say we have the rights. What we mean to say is we have the moral authority to sell your position.” I’m like, “What?”

Jason Lemkin

You might be oversharing here, dude.

I love it. I’ll keep sharing. I’ll share everything except the name of the LP. I’m happy to share.

Rory O’Driscoll

But I do think you’re right.

Because it’s a gem of a story.

Rory O’Driscoll

Stepping back, better liquidity is an advantage to both sides, especially as these funds drag on. We’ve been supportive of some of our very good LPs who’ve continued to re-up in new funds. After 10 or 12 years, if you’ve got a tail-end residual with one company left, does it make sense for them to clean up their books and sell to a secondary?

There’s going to be a lot of this kind of process happening because it just makes sense. It’s just like companies staying private longer. The consequence of that is that you end up having to facilitate secondary shares for employees because, when your time period of being private is longer than half your working life, you probably want to get some equity money in.

In the same way, for any of these companies, even if they have a long-term perspective—which obviously LPs should if you’re entering venture, given the thing—sometimes you talk about the long term as 8 years, and it turns out to be 12 years. I think it’s healthy to be able to conduct those sales.

Even if they were efficient—and it sounds like yours weren’t—the truth is that, no matter how efficient they are, there’s still a drag to it. There’s still a price discount. Therefore, it’s not going to be the new norm in the sense that people aren’t going to run into it saying, “I love to do this.” But either because you need capital or you want to close out an old fund, it’s just going to be a part of the world.

There’s already a substantial secondary business with big players. There’s always going to be secondary, and it’s probably going to increase over time.

Quick summary. We’re acting like there’s infinite liquidity. It appears there is for OpenAI and these secondaries, but this is a world of scarce liquidity outside of a few names, right?

My learning from this process is that everything would be better if there were more liquidity down the stack. Everything—no matter what the discount, who cares? The market can decide the discount. Everything would be better in an age where companies are going longer to exits. It would be better to have more liquidity for all players.

Jason Lemkin

That’s also true. I think, implicitly, you’re saying that for companies, which of course is why they should go public, because you’re exactly right. I always say to people, when people say there’ll always be liquidity because there’s lots of money, I’m just going to say it so you remember it in a year or 2 when it happens: liquidity doesn’t evaporate because people run out of money. Liquidity evaporates because people get scared and want to keep their money.

5. What Does it Take to IPO in 2025: Why Snyk Will Not IPO

At some point, when that happens, you’ll go, “Oh, that’s what the public markets were for.” So, you have a little more opportunity than this.

People should just IPO, right, Rory?

Rory O’Driscoll

Yeah, they probably should.

Then you’ve got a company like Snyk, which has slowed down growth to 26%. I think it’s about $300 million in ARR. Its growth is down from 150% in 2022, so it’s taken a big growth hit.

Jason Lemkin

First of all, I would actually go with the more conventional English, “has slowed down,” rather than “has slowing down.” You’re sticking with the Queen.

It’s slowing down, I think.

Jason Lemkin

Yeah. But that was just me being me, Harry. Sorry. I have to do it once per show.

Dude, you’ve corrected me on every sentence.

Jason Lemkin

Harry may have this weird accent, but he’s one of the most well-spoken people I know.

Yes, every—

Jason Lemkin

I’m full of malapropisms in every paragraph, and Harry just nails it in every question. I don’t know how he does it.

I was just being snarky. Let’s—

Jason Lemkin

Just beating me up.

I will admit you can calibrate almost everything by coffee consumption before a show, and this is a 3-coffee day, so it’s just going to be tough. I’m sorry about that. Normally, on a 1-coffee day, it’s easy, but I’ve got a lot going on, so sorry about that.

What are you saying about Snyk? Let’s go back to the task at hand. At 26%, has growth slowed down to the extent that they are now no longer able to IPO? There are rumors now—and this was the news—of growth down to 26%, revenues at $300 million, PE buyers circling, and PE as the option. Are they at a stage now where they’re looking for a PE buyer and the IPO is off the table?

Rory O’Driscoll

The interesting thing is that $300 million in ARR and 26% growth is about the low bar. We were just looking at this now. There have been 15 IPOs year to date, and the median IPO this year was a stunning—get ready for this—$931 million revenue run rate. $931 million.

The median blew me away, right? The cutoff there was a couple at $200 million or $300 million, growing at around 30%. The first comment, just to put it out there, is that it’s not like they’re miles away from it. If those are in fact the numbers, it’s just below. If it’s just below that line, then you’ve got 3 routes, and we discussed one of them last time: either you have PE, you have, if you’re lucky, a strategic buyer—but if you’re not in the AI world, I don’t think people are buying strategic stuff at this point—or you have to consolidate like dbt and Fivetran to get to scale.

What price does that go for? They raised at $7.2 billion last time.

That’s the tough thing, because PE buyers are 6x to 8x, plus or minus. You multiply $300 million by 8, and you just end up with a different number.

I’d say it’s in the 2s, based on Netskope as a rough comp.

Rory O’Driscoll

Yeah.

Right. Netskope’s at $8 billion. It was at $700 million, growing 33% at IPO. So, this is $300 million growing 25%. I’m just using the roughest VC math, but if that’s worth $8 billion, this is worth in the mid-$2 billions.

Rory O’Driscoll

No, 2x revenues.

No, $300 million in ARR would be worth $2-something billion if it IPOs, if there’s appetite for an iconic company, right? Just not a Wiz, right?

No, you’re exactly right, Jason. I’m using a security comp. I know it’s not the same application. It’s the last security IPO, Netskope, which is S-tier, but it’s not quite Rubrik.

