Global Macro: The Next Era of Market Growth ft. Ben Miller

28:25We saw Silicon Valley Bank collapse. We saw Silvergate. We saw a few other banks collapse. But really, it didn't become systemic. And that's because the big banks were so heavily regulated and it stopped them from basically having the kind of risk-on behavior that caused 2008 financial crisis. So I worry about that. But on the other hand, if you can go and increase leverage by... Right now, the leverage banks will give you today really low. Like if we have borrowing, we have fundraisers, billions of dollars of borrowings across our funds. And banks maybe will lend you 60%, 55%. And in 2007, when I was in real estate, banks went on 83%.

29:15And so there's a lot of room between 55 and 83. And so it's possible you see the banks start to creep up. And like all things, they start out wise and end up foolish. And so it's just a question of degree. So in the short term, if you can start to put more leverage in the system, That basically is the same as putting more money in the system, and that will drive more growth.

29:48What I worry about is that it just goes too far, especially if it looks like it's working in the short run. Hey, do you have a sense of what those thresholds are in terms of what too far means and how you measure that leverage in the system? it's really difficult to measure so much of it is now happening in private markets with private private credit and um and the formula is is changing so what will happen is the bank will will lend to a private credit uh institution the private credit institution will then lend to the to the borrower and so there's like um kind of like a different mechanism i think of