The IPO Wave That's Harder to Price. What Investors Are Paying For | Aman Verjee

We now have a number of companies that are worth over a trillion dollars market cap in the stock market. We have a few companies. SpaceX just went public over two trillion dollars of valuation, and Anthropic is about to go public. It could reach those type of valuations. Today we sit down with Aman Virji. He was the CFO of PayPal and he took a number of companies public. And now it's at the table I think for Anthropic to go next, another amazing company. They're up 10x year over year, which I've never seen before for any company that big at this scale. I don't think it's ever happened in US or any corporate history. They're growing probably another two to three x next year. So I think when they go public, they'll tell investors they're targeting probably 110 or 120 by the end of the year on their way to maybe 300 billion in 2027, 2028. I'm Siamak. Welcome to Marketing Insider. Aman, it's great to have you back and welcome back. Thank you. Great to be back. We want to talk to you about a lot that's happening in your area. There's a lot of IPOs and you're in the middle of a bunch of them. You're a major investor in one of them that we just, SpaceX, we just went to public and you had some stakes in it, in your organization. And there's Anthropic going public too. Can you tell us what's going on with these IPOs? What are you seeing out there? Yeah, I think overall it's a very exciting time. We've just had the biggest IPO in history that was SpaceX. And as you said, we were investors in SpaceX. You know, normally in a good year for IPOs, the companies that go public will raise 50 to 75 billion, like all of them all together, all together in one year, like before 2020, that was like a, that was like a good year. 2021 was a really big year, 150 billion. And then all the next three years were zero to 20, like very, very low zero to 20 billion. But I think we'll be back to a record this year. We'll be over 150 billion. SpaceX just set the record for the biggest IPO of all time. They raised 75 billion in their offering, raised another 11 billion because it had a follow on or a green shoe. And then they raised a debt offering to get to a hundred billion altogether. And they've done really well. I think it's been a company that's been executing really well. We invested in them as a space rocket launch company. They then added Starlink, a communications company. And then earlier this year, they added an AI company, XAI, and are now really an AI company as well. So they have three different business units, but I think all of them are progressing well. They're working together and it's now like a $2 trillion company, which is amazing. And now it's at the table, I think for Anthropic to go next, another amazing company. And then I think we're, there's a long list of companies that we're looking forward to getting public as well, but this year is so crowded with just those two and a few other big ones like Cerebrus, that I think 2027 will be an even bigger year for IPOs, but it'll be a lot of smaller, but more interesting companies. What do you think of the Anthropic IPO and the valuation behind the scenes? I think they're amazing. They're running at a $74, $75 billion ARR run rate as of now. So their revenue is $75 billion. Yeah. They call it annually recurring revenue. It's the kind of the current month revenue times 12, but they're up 10X year over year, which I've never seen before for any company that big at this scale. I don't think it's ever happened in US or any, any corporate history. They're growing probably another two to three X next year. So I think when they go public, they'll tell investors they're targeting probably 110 or 120 by the end of the year on their way to maybe 300 billion in 2027, 2028. And so they should price like a Magnificent 7 company. They could probably be $2 trillion on the day of the IPO. And I think they're probably worth $3 trillion. They probably got close in Q2. I haven't seen the S1 because they haven't made it public yet. I think they will probably be there for the full year. I think they could, most of their, most of their revenues are now from inference revenues. That's 80, 85% gross margin. If they're making 120 billion in ARR, I don't know how you can spend so much money that you're not profitable on those gross margins. But they are pulling in a lot of cash. They're spending a lot of capex, but it seems like they are already at, at, you know, cashflow breakeven or will be by the end of the year, which is, which is amazing on top of the growth rate that they have. Is it just the $20 per month, $200 per month subscriptions that's driving this growth? Or is there a token customized services that they're offering companies that we may not know about? Is it like specific deals? Yeah, it is. It is all tokens. They don't really have much of a consumer footprint compared to OpenAI. They're way behind, or even Gemini. They're way behind on consumer and consumer adoption. They were behind about a year ago. At the end of last year, I would have said it was OpenAI number one, Anthropic way behind at number two, and then Google Gemini trying to get in the game. OpenAI really had the lead on the consumer side where they charge $20 a month and have some other tiers for business users. Anthropic figured out something very different, probably around February, they launched something called Opus and they have a coding agent, Clod, which is basically it's a, it's a coding assistant. So it helps computers, software developers to code things more easily to automate workflows that are