The problem is that we start to see this triage, right? Snyk, or whatever—it’s great, but it’s not Netskope, which isn’t Rubrik. I don’t have your skills, Rory, but I can pull up the valuations and the numbers and spitball it. At $2-something billion, that would be its IPO valuation, right?

So, to play it out, there are 2...

So the question to play it out is, there are 2. One is, does that fall just below or just above the line? Can you get a public deal done at that size? And if you can't, are you in private land? Then, obviously, instead of all the preferred converting to common and cleaning up the cap table, you have the whole preferred stack. And then the next question is, how much have you raised? What does that mean for all the folks? So you've got all that drama to do.

And even if you can't—let's say you can't IPO because you're at the edge, right? It's at the low edge. Do you want to deal with that crap and be ignored by Wall Street if you're a rung and a half below Netskope? Is it worth it to be one of these invisible public companies?

I'm just going to put it out there because I really don't like the idea that these companies are no good because they're not worth $7 billion. I mean, stepping back, the other alternative you have is to compound for a couple more years at 30%, right? And if you've got the stomach for the holding period, try and do some acquisitions, build the thing up, because at $400 million at 30%, it's a little more compelling. You're well above the threshold line.

So that is one option. And then you've got to assess, obviously, do you have obsolescence in your future, right? The point I'm making is this: every single one of our quote-unquote successful companies, other than the most successful, are dealing with this reality. I can think of plus or minus 8 or 9 companies in our much smaller portfolio where we're all at this kind of stage and having these dynamics, right?

I think I said it: there's just a whole bunch of cutting and wood to chop to figure out what these companies become over the next 2 to 3 years and how you get liquidity. Is it private to private? Yeah. So there's just a ton here. I'm kind of—I suppose what I'm saying is, we're quote-unquote picking on Snyk, but it's a high-class company. $300 million is the high end of great; 30% is the high end of post-pre-IPO critical mass. I mean, we see lots of companies at $150 million going 10 or 15%. It's not just a crime of price, which is like overly exuberant capital-market stuffing cash down a company.

I'm curious to see what you see in your portfolio, but I only have 3 companies that I would say are in a different era but are in the same bucket that you're talking about. What I'm worried about is that 0 have had PE offers. 0—0 of these 3.

Now, if this was 2021, your phone would be ringing off the hook, even into late 2022, even into early 2023, pre-IPO. Your phone would be ringing. The valuations might have gone down into early 2023, but none of these deals—I can think of 1 at Snyk scale, 1 about a third, and 1 about a tenth. But they all should have gotten PE offers, right? They all have the right Rule of 40 numbers, the right NRR, right? Why not buy them? They're not going to IPO, but they're good companies.

Crickets. Crickets from the PEs. Crickets. Is your phone ringing off the hook from these PE firms? Are they beating down the door? Are they out there? Can I see them over there? Are they banging on the door to get in this morning? You've got to get off the podcast to sell a few portfolio companies at 8x.

Rory O’Driscoll

I think you're spot on, Jason. Which is why, if I was sitting on that board—as I always say to people, when you're private, the liquidity window opens only rarely. Whenever it opens, you should pay attention. You can decide no, but you should pay attention.

And you're exactly right. You're not seeing infinite demand. It's not like PE is dying to do this. They've got lots of capital, but they're not in any rush to buy subscale assets that aren't defensible market niches.

And I think, going back to this, maybe what—as I articulate this—the number one thing I'd say to these companies, including ours, is you've got to take control of your destiny. What does that mean? It means a couple of things, probably 3 things, maybe 4.

One thing, and probably the most important thing, is you've got to make sure the management and founding team are excited and have something to play for. Maybe I put that first, right? If the management, if the team doesn't think that they can build value here, then you know what to do, right? And you should incentivize them to build value.

We've done a fair number of these for these kinds of companies. I call them EFGs: equity for growth, right? You say to someone, “You're fully vested. You've long since been fully vested, but you're the founder. You want 7 or 8%. Conventional wisdom says that's it. Let's just all keep working.” But they're like, “I'm not getting any more for more time.”

I've done significant equity grants that are linked to delivering growth, so they have something to fight for again, because these guys are fighters. That's what you want them to do. I want the founder to say, “I thought I was on an 8-year journey. I'm on a 15-year journey, but fuck it. I've got another 7 years of equity ahead of me.” So I take 3 or 4% dilution, but the guy who's in the trenches is incentivized, and so is his team.

That's the first thing: make sure the team is excited to keep going. The second thing is, make sure you have control of your destiny by being profitable. And then probably the third thing is, figure out a second act.

Typically, it should be a second product, or typically something that links to the AI trend. We're saying this to our companies: you should link to the zeitgeist, right, if at all possible. If you're in somewhere like financial services or fintech, it's not really that relevant. You'll use it for the back office, but it's not going to change your product.

But a lot of these software companies—you have to assume you need a second act. You have to assume it gives you 7 more years of growth. It's almost certainly going to be related to what's going on in AI and how workflows become software, become agents.

I think if you do those things, then as a board member you feel a lot more secure because I have a plan. I have a plan that doesn't rely on the kindness of likely Thoma Bravo—the kindness of anyone, right? I'm building my independent company here, and everyone's aligned. So that, to me, is the job of the board right now, and we've done that at a couple of companies.

What's the first one? You called it an EFG? What's the—

EFG

equity for growth.

No, you inspired me. I proactively did one of these for that reason. You learn a lot from the process, but you've got to do it. It's not enough, but you've got to do it, right? You've got to do it, especially because a lot of folks won't ask.

Some founders are very aggressive in asking, “Where's my next 10%, Harry?” But a lot of them won't ask, right? A lot of folks just won't ask. I think I learned it's good to be proactive.

You've got to. And in return, it's the only time you get the right to say, “In return for this, you've got to dream big again.” But if we're going to be at 20, going to 15, going to 10, then with dilution, you're way below my cost of capital. Let's admit that and go for a sale. Then you get what you get and you don't make a fuss.

But if you think you can keep it at 20 and walk it up to 25 or 30 and keep going for 3 more years, that's value-creating at a very different level.

The astonishing thing for me about your Aaron Levie, your Drew Houston, or your Mike Cannon-Brooks is the longevity.

Totally. There are just so few. We mentioned Snyk there, but how few actually really do the 15-year journey?

Because it's hard, you know. Can I ask you? You've seen more than me, with more experience and wisdom. When the founder leaves and there's a CEO brought in, what percentage of the time does growth reaccelerate and the positive outcomes come?

Rory O’Driscoll

Break it up into 2 categories. If you have product-market fit and you have a CEO who's very entrepreneurial but just not a great manager, then it can work because they've done the entrepreneurial act, but they're not great at managing. Maybe they're so bad at management that a competent manager stepping into a post-product-market-fit company can give it a lift. It's not ideal. I'd much prefer to err on the side of making that founder work by surrounding them with good people, but it's not crazy.

If, on the other hand, you don't have product-market fit, the founder's not working out, and you think you're going to hire someone to get product-market fit, you're deluding yourself. Don't spend a second, because it's just too hard and too unlikely. You are what's called wrong, and you should sell for what you get and move on, because that's not an act that a professional manager does. If they were capable of doing that, they'd be founders. So the answer is 0 in that case.

I think it's bad in all cases, but for the first one, I just think—listen, you know, back when Andreessen wrote these handcrafted blog posts himself, the point was, “We always bet on the founder.” Sometimes the founder doesn't stay, right? This is more about the second case. Sometimes they don't stay, and we have good outcomes there, but the best outcomes are when they stay.

So we do everything we can to surround the founders with other executives to help them. But if they can't finish the journey, we'll support them either way. I think that's the best answer I can come up with.

That’s my answer, right? I never want the CEO ever to leave. Good God. But if they raise their hand, and you’ve given them the EFG thing—or whatever you call it, the re-up—and you’ve had the conversations and you’ve begged them to stay, and they still just don’t want to do it, there’s no point, right? Most humans should take the millions and relax, right?

They may be dissatisfied, but most humans are not wired this way. They’re not wired this way. Which is why the most useless question VCs ask founders is, “Rory, if you were offered $500 million for the business today, would you take it?” And then they go, “See, I told you he’s not here for the long term,” or, “She’s not here for the long term.”

So I’ve got to say, to be clear, I never asked that question. I think it’s a stupid question on its face, on multiple dimensions. Let me give you just 2 ways in which it’s a stupid question.

The first thing is—and I know I said this before—people’s opinion on what they’ll do if they’re offered $500 million at a point in time when they haven’t been offered $500 million is meaningless because it’s a purely theoretical discussion. Right? I’ve seen people go both ways. I’ve seen, “Hell, I’ll never sell,” and then, when they’re offered the money, “Fuck, I’m out of here.” And I’ve seen people where I thought they were merciless say, “No, we can keep going.”

So I literally don’t ask that question because there’s zero information content to it. And the idea that you wouldn’t back a really good founder just because you’re afraid they might sell in a good outcome that doesn’t become great, that’s just so low down the list of things that you’re worried about when you’re doing a deal.

You want to know: Is the market good? Does the founder want to build a big company? If they start on the journey to doing that and someone gets interested, it’s a high-class problem. There are a couple I look back on and go, “I wish we’d run that longer.” But in the list of things that will reduce the return of my venture career, that versus backing B-level people who don’t even make it happen? It’s a concern, but it’s also—how do you make it actionable? I’m sorry, that was a little long and raggedy, but I just don’t think it’s a useful question.

I completely agree. I also think that you can tell your best and your worst within the first 60 days. The messy middle, you don’t know. There’s a TBD, but the ones where you’re like, “Shit, we regret that one,” you can tell pretty quickly.

And same with the, “God, they’re on it.” That first update is so good.

Yeah. The first board meeting. This is—I do agree that at the 70% level.

6. Vercel Raises $300M at $9BN: Suicide Round or Strategic

In other words, it’s something. But I totally agree: on all my best deals, there’s been a moment somewhere in the first year where I remember sitting around the board table going, “Oh, Rory, you clever boy. You’re going to make money here.” And all I have to do now is cheer them on and not have to do a lot, because this is just humming. You’re exactly right. And to that end, that’s always a good feeling.

Speaking of those moments where you go, “Wow, well done. I’m in a great company,” there are a couple of standout raises that I wanted to talk about. One is Vercel raising $300 million at $9.3 billion. Jason again highlighted this; it was announced pretty quickly after the post from Guillermo Rauch, the founder, which we’re not getting into because we don’t do politics, but the timing is interesting.

I want to hear your thoughts on, 1, the timing, and, 2, I’ve heard these described before as suicide rounds, just because it’s a super-high price with not such a huge amount going in—$300 million—and so it sets a huge expectation with not a massive capital injection. How do we feel about those 2: the timing and the suicide-round status?

The suicide one is an interesting one. The Vercel thing, I don’t claim to be a total expert, but I am 200 hours into my vibe-coding journey. I think Vercel and Supabase are actually tied in a sense, which is that these are not the most profound bets; the world of software development has completely changed, and everyone who’s going to build a web app going forward is going to build on Supabase.