very expensive or that have a lot of labor costs. And companies have just been eating that up. And they're not really pricing by seat or by subscription. It's mostly outcome-based. So if it said that every token has a certain price and they're pricing by token, and that has just exploded. And companies, a lot of small businesses are using it. SMBs, enterprises are beginning to use it as well. And that has just been the most popular products, probably in software history. We're just driving all the revenues and all their profits. A lot of people wonder what happens to our data when we're using these agents and they're working with the data. Do you have any thoughts on that? Where does the data go and who has access to it? Is it clear? It's a little bit unclear. There are different, different ways you can do it. If you work with an open weight model, for instance, what you will get is the algorithms and it'll be, it'll be something that you can download and control on your own premises. You don't have to put the data in the cloud. You have full control over your data, your privacy. You're just essentially getting a software and. And it's your own computer environment. You can keep it in your own environment and you can have full control over it. That's not how cloud works. That's not how Anthropic works. It's not how open AI work for them. It's all proprietary. It's a closed source model. So they have control and they have use of your data. It's, it's limited. There are going to be limitations on how they can use that data, what they can use it for all based on contracts in terms of use. But it is very easy for them to use that information in a way that you wouldn't have expected. There've also been cases where in, in both cases, now open AI and cloud, they had models that broke out of their sandboxes and created a cybersecurity risk. And then in the case of, uh, in the more recent case with open AI, the model got out of their sandbox. It actually took, it actually jumped into hugging face, which is a repository of code, hacked hugging face, got into their software and then opening. I had to, I had to shut them down only after, you know, they discovered it and hugging face had to discover it ironically with an open source model. But, um, these things are very tricky to control, to manage. I think we've seen how social media companies use data in ways that you didn't really expect. And so a lot of, a lot of people in the Valley are concerned about working with cloud and working with Anthropic and working with open AI because they're not sure that they have fully safeguards on their data. And what about the Chinese models? Do you have any thoughts on the Chinese models? Yeah. You know, it's funny. A lot of them are open weight. Um, people use the term open source and open weight a little bit interchangeably. That's not really right. Open source is what that really would mean is that all the code, all the algorithms, the training data, it's all available to the public for everyone to see in the public domain. That's not really how the Chinese models work. So deep seek and, um, Kimmy are probably the two leading ones right now, cleanse another one. They're open weight, meaning they, they, the code is proprietary. The training data sets are proprietary, but the, uh, and the algorithms are proprietary. It's just the, the weights and assignments and how they build the models and how they wait, the different answers are open to the public. So you can control that. What you can do with those open weight models is again, I think you can have, you can have a lot of control over your own personal data. You can download the model from them. You can install it on your own, um, on your own premises. And then they won't be able to touch or use that data. All they'll be able to do is provide the, the code and the algorithms that manipulate the data, but you have full control over the outcomes. It can customize it for what you want to do. So if I were to use a Chinese model, that's, that's the only way I would do it. I would not hand over my data to a Chinese company because the governance there, you know, is very different. In a sense, are you seeing these, um, the spending in AI that, that is the main drivers, open AI and Antropic and these companies that have the customers or is there other drivers? Yeah, that's a good question. You know, AI is, um, it's, it's more complicated than the software business because it's software and hardware. So let's start with the software piece. There's an application layer. These are companies like Harvey and Cursor and, uh, they face consumers and you use these things to do things like legal document review or a coding assistant. Uh, my daughters might use it to, you know, research school projects and stuff like that. There's still not a lot of money being made in that part of the, that part of the AI stack, the application layer. It's probably not more than 40 or $50 billion in revenue, current course and speed, but it's a growing very, very, very, 40, 50 per year. If I add up Harvey and Cursor and you know, a few of these other consumer applications, the next piece is the model layer. So open AI and Anthropic is most of where it is. That's probably 110, 120 billion a year, but again, growing very, very quickly. I think Anthropic's at 75 billion. Open AI's at 40 billion. So add those two up, you're at, you know, 115. And then there's a few others that are in the running, but really that's where most of it is. So that's where the, that's where most of the spend is. And I think it's, again, it's growing like three or four X a year. And then all that infrastructure