Jason Lemkin

It is the default choice for how to host and manage Postgres in this world. And when folks want to host apps live, they’re going to use Vercel. It’s super easy to use. People do love Vercel.

You’ve got to host this app and you need a database. These are 2 structural components, and the number of apps is also exploding, which it is. So if this is the future of hosting, management, and databases, and it’s exploding, these are good bets to make. These are the leaders. These are 2 leaders.

You could debate the valuation, whether it’s a suicide round. But, valuation aside, I think these are actually Captain Obvious bets. I think they’re both Captain Obvious bets because this is where developers are going. That’s how you make money.

I mean, Mike Cannon-Brookes would probably reiterate that from a few years back: follow where the developers are going. If you become the leader, bet there, and you’re going to make money because those markets are pretty damn large if they’re growing. Beyond the suicide round, I think these are good bets.

Vercel was at $9 billion. It’s still a lot of money. Hopefully, you have more than 1 of those bets at $9 billion. To Rory’s point—

Rory O’Driscoll

First of all, I agree, Jason, with what you’re saying. I was talking to one of my partners yesterday, and we were just talking about it. The more you do this, the more you just say to yourself: You just need to do big, exciting deals in trends that are absolutely obvious.

Every time you try and make it harder than that, you lose money, right? Both of these things are exactly right. They are on-trend, Captain Obvious: there’s a new wave of people building apps, and these are 2 parts of the infrastructure that people will use to build those apps.

Both of these companies have done a brilliant job. Vercel and Supabase were, just as a reminder, infrastructure components, for lack of a better word, that existed pre-OpenAI, pre-the AI trend, pre-vibe coding, but have just inserted themselves into relevance as components to build these next-generation apps and are riding the wave, for lack of a better description.

Because of that, they have momentum. They have growth on their side, and yes, the price might have doubled in 6 or so months. It may also be that the company doubled in 6 months.

So the revenue multiple is the same, which is actually a super-interesting discussion. What you’re effectively saying is, normally, you figure you start with high revenue multiples because, by definition, your revenue multiple is infinity on the seed round, and your multiple is 6x when you get 30% growth. The idea is that your revenue multiple is coming down as the rounds go up.

But what you’re seeing in some of these AI companies is people are effectively saying, “When they were doing $100 million, I paid 20 times or 50 times. When they’re doing $1 billion, I’m going to pay 20 or 50 times because the growth is still the same,” right? And it’s logically correct.

I mean, there’s a market-size question at the end of it, but from a growth-rate perspective, I’m willing to bet that you look at those 2 rounds and they’re not wildly different. Is that logical?

Totally agree. That’s why I said, actually, you said it correctly, Harry, and you’re agreeing with me exactly, which is the odd thing. It applies even though the growth rate’s remaining the same, and normally you expect growth rates to decline.

So you kind of have this mental model—we talked about growth persistence—but they’re not declining here. When you do these rounds at the same multiple, but with an order of magnitude more in the valuation, the thing that can go wrong is you hit a market-size wall or some kind of wall, and they decelerate rapidly, and then you’re wildly wrong at scale, right? That’s the risk.

If the Vercel market or Supabase market is finite and can support a $10 billion or $20 billion valuation, even if you have a 100% share of vibe coding, then you could see very abrupt valuation changes. I’m not discounting that. I’m simply saying I know how to get to doing the same—these quick rounds—because the underlying growth is so quick.

We’ve been looking at some markets where we’ve seen 2 and maybe even 3 rounds within the year. Part of me goes, “That’s crazy.” And part of me then goes, “The B was at the same multiple as the A. Maybe, Rory, you need to update your priors to the world as we see it now.”

And accept that, if you’re getting this growth rate—and these growth rates are, and really, Jason, you called it in the discussion a while back, well beyond triple-triple-double-double-double—maybe you can lean in, and maybe it does make sense to have these rounds.

So, not what we do, but I definitely wouldn’t call them suicide rounds, Harry, because, A, the growth might justify it, and, B, the other part of what you said is, are they raising enough? Yes, they are.

If you think about it, take Supabase, where they did 2 rounds. Some smart dudes at Accel looked at the $2 billion and said, “You need whatever, $300 million,” and then 6 months later they picked up another $100 million.

They still have the $300 million. So it's only a suicide round if 2 things happen: 1, you decelerate quickly, and 2, you start losing money such that you're forced back into the market in the next period of time, where you would have to take a down round. So I don't think it's wildly, outrageously, stupidly risky.

7. Does King Making Really Work in Venture Capital: Legora vs Harvey

I mean, if you want to talk about risk, there's a lot riskier things going on in infrastructure land than Vercel raising $300 million at $9 billion. I don't think that will be the problem that brings the whole thing to its knees.

One interesting lesson for me is kingmaking: when you have very quick rounds and a lot of money going into categories, it absolutely exists in this space.

Yes, it does.

But it doesn't enlarge enough categories, is what I'm seeing. And what I mean by that is, in law, for example, there are many that have got a lot of funding and a lot of traction. Same in healthcare, same in customer service, same in coding.

But then, as you go to smaller markets, kingmaking really becomes more prevalent. The smaller the TAM, the more prominent the kingmaker ability is.

I'm not sure I buy that at all, right? I think it's—sure, I think it's obviously easier to be a kingmaker in a small market because it takes less money, and more people have small money than big money, right? It's the Julius Caesar quote: "I'd rather be first in a village than second in Rome," which indicated he was a psychopath, by the way, but we can come back to that.

But I disagree, because I think kingmaking is going on in the biggest markets. I think, to some extent, everything OpenAI has done has been both technically brilliant and financially kingmaking, right?

Just pause on that. What do you mean by that? They have not been kingmade, so to speak. They have many competitors, and it's by no means a monopoly.