layer and that software layer needs compute. It needs very, very powerful computers to run on, which is very different from the internet. And there is hundreds of billions of dollars being spent every quarter, like just the four big hyperscalers. So it's Amazon and Alphabet, Meta and Microsoft. Last quarter, they spent $168 billion on CapEx for the compute around all the stuff that's needed to run these computer programs. So they're AWS and Azure and Google cloud. They're making a whole bunch of money in SpaceX now. We're all making a whole bunch of money on just providing that infrastructure layer. And that's going to be a trillion dollars a year. So when people say AI spend, you know, the first piece of it is like, what are we, what are we buying as consumers and small businesses? And then there's all the other AI CapEx spend. That's like a trillion dollars a year and counting, which also matters. And that's very important for investors because they're asking right now, will all that CapEx spend be justified? Will there be enough revenues from the models and the applications and the school homework assignment, you know, applications and programs and the tokens, will that ever justify the trillions of dollars that are going into the CapEx part of the stack? What do you think of that? Do you think it's justified? I think it's, you know, it's hard to tell right now. It's still early, but the amount of money going in is probably, I don't, I don't foresee a trillion dollars a year going into the CapEx spend on an annual basis. I think it's, I think we're in year three of a, of a buildout and maybe it goes to year four or year five. But there will have to be a reassessment where the amount of money being generated in revenue has to recalibrate to the cost. We've seen in prior bubbles, we've seen like in the telecom boom in the late 1990s, we spent about 10% of US GDP on telecom and fiber optic cable thinking that the, Oh, well the internet will have so much demand that all that will make sense. And at the peak of the bubble, like 97% of the fiber was dark and only 3% was being used for the internet. And then, and then, you know, the market crashed and companies like Cisco and 360 networks and WorldCom and MCI, half of them went bankrupt and half of them barely survived. But then we came out of it and eventually the demand, you know, it took a while to the The difference this time is that every GPU that NVIDIA rolls off of its assembly line, it's lit up the next day. So there's no, there's no dark fiber, you know, being laid. So far the demand has been enough, I think, to justify the supply. Not everyone's paying for it, but I think there are a lot of business models we've seen now with Anthropic, OpenAI. I think we'll see a lot of willingness to pay over the next couple of years. So it feels like we're, you know, we're in a better environment than 2000, but I think within the next year or so, there'll have to be a recalibration of CapEx versus the revenues. And you mentioned Anthropic IPO and SpaceX. Some people said that the valuations are not justifiable, especially for SpaceX. And you're an investor there. So what are your thoughts? Well, I like SpaceX. So just in disclosure, we invested in them. I got shares personally from the IPO and SpaceX. I feel really good about the company. I'm not selling those shares. You let them keep them for a while. We have a lot of investors and they're going to ask us, what should we do with our stock? You know, when we do the distribution, I'll tell them, you should do whatever you want to based on your financial situation. Here's why I like, I understand the valuation concern. They did $18 billion in revenue last year. They IPO'd at $1.8 trillion. So that's a hundred times revenue. Yeah. So just based on, if I stop right there, I'm like, I can't, the spreadsheet doesn't make sense to me. Like the spreadsheet that stops next quarter doesn't make sense to me where this valuation makes sense. Why am I buying a company at a hundred times revenue? What they said on their earnings call at Q2 was, first of all, there's three business segments. There's the space launch business, there's Starlink, which is basically going to be $10 billion in revenue and doubling year over year and profitable. And they have a lot of room to grow on Starlink, I think, which is just a game-changing connectivity program. It's internet, it's broadband, it's rural areas, it's countries that don't, you know, don't have access to broadband. So they can, that can be, and it's the, it's by far the best and most reliable internet in the world. Pretty soon United and all, you know, American will have Starlink flights and you'll like, you'll be able to get the internet on airplanes. Have you ever tried to use the internet on airplanes? They don't work. They never work. Sometimes they work and then they stop. Like, and that, by the way, is a, that's a metaphor for the internet in many parts of the world. Like I know 3 billion people in North America and Europe and Western Europe and wealthy countries have internet that works and it's a, it's a commodity and it's like, you can't do without it. There's 3 billion, 4 billion people in the world who have airplane-like internet, you know, Saharan Africa, parts of Asia, where you just don't have reliable internet. And Starlink really has the opportunity to change all of that. And then most of the revenues come from this AI business, which we didn't even know when we invested in them 3 years ago, 4 years ago, 5 years ago, we didn't even know they had an AI business. This was like, you know, Elon had bought Twitter and then he started XAI on the side and it was a side project. Like, what are you doing with a side project and Twitter and why are you messing with all this stuff? Just focus on getting spaceships into outer space. And then it's like, you know, what I think you realized was open AI and Anthropic, all these computer software players, they need compute, they need capacity. And that means you need data centers. And that means that you need places where you can actually run all these algorithms and programs. And Amazon and Google and all these guys can't keep up. So he stood up Colossus, this great big computer in Tennessee, which is providing 1.3, 1.4 gigawatts of power. All of Anthropic's $70 billion business today runs on about 1.3 gigawatts of compute. And he's able to provide $1.3 billion of the capacity. And Anthropic is paying him $12 to $15 billion a year just to rent that facility. And then Google is like, we'll pay you $12 billion a year to rent that. And now he's talking to, you know, all these other providers are like, we need capacity too. He can build and print these AI factories and data centers faster and better than anybody. He wants to put them in outer space. So not just terrestrial data centers, but put them into orbit. And in his earnings call, he said, we'll be at a hundred billion dollars in ARR by the end of the year. So now we're talking 16, 17, 15 times ARR, which if I look at the broader, you know, software index, like that's not, that's not that crazy. Morgan Stanley says they could be at 300 billion in revenue over the next two to three years. Elon is saying they could be at a trillion dollars in revenue by 2030. Now, Elon makes a lot of promises that he doesn't always make, but even if he's wrong and Morgan Stanley is right, and this is a two to $300 billion revenue company within the next three to four years, the stock may go up. The stock may go down, but I think over time, like if you have a five-year adjustment. And here's something else to think about. And, you know, I reflected on this a lot when I, when I made the investment, if you're investing only for one or two years, then there's going to be a lot of volatility and you can debate whether or not that's the best place to put your money. And maybe you should, you know, you should, you should, um, if you are an investor in SpaceX, it would take some money off the table to manage your portfolio. My, my, my mother is 78. My father is 82. I tell them, be careful. Bonds are really good for you guys. Like you need money. You can't have money. You can't have the stock going up and down 50%. That's going to make you angry. And when you're old, you don't want to be angry and stressed, right? But I've got a 25-year horizon. My kids maybe have a 50-year horizon. I think we're at the beginning of an AI and robotics investment super cycle. And that means trillions of dollars will be invested in this new technology over the next five or 10 years. We saw the same thing with the internet in the 1990s. Um, and you wanted to invest in Jeff Bezos and you want to invest in Bill Gates and you wanted to invest in, you know, these, uh, these once in a lifetime entrepreneurs at the beginning of an investment super cycle, even though Amazon looked expensive on paper, it always felt expensive. Um, it turned out if you invested at the beginning of that super cycle and just held during the ups and downs, you would have been, you would have been fine. Henry Ford, the same thing in the 1920s, Alfred Sloan, go back to the 1870s and 80s. You have Rockefeller, you have Vanderbilt, you have Carnegie. Again, at the beginning of these investment super cycles that go 20, 30 years, all of them looked expensive, right? John Rockefeller's company never looked cheap on, on paper, but, um, but it turned out that was another great investment. And I really think Elon Musk is a once in a lifetime entrepreneur at the beginning of a very long investment super cycle. What about open AI and Antropic? How do you see Antropic IPO? Do you have the same thoughts? Like, are you going to invest and are you going to get involved with them? We did not invest in Anthropic as private, as a private company. Um, we, we missed that one. There was a little bit of tension between SpaceX and Anthropic. And so we ended up picking SpaceX and could never quite get into Anthropic at the right time, but I think they will do great. I think they are, you know, a hundred billion in revenue again by the end of the year, similar to SpaceX, but I think growing again at three X year over year, um, profitable as we said. So they could be at two to three, two to $3 trillion. And I'm sure their IPO will be volatile and it'll go up and go down. But I think if you can get in at that IPO price of 2 trillion, I think it'll be fine over the long run for the same reason. I don't know if Dario is the same kind of once in a lifetime entrepreneur as Elon Musk, but he's made every right decision over the past, you know, six or 12 months. And they've got a big lead now on, uh, agentic AI, which is really where all the money's being made in this area. It wasn't really around six months ago, but they've broken through in a way that I think is very, you know, very special. And I think they'll do fine. OpenAI is tricky. They are a little bit behind on the revenue side. They're probably 40 billion in revenue, not 70 or 80. Their last raise was 852 billion. Just did a tender offer at 852 billion. So, you know, they, they are able to do a buyout of employees, but I don't think they'll get public at the same time as Anthropic. We'll have to wait and see. So they'll, um, they're a little bit behind on