I think it's a duopoly.

Okay, so controversial.

Yes. I think their capital strategy has been to the point where it's going to be very hard for anyone else to attract that kind of capital. I mean, if you take someone like xAI—if they really do need $100 billion and they've locked it up, I can't remember how much xAI has already raised, but there's a lot more to be raised to do that. So I think that is a kingmaking strategy. I think you mentioned—

You don't think Anthropic could, and Grok couldn't?

I think it falls away very quickly. I don't think it's a monopoly, but I think it's definitely an oligopoly, to be clear. So, yes, number 2 clearly can. It's a differentiated number 2 with an overlapping but different strategy: one is winning in consumer, one is winning in business.

But do I think all of the other recent, very high-profile startups are going to be able to attract the kind of capital it takes? No, I don't, because there's not that much capital out there. So that's an example, I would say, of the largest-market kingmaking.

There is so much more capital in venture, but it's so stratified, right, into 20—however we define it—20 companies, 50, 100; we could go down a lot. But it's one thing back in the day when you would kingmake with a $20 million round or a $50 million round, right? Or kingmake very late stage with $100 million.

But now, even though there's more venture than imaginable, you really can exhaust the capital in a category at 9 figures or more. How many of these categories can support $200 million, $300 million, $400 million rounds? If it is capital-intensive, it's tough to compete.

I mean, it's crazy. If Base44 has 10% of the vibe-coding market as part of Wix, if that's accurate, it's crazy. But where would they have afforded those tokens? If Replit and Lovable are losing some money, where would this poor guy in Israel with 8 developers get $50 million of tokens?

I don't know. I mean, there is kingmaking happening here, right?

I've been thinking about it, too, and I'm actually going to change some of what I said, Chris. Actually, listening to what you said, I think I don't like the word kingmaking. I've decided, because I think it imputes way too much value to venture, right? Like, as if we're making the difference.

I think there are often mild versions of that. You see something where, "Oh my God, Sequoia led this round." Maybe people back off the other competitors. But in general, the entrepreneur is the king and the entrepreneur makes the good company.

Hang on, right? With so much capital available, they can get into this virtuous circle of getting the prestige names, a prestige amount, and significant amounts of capital that help build barriers to entry and deter invaders. But the fundamental act of creation that allows that to happen is the entrepreneur and the revenue success.

But that's just not true. I'm sorry, but I mean it in the nicest way. The cash enables that execution, and execution wouldn't happen without cash. And 2, what you've seen is kingmaking pre-revenue, which is companies that are at $3 million, $4 million, $5 million in revenue, which is great, but bluntly not a huge scale, getting $50 million to $200 million successive rounds on the back of a Tier 1, with an Iconiq—or you name your multistage fund—coming in very quickly afterwards.

And at that point, it actually is the venture investor that is doing—

First of all, I agree with what you're describing as a phenomenon. We've seen it in a bunch of markets. We've looked at those markets. We've struggled to find a way to compete in those markets and figure out what to do, to be really direct, right?

But this is my point: everyone knows them and goes, "Oh, shit. We don't want to go in after Sequoia and Iconiq into a company that competes with the Harveys of the world."

That's exactly it. But I think, going back to kingmaking, in every one of those cases—I just, we may be arguing semantics, but it's an important comment—you start with the company doing an excellent job. They build a good product. You're right: they get to $2 million or $3 million in revenue.

In other words, the moment of gestation is the company doing a really great job and having a small early lead, right? And then, you're right, the wall of money allows you to build on that lead and defend it. The brand-name firm allows you to raise a follow-on round very quickly. Arguably, the growth allows you to do it.

So there is this self-reinforcing thing going on. I think we're describing the same phenomenon. I just wouldn't call it—

I'm just saying it's earlier and earlier, because I don't even think, in some cases—in some cases, literally, Harry—it is pre-execution. Its founder has unique insight on GTM, on product insight, on you name it, and that is enough to catalyze the fire.

Yes. And in most cases, I think you'd agree, you describe more correctly: it's a couple million dollars in run-rate revenue with prestigious customers, a strong founder, growing quickly, who has done one seed, raises a good Series A from a top-tier firm, gets a Series B 6 weeks later. Now you've got perceived momentum.

And, yes, we struggle with those and, frankly, both struggle to know: should we compete? And then, secondly, we struggle to compete because once you have the top-tier firm, you get a whole bunch of people willing to pay up. I'm meeting a lot of founders who are going, "My God, I did not realize quite how powerful kingmaking is."

Because I'm like, everyone is just saying, "Whoa, we don't want to compete against that."

Well, I always remind people it's a long way from here to $300 million in ARR in a public offering. It's just a comment I always make. And I don't know if it's the right attitude to say, "I just can't compete."

Harvey looked like they'd been kingmade, if that indeed is the past tense of it, right? And then likely Legora came in from Sweden, for God's sake, and killed it. They did really well. They shipped that product. They took out what looked like a monopoly and took it into a duopoly.

Those 2 companies are playing aggressively in the legal space. So there's an example: they were second to the market. Harvey had established a lot of mindshare plus capital. They had Sequoia, they had conviction, they had really strong people.

And give Legora credit: they shipped a good product, a very good product. They got Benchmark in. They've just done a third follow-on round with—I want to say—Bessemer, after taking Redpoint. So there was room for a second person to be kingmaker.

And by definition, if you have 2 people being kings, it can't be a king, which goes to my original statement that the size of the market does impact the ability to kingmake. Law and health are difficult to kingmake; smaller markets are absolutely more plausible to kingmake, validating my original statement. Mr. Very Cool, thank you.