revenue. They're a little bit behind on the fundraising, on the IPO path. The thing that's interesting about, you know, OpenAI, um, they have a big lead on the consumer side. So more people will use chat GPT then, you know, then cloud code for instance, and having that consumer is very, very powerful. We, I think we learned that with internet 1.0, but the thing that you have to be, I'm a chief, former chief financial officer. So I see public companies, not just as an investor, but also an operator. And I'll tell you the thing I don't love about Sam Altman. Um, and you might remember back in 2020, I guess, 2023, there was a board, it was a big shift on the board. He got thrown out, came back into the company and then two board members left, Helen Toner and Tasha McCauley. They brought in some new board members and then he brought in some new, uh, new members of the team and he kind of rebuilt the team. Their, their, one of their chief tech technology leads was, um, Moradi. She left the company. He brought in a new COO, just brought a new CFO, Sarah Fryer. She seems really good. The new COO, Brad Lightcap was shifted to special projects and then left. He brought in a head of applications, Fiji Simo. She had a health incident. She left and she's on leave. They brought a new head of marketing, Kate Rauch. She left. I'm like, you know, I don't know what it means. If you have one or two incidents, like where senior members of the management team board members, like they're leaving, there's, there's always some noise in any company, but if it's like one or two, I understand if it's like five, six, seven people on the management team, all changing roles, moving around within a year, 18 months, that just feels weird. It feels like a team that's not quite clicking on all cylinders, even though, even though they're all fantastic and they're superstars. It's like, you know, the U S Olympic basketball team, they win the gold medal every year, I think, but it takes them a couple of games, right? Just to kind of get there. Who's got the ball? I got the ball. I was supposed to be there. Okay. We'll get there next time. I can see it. And Sam and Sarah, the CEO and CFO, they're a little bit out of step on guidance on revenues and financials. And when do you, when are you going to go public? A very simple question, like, when are you going to go public? They gave slightly different answers. And if you're not a seasoned investor or a CFO like me, you might be like, ah, there, you know, I don't, I don't, I don't see what the problem is. It seems like a great company growing quickly with a lot of customers and the end. And why not? And I'd be like, I'm just seeing enough, you know, enough that even though it's not obvious to a lot of people, the team isn't quite gelling yet. I think they still need a little bit of, I'm going to guess six months to a year of just getting their game together, having all those people be in their seats, working together, giving guidance, hitting numbers. When you're public, everything you do is under a microscope. Every quarter, every guidance, you know, statement you make, you can't be off by a little bit. But what I want to see from OpenAI is like the Spanish World Cup team, where it's like, I don't know the name of that guy, but, but that, I don't, some of their players are stars. They're all good. All good. And they're working. They might not have one or two big stars, but they're all working together. They're, and I don't, and they're seamless. Exactly. They're working together. Their teamwork is flawless and they know where they're supposed to be. I think that just takes six months to a year for any company. And I feel like a lot of OpenAI is still a little bit new. So my advice to Sam would just be keep your head down and execute for the next six months. Don't worry about the IPO. Don't, you know, don't get in front of Anthropic. They are a juggernaut and they're going to raise a lot of money and stay, stay, keep your head down and let the capital markets, hopefully that'll be a good transaction. Let the capital markets sort that out. Take a fresh pause at, you know, 2027 is a fresh slate. You'll be the first and most valuable IPO in 2027. And all this money coming from SpaceX and Anthropic will go back to investors, right? In distributions, it takes six months or so. Out of lockup, let them get their money. They'll want to reinvest in the new hot company and you'll get your chance in early 27, but do it right. Don't try to be, you know, too early. I think Sam gets a little bit impatient sometimes, but that said, I love the company and the product. I think they're growing really, really quickly and they're just hitting their, they're hitting a stride now with their new model. So I think it's good. Now, Aman, you're part of this ecosystem. Now, a lot of these companies raise a ton of money before going to public and they take longer and you're involved in this, right? Can you tell us more about what's happening right now with these companies that are having these gigantic IPOs? And you were involved with an IPO yourself back in the day. Did you think we'll get to a place where we, people can raise billions of dollars? I didn't. That's, it's a surprise. You know, my first company was at PayPal in Silicon Valley and we were public in four years. So we went, we started in 1998. We went public in 2002. It was a billion dollar IPO, which felt like a big number back then. But of course, PayPal is worth 50 billion today. So 49 50ths of all the value ever created at PayPal was as a public company. And then I took, I helped Sonos get public, which is another consumer electronics company as their chief financial officer. Now these companies like Airbnb, Anthropic, we just talked about Palantir, Snowflake, they're staying private 15, 18, 20 years before they finally go public and they can raise billions of dollars as private market companies. And so the good news is, I guess, if you're a company like that, like Anthropic or OpenAI, you can raise the money. You don't need to go public before you're, you know, before you're ready. And it's great for the CEO because then as a private company, you don't have to disclose all the financials. It's an easier compliance burden. You're dealing with a smaller number of investors who really know your business and you don't have the volatility of your stock price. But if you're a shareholder in one of these companies and you're locked up 15, 18, 20 years, SpaceX has been public after 23 years. For all those SpaceX investors who were in early or employees to be illiquid for 23 years, it's just not tenable. So there's been a big rise in secondary market activity where buyers like ourselves, I run a firm called Practical Venture Capital. We go and find the companies like that, like the SpaceX and the Airbnb and Anthropic or OpenAI or Databricks or Stripe, and we will buy out those early shareholders. And we'll usually give them, you know, a price that the company agrees to. And then the employees or the early shareholders sell. They sell some of their stock. They sell all their stock. Maybe their timeline doesn't quite work for where the company is, but they want to de-risk their portfolio. And we'll buy those shares and then we'll ride out the endgame and let the company go public. A lot of venture funds, when you first invest in a company directly, it's tough. The first five years are complicated because companies come and go. You have winners and losers. You're right after losers. A lot of fees come out of the fund. And your returns as a fund don't look very good. They call it the J-curve. It's the part of the investment cycle in a fund where your returns aren't typically very good for the first five years. But then the J-curve means your IRRs accelerate, usually after year five, year six, year seven. We like to skip the J-curve and come right into a fund or a company just as those returns are picking up and things look good. So PayPal just had a deal with Stripe. Can you tell us more about it? And what do you think of PayPal? How's PayPal doing? They're working on the deal. They haven't agreed to anything yet. PayPal has been, I'd say up until 2000. So I left the company in 2010. They were growing every year. They were profitable every year. When I was running finance there, we would grow at 30% a year was our average. So things were good. And eBay had acquired us. And so we were part of that eBay company. And then they spun off from eBay in I think 2015, 2016. This is after I left. And I think they did fine for a while. They actually built a really good payments franchise away from the eBay platform. They had become one of the most popular ways to pay for consumers, 440 million consumer wallets and all over the internet and all over the world. And then during the bubble, during the pandemic bubble, they got to $300 billion in market cap. And then I think they began to make a couple of mistakes. They moved towards a unbranded checkout solution, which essentially means you're white label processing on the back end for payment providers. Very thin margin, very difficult to make money doing that. So the gross margin compressed. The rise of like Apple and native consumer wallets, Apple being the big one, but a few others meant that they were losing share on the consumer wallet side. So if you go and buy something, you know, on like Fandango or something, you want to buy a ticket on Fandango, there is a PayPal button there, but all you have to do is hit the Apple pay. And it just, it is a very quick process, you know, process payment. And PayPal couldn't quite figure out that one click checkout solution that, that Apple pay was able to get to. So they're losing share, I think of the market space, the gross margins are compressing and the stock price went from the value went from 300 billion back down to 50 billion. So that's not great. They made a couple of mistakes, I think on the crypto decentralized world, they missed the chance to be, you know, the next Coinbase or Gemini. They're, they're a wallet with currencies in them, but they didn't see crypto as a currency until it was too late. And then by then they, you know, other people came along and took over the crypto exchange space. So they're a payments company that's not growing very quickly. That's, you know, profitable, but not really driving efficiency or profitability. In fact, their profit margins are 21, 20, 21% based on the last quarter. When I was there in 2010, we were at 21%. I'm thinking, how do you, how is 16 years later with all the scale, all this growth and AI to make your business better and more efficient, you're still at 21% margins. Like that doesn't feel like you've made a lot of progress. So they've, they've changed CEOs in the last year, in the last six months. I think the, uh, the new CEO has just taken is just taking the role. And I think within, within weeks, there's an offer on the table from Stripe and, uh, Advent to buy PayPal, which I like a lot. I think Stripe's an amazing company. They're probably 150 billion in market cap, private company. So the only, they're only offering private stock. Advent's a private equity firm. They have cash. I think that could be a really good lineup because Stripe has done a great job of acquiring