I'm not sure I agree, but you said that so confidently. I'm just going to give it to you. Would you do a third in that market, in the law software for corporate law, for law firms?

I think that's hard. I think there are other markets in law equally interesting that we'd like to play in, but you're right. At some point, it gets kind of cooked.

You know, I remember when Mike Cannon-Brookes came to SaaStr, which was a long time ago—SaaStr Annual—and we had the CEO of Trello interview him right after it got acquired. It was interesting. It was like a board meeting at SaaStr Annual, and he asked him, "Could you do Atlassian today?"

He said, "No way I could do Atlassian in the same way today, because I had 5 years to be left alone."

Correct.

That's why bootstrapping worked: everything was so slow the first 5 years.

So the flip side today, I think, is that a lot of the AI startups we're seeing start off very, very low-capital-demanding, right? A couple—2 or 3 folks—some free tokens, some free Google Cloud. It costs nothing, but then they consume a lot of capital. If the question is whether you could be a new entrant to the market, a number 3 or number 4, and choose to be capital-light, that's great.

But if the space requires capital to win, or you, the founders, believe it does, then kingmaking becomes a prophecy because you just bow out. If I believe I need $100 million to scale Base44 to compete with Replit and Lovable, then I'm going to bow out and sell to Wix because there's no option, right? That's what's a little bit different from classic 80% gross-margin software: if you were lucky, there was a third path. If you were lucky, someone out of left field—you hadn't heard about them—got to $100 million.

Well, it took 5 years longer in the early days, but they caught up. They used to catch up around $10 million or $20 million ARR, and actually, in some cases in SaaS, you would lose your capital advantage around $20 million ARR because you would catch up, right? Like in Atlassian or Qualtrics.

I'm not sure that's true in the age of AI. I'm not sure that you lose the capital advantage. I think it often is inverted, where that capital is more helpful, right? In the age of AI, that's what makes it harder to invest in the number 3 or number 4. You might be like, listen, yeah, but those guys only need $10 million, and we'll see how it goes. If they're going to need $200 million, it's not fun to invest in the number 3 or number 4 player, is it?

Jason Lemkin

I agree with that. I think what you're saying is exactly right, which is that you could do Atlassian and bootstrap it for 5 years because no one was going at it and you had an uninterrupted run. Today, I think the direction is obvious and the capital is available, and therefore, even if you don't want to do it, someone else will do it, so therefore you have to do it.

It's like nuking the cities. I didn't want to do it, but I knew they were going to do it, so I had to do it. Pretty soon, everyone's launched $50 million Series A checks at each other, right? That's what's happening because—and you're right—that money can either go because you need it to build the product or deliver the product at scale if you have a token cost. It can go because you need distribution, or it can go just because the other side has it and you feel the need for credibility.

I know in some of these wars, having looked at some of these deals in the ERP space and some of the other spaces where you're dealing with enterprise customers, the balance sheet becomes a criterion for qualification. So there's a whole bunch of reasons to say this is the way the game's been played now, and I'll admit I've taken a while to process and internalize that. That's just the way it is.

Rory, Jason, you're in a portfolio company and then you get one of the big funds that is trying to kingmake with one of your companies. Do you say, “Take the money. Fantastic. When the money's on the table, take it at a high price”? Or do you go, “This is a good company. That's a lot of money that could distract them and defocus the plan”?

Rory O'Driscoll

Pragmatically, in this market, I think you err on the side—if you have something that's working, and even if you don't know where you're going to put the money, if it's on attractive terms, you probably err on the side of taking it. If you're in a reasonably big market, you're going to have to grow and you're going to take capital.

So I think you err on the side of aggression precisely because you have to play the game theory. It's not just what do you think, but what are they going to do? Where is your competitor, right? You can say you're going to be careful, slow, and rational, but if they're not careful, slow, and rational, then you just end up outclassed.

As I say, I've resisted the kingmaker word, but I think where you are correct, Harry, is that capital has consequences. It has consequences for customers because it has consequences for hiring. I think one of the biggest drivers of when you take capital and when you don't is, unfortunately, that you have to look at the competitive dynamic, which I don't love because you want to steer your own ship. But if you're in a competitive market, it's hard to—

Take the money. Jason, should founders take the money?

Jason Lemkin

That has changed over the years. There’s so much more information, there are so many more founders, and there are so many more quickly growing companies. Founders have already made their own decision. Now, I haven't had these conversations in a while, Harry. I've got, I think, 2 of my best current portfolio companies. One will consume an infinite amount of capital. The other has 6 years of runway. It's their DNA.

I could argue one is overspending, and I could argue the other is underspending. That'd be a very easy argument to make, right? It doesn't matter what I think. Their DNA is different. Their customer is different. The market is different. All of it. But especially their DNA is different.

Some folks want it, and they understand the downside—or they don't—but they get it. They get that there's some risk, but they're effing going for it, or it's what their friends are doing. But I don't find that it matters. It no longer matters what I think. I can't influence it.

8. Chamath Raises Latest SPAC: SPACs are Back

So the best I can be is, if someone asks me, I can tell them at the last minute and the first minute, and I can try and influence it, but I can't change how they're going for it. There's just too many. There's just too many modes out there today.

Chaps, is there any other topic before we do a quick fire that you would like to cover or think we should cover?

Jason Lemkin

Oh, okay. I'm just going to say it here: Likely Chamath's terms are almost legit. There you go. It's like the new SPAC. I mean, I will say, the new SPAC terms—the SPACs, which were an alternative mechanism for going public in 2021—they've been around for a long time, but exploded in 2021. The subsequent return of those investments was miserable.