merchants and payment processing. PayPal is strong in the merchant side, but very strong in the consumer side. Stripe has never gotten good on the consumer side. If you take that consumer network that PayPal has and that merchant network, right? That Stripe has and PayPal has now you have a much stronger network and you can move money from those PayPal wallets to the Stripe merchants. And if they get, if they're smart about it, they can, they can do an end around right around Visa and MasterCard. Anytime you move money from a consumer to a, to a merchant, you really have three ways to go. You can go through Visa MasterCard rails. It's going to cost you the processor 2%. So whatever you're charging to the merchant two quarter, two and a half percent, most of it's going to go away and interchange fees to Visa MasterCard. So for people that are, whenever you're buying something using your Visa card, that merchant, whatever it is, a restaurant, coffee shop, they have to pay 2% to Visa or MasterCard. It's a little worse. The coffee shop pays a merchant acquirer or a bank, and they're probably paying them two and a half to 3%. Then the merchant acquired in the bank said, thank you for that 3%. I'm now going to give 2% of that to Visa, MasterCard, Amex, you know, Discover, whoever you, wherever your credit card provider is. And then I'll keep that little extra money as my profit. But for that, I have to go through the, you know, the marketing, the acquiring business, the underwriting, that machine on your, you know, on your terminal. I have to, that's a lot of expense and processing payments. If there's fraud, they're on the hook for it. So all the fraud models, all their collections, you know, all that infrastructure they have to pay for. And then Visa, MasterCard just make a ton of money, 2%, 2%, 2% every single time. And they don't, you know, there's a lot adds up on all of your transactions we use on these cards. Yeah. So they're trying to get into that. I think Stripe would like to do less Visa, MasterCard, and like to route it through either a bank, ACH funded, or just their own balances. Like if you have money in your PayPal balance, you don't have to go through Visa, MasterCard. It goes right from your balance into the merchant's pocket. And then you can avoid that 2% fee. And the best thing of all, I think is they could also have a, the third way is a crypto solution. Stripe has messed with that. They haven't really figured out how to, how to get into crypto, but they've built a lot of the rails for a U.S. stable coin or a cryptocurrency token. PayPal actually, I think in 2024 launched something called PYUSD, PayPal U.S. dollars. It's a stable coin. So it's a PayPal currency backed by the U.S. dollar. So it's valuable. And they're one of the top five or maybe top 10 stable coin issuers in the game today. They're not making any money on it, but if there's an, if you have enough stable coin and you have a Stripe and PayPal merchant base, and you've got stable coin in your PayPal wallet, along with Bitcoin, along with euros, and maybe some U.S. dollars, maybe you travel to Europe, you've got some euros in there, and they can route money from those balances to the merchant and avoid Visa, MasterCard. I think they can actually do, I think it could be a really good deal. So I kind of hope that they get to a good answer. I don't know that PayPal wants to sell to Stripe because their CEO is still pretty new and maybe he's holding out for a better offer. I can't blame him for that as a PayPal shareholder. I would like him to get a better price, but I think it's a really smart acquisition. You recently wrote a book on bubbles. Going back to the AI conversation, do you think we're in a bubble? And you went in and did a lot of work in researching the history of bubbles in your book. Yes. So are we in a bubble? You should buy the book and then you can answer. I'll give the answer for free right now though, I think. I think there are a lot of differences between where we're at on AI and then the prior bubble. So first of all, what is a bubble? A bubble really is you have a big run-up in value and people get very excited and overpay for something at the peak of a cycle. When all your neighbors and everybody's getting rich and you're the last one left. Sure. This is some psychology about a bubble. And then the bubble pops, which means the value goes away and doesn't really come back. And then you're like, ah, I got excited. Baseball cards or tulips. The first bubble is the tulip bubble in Amsterdam, where there's a big run-up in the value of these tulips. It goes up by 10 times from like late 1636, early 1637. And then it all just goes away like that, like a bubble. Well, let's compare AI to, let's pick one like the 1999 tech bubble. So a couple of differences, the valuation environment is very different now. Like the NASDAQ 100 right now is about 25 to 28 times earnings. In 1999, 2000, it was 75 to 80 times earnings. So we're in a very different environment now than 1999. Second thing is the company leadership is very different. Back in 99, 2000, Cisco was the biggest, most valuable company in the world. They were laying the fiber, the dotcom fiber, you know, the networks and the internet. They were building supply ahead of demand. They made money. They made about 2.7 billion in gap net income in 2000. And then they lost money in 2001. So at their peak, they were at 200 times earnings. And then they lost money in 2001. The company in the middle of this bubble, if you will, this AI cycle is NVIDIA, which is trading about 18 times this year's fully taxed earnings, maybe even 15 times. I have a different set of projections, maybe than the Wall Street analysts think. I think when all is said and done, NVIDIA will be at 15 times earnings. Wall Street now is probably more like 18 to 19 times earnings. Either way, that's not the stuff that bubbles are made out of. NVIDIA has spent 113 billion buying back stock over the past decade or so. Cisco was not buying back stock. They were not paying dividends. The NVIDIA is selling mostly to the hyperscalers like Amazon, Alphabet, Microsoft to Meta. Collectively, those hyperscaler customers who are building out the AI infrastructure have 300 billion in cash on their balance sheets. Back in 2000, the customers were telecom companies who had a lot of debt. They had $300 billion in debt on their balance sheets in 1999-2000. It turned out those telecoms weren't making money. We kind of thought they were, but they all went through accounting scandals in 2001-2002. Enron, MCI, WorldCom, Williams, all of them went bankrupt in 2001-2002. Quest had to restate earnings and almost got zeroed out. AT&T did a bunch of impairments. 360 networks had to do a bunch of write-offs. One after the other, all those companies were built on fictitious profits and cash flows. If you look at the hyperscalers, they are all making money. They're all profitable. They're all buying back stock. They've all got net cash. It doesn't seem like there's any major issue there. For those reasons, I think it's a different space. The last thing is, I think AI is a real technology that is going to leave a real, lasting, positive impact on the world. You can say the same thing about the dot-com internet. Although the valuations went up and went down, it left a very valuable technology in place. The UK railway boom, I studied that in 1845-1849, left a very valuable technology in place. Prices went up, valuations came down. A lot of people lost their money. From 1850 to 1873, the UK went through a very long, they call it the Great Victorian Boom, a 25-year expansion in economic activity at the end of the Industrial Revolution that made everyone better off. Coming out of the internet, I think we're all better off because of the internet and connectivity. If you just bought those companies during the internet boom, Nvidia, Amazon, PayPal, where I worked, eBay, Google, Netflix, Salesforce, Broadcom, they were all founded in that late 1990s period. Even when the bubble popped. My first two bosses were Elon Musk and Peter Thiel. When the bubble popped, PayPal survived. Peter Thiel took his money and became the first big check into Facebook. Elon took his money and founded Tesla and SpaceX and the Boring Company and bought Twitter. The value created created a lot of value for shareholders. You have to have the discipline to buy those companies, hold. You would have had a lot of losers. You would have had Webvan. You have to have the discipline to hold during the longer period, not sell in 2001, 2002, 2003 when everyone just dumps the shares. In fact, you should use that drawdown to buy. I think if AI does have a drawdown, people would be wise to buy into that downturn. From what you're saying right now, even though the valuations have gone up significantly, it doesn't seem like the market is valuing these companies the way that we were valuing them in the dot-com bubble. It seems like the market is cautious. I think the market actually is. And they're assuming that these earnings are not true or they're not going to last. They're probably more the latter. They're careful about the longevity. I think there's a lot more skepticism about AI than about the internet. Everyone was very excited about the internet, about social, economic, all these profits can go on forever. AI, I think there's a lot more, there's more of a wall of worry around AI than there was around the internet, which is good for the sector because that skepticism keeps the bubble from forming. The NASDAQ is up about 15% year to date. The earnings that are just got declared in Q2 are about 50% up year to date. So on a multiple basis, the multiples have actually compressed because even though the stock market is up this year and last year. But it hasn't gone up 50%. It hasn't gone up 50%. So earnings are going up faster than valuations, which was not the case in the internet. Now your book is climbing up on the ratings, right? Can you tell us more what's happening with it? You just released it a couple of months ago or a few weeks ago, right? Yeah, we just did the release August 12th, I think was the publication date on Amazon. So it's only been really two weeks. It's in the top 50 now on the bestseller list on Amazon, which is great. But we're scrambling to get it out because it's a brand new launch. We had some launch events and we only did the paperback. So only the paperback is selling. And I'm told everyone wants to buy the e-book and everyone wants to do the audio book because they don't like to read. People don't like to read. I don't know why. So when we do the e-book in the next week and then the audio book the week after that, I think it'll do even better. I'm really happy with how it's going so far. Excellent. We encourage people to buy it and read it and tell us what they think. It was great to have you on back on Aman Virji. It was great to have you back on. Excellent. Thank you very much for hosting me. I really enjoyed it. Tell us what you think about this conversation with Aman. Do you think these companies should be valued at these valuations? And what do you think about companies go public a lot later? Make sure to tell us in the comments. I'm Siamak. This is Market Insider. We'll see you next time.