It became obvious that one of, not the only but one of, the primary reasons for that was that the incentives between the person sponsoring the SPAC and the investors in the SPAC were misaligned, and the sponsor made money simply by getting a deal done. It's like a venture capitalist getting paid your 20% just for investing money in the ground.

It turns out that if you do that, money gets invested, and now the terms are still not cheap, but you only make it if the stock—if you make at least a 50% appreciation on your stock. So it is a more rational structure. I still think there's issues with it around incentives and uncertainty. I'm still skeptical it'll be an amazing replacement to IPOs, but it definitely is less egregious and less misaligned than the last time.

Does it prevent Chamath from having the ability to pump and dump, like he's accused of doing?

Jason Lemkin

That has nothing to do with that. It doesn't. You use the wrong word: “prevent.” There are 2 words used: pump and then dump. Pump is interesting in the context of SPACs. The whole thing about SPACs is entirely separate.

The SEC has really tight laws about what you can say about an IPO. Really, really tight. You can't make any forward-looking statements at all. But bizarrely enough, SPACs are exempt from that because it's a merger, so you're allowed, from a legal perspective, to articulate any forward-looking story you like.

So anyone—let's not pick on anyone—can pump all they want because they can. I mean, it's such a weird difference. You've got one company going public in an S-1, and you can say nothing about the future, and then you have the other one going public in a SPAC, where you can say it's going to be freaking amazing. The next 10 years are going to be enormous. I'm tweeting like crazy.

So the pumping takes place because of the regulatory thing, but you're right: as for the dumping, you can still dump, but at least you have to get the stock up before you can dump it. Before, literally, you had the odd circumstance: the investor could come in at $10, and the stock could go to $5, but the sponsor got their stock at a penny. So even at $5, they could dump, and the investors lost half their money while the sponsors made money.

That's not a thing anymore. Now the investor comes in at $10. Until the stock gets to $15, the sponsor gets nothing. But once it gets to $15, they get a 30% promote. So it's not cheap, but it's a little better than before.

This probably makes SPAC a marginally more attractive competitor to the IPO, but I still think a well-run IPO beats it by a head.

9. Polymarket Raises $2BN at a $9BN Valuation

Rory O'Driscoll

Can I just add one last thing to deals like this? The SPACs—the reason it comes up is SPACs are back, right? It's really a sign of the times. We'll run out of time. Related to that, good God: the New York Stock Exchange invests $2 billion in Polymarket, which was essentially illegal—illegal—last year. It was essentially illegal.

The Biden administration decided this was gambling, offshore gambling, and was going to, as I understand it, shut down Polymarket. Trump—and listen, no politics here, okay? But it is interesting. The world changes. Trump comes in, his son joins the board and invests in the company.

David Sacks, whom I am a fan of as a SaaS founder, is trying to remove all the regulations from crypto and all this. Now you go from something that was essentially illegal—illegal—last year in the US to something that Trump Jr. owns a significant share of, and now the New York Stock Exchange is investing $2 billion at a $9 billion valuation.

I mean, if that isn't a sign that SPACs are back, if that isn't a sign of the times, it's such a change. Forget about AI: AI isn't the only thing in the world. Just removing all this regulation, and now that Polymarket's on the inside, with the New York Stock Exchange investing, to me, that's quietly the craziest story. And the legitimized self-dealing in it—it's just a different world, right? I don't know if it's better or worse—no politics—but it sure is different.

I think the deregulation is great, to be clear. I'm not woke. I think the fact that the Biden administration chose that hill to die on was both probably wrong and definitely stupid, which is the low quadrant and goes a long way to explaining their terrible polling numbers among the 20- to 30-year-old male demographic.

I actually give credit. I think the deregulation is one of the few joyous parts of all the rest that's coming out of the current administration. I think it's great. Right. Go team.

Whether or not—I'm not going to comment on the interpersonal relationship it takes to get something done—I don't think it impacts the deregulation. But, separate from that, yes, that's a morass that I have no doubt would be unpicked on another day by someone other than us with judicial powers.

I think the interesting thing about those businesses is, if you look at the volume—and I didn't check, I think I checked—Kalshi is still 70–80% sports betting. But the real question is, can you build prediction markets for other things? You're seeing that, and to the extent that you can do predictions for other things, like the questions we're going to be asked—I just saw one: “Who's going to be the next prime minister of Japan?” It was running because of the election.

Whether or not there's a, quote, legitimate, non-sports gambling business there that someone like the—was it the NYSE, which is really Intercontinental Exchange, that Atlanta company—is the core owner of that, whether or not that's a non-sporting business is TBD. But I think it's super interesting. I think those are 2 really interesting companies, and credit to the venture investors who stuck with it during the regulatory time. I think you've got a really nice asset now.

The thing that I did find a little bit confusing is that $2 billion is a huge amount of money. Again, comparing that to Vercel's $300 million—and this, I don't know, but there isn't a compute infrastructure spend, I guess, in a lot of cases. But $2 billion? Did you read—

Jason Lemkin

I think we have to see the details. I think there's some level of implicit exclusivity here. There's data sharing, right? You invest $2 billion for it to be, in essence, a subsidiary—well, maybe a subsidiary, an affiliate—of the company, right?

There may be some vague similarities to the beginning with OpenAI and AMD. It's a bonding investment. This is buying a quarter of the company, or 20-some-odd percent, and some sort of exclusive. Someone's putting something in here more than money. In return, they want to own a lot, right?

These guys are strategic investors. They're not doing it for a 2× return. They're doing it because they own—I can never remember, is it the NYSE or the Nasdaq? I should have checked—they own a bunch of market-making companies.

Intercontinental Exchange, owner of the New York Stock Exchange.

Jason Lemkin

Yes, exactly. ICE. They are called ICE, and I was going to say “ICE,” but of course that would be confusing to the average reader—listener, right? Intercontinental Exchange. They bought the London International Financial Futures and Options Exchange about 20 years ago. They're like, “If there's an exciting and interesting financial market where people buy and sell electronically—really interesting stuff—we like to own some of that.”

So it totally makes sense strategically. Good for them.

10. Quick-Fire Round

Okay, we're going to do a quick fire. Good addition there, Jason. By the way, I totally agree with you. Nuts deal. Nuts.

Number 1: This isn't official—Tim Cook leaves Apple this year. Yes? $100 turns into $879.

Jason Lemkin

No. $100 turns into $107.

I mean, is he already confirmed? Those odds are terrible.

Agreed. “The belief” was the word, right? I hate these things because I'm riffing, but Jason can do it in real time and look things up. Wasn't there some kind of succession-planning announcement? I can't remember whether it was that the VP of engineering would, in time—I should have looked it up. I didn't.

So there's clearly a succession path here, as one would expect if you're a competent board and you've got a chief executive over 60. But leaving this year, I think, is a very tight bet. The only way that would happen is if you thought things were failing.

So I actually know why it's a very modest payout, because it's very unlikely. That's different from saying, as a separate comment, that the board has started to think about what the transition is and what skills are required for the next leader. I think—I could be just imagining this; it's been so crazy this week—but I think it was the SVP of engineering. So it's very much a—

Yeah, that was a leaked rumor that he was a successor. He's 50, and Tim Cook's turning 65.

Yes.

So it's their job, but there's no way it's going to be this year unless it's a health issue. There's no way it's going to be this year.

You're exactly right, Harry. I think both sides of that are great. It's not this year, but it's exactly their job. If you're on the board of the second- or third-most valuable company on the planet, your CEO is hitting 65, and you're not thinking about succession, then just call yourself the Disney board and give up.

You're still a holder. I'm still a holder. I trimmed a little when Warren did. I got in before him.

Trimmed a little—99%.

No, no, no, no, no. And a whole lot. And, you know, I continue to worry about the growth rate and continue to worry about the AI story, but fundamentally, you need instantiated physical products to consume all this stuff, and they're the platform of choice for middle-class consumers and up worldwide.

So it's still been good. It's bounced back nicely from where it was. It was a little low there, and I'm like, “Ooh, maybe I was wrong.”

I don't get these VCs that invest in the public markets, but it's good content.

I feel the need to defend myself. I have almost no individual public stocks other than companies that I get distributed to, but I also wisely bought in 2009. My basis is so low now that I just can't bring myself to pay the 37% tax.

Don't sell it until you move to Puerto Rico or whatever.

No, I'm not going to move out of California, but I just can't bear to give it up. But maybe I'm just stupid.

Just borrow.

Okay, next one: Replit and Lovable, over or under $250 million in ARR by the end of the year? They're both in the $160 million range now. $160—

Jason Lemkin

Is there a caveat, or do you just say—is this a binary question?

This is a binary question. This is a Harry question.

I'll tell you the only interesting thing. I would say over $250 million, but barely. There was an interesting report this week. It wasn't Bloomberg; it was someone at Barclays. Barclays tracked what they believed the web traffic was to all the categories, and what was interesting was that it looked very accurate.

It had Base44 hitting the numbers that tied to what Wix publicly disclosed, so we can assume that's accurate, right? It had Replit tying to when they launched V3, which was a big boost, and then it had Bolt's numbers, which roughly tied to what I think they are.

What it said is that traffic had flattened or was down. Initially, you might say, “Oh my God, sell your stock.” But I actually think it's a good thing, because the looky-loos aren't going to renew.

The problem with these products is that you have to segment churn. There are folks like me who will never turn off Replit. I'm 200 hours in. I've got 8 apps into production. It's impossible to leave. The odds are that I'm going to spend somewhere between $300 and $3,000 a month for a long, long time.

But Abigail, who wanted to build her own CRM and was told she could do it in 60 seconds, and it didn't work—she's going to churn, right? Or he's going to churn. So it's good to get rid of the looky-loos, because even though they got these guys to $100 million, they aren't going to get them to $1 billion. They're just going to churn.

We've all had companies where this is the case, where we had a segment of customers that were very high-churn and a segment that's very sticky. So I think it's okay that interest is flat, because it'll be higher value. But it is something to reflect on.

It looked right to me that we've seen at least a temporary plateau in interest in these platforms, and it makes sense to me because a lot of use cases aren't viable. I love Replit to death, but all of this is at the edge of marketing misrepresentation.

So, rambly answer: I think they'll hit it, but I do think there's deceleration. It's not like ChatGPT—there's going to be deceleration in lay users wanting to use these platforms.

For the record, there's deceleration in ChatGPT too. It goes back to what I said earlier: what rate of deceleration gets you from $10 billion? How much deceleration can you have at $12 billion to make sure you still hit $100 billion? That's my question. But we don't have time for that today. I'm thinking about it.

Next, Harry, I don't have anything to add on Lovable and Replit. I'd defer to Jason entirely.

Jason Lemkin

They're getting better. Here's the thing people miss: I'm almost 100 days into vibe coding, and the platforms are so much better than they were 100 days ago.

In a lot of things in venture startups, you've got to be careful how you predict, because these are not—this is not SaaS of 2016.

Okay. The rate of improvement is so high that’s why I feel pretty good about the ARR numbers, even if I think the latecomers may fade.

Guys, thank you so much. This has been awesome.

All right. Rock on, Harry. Thank you for the time. Thanks for doing this for us.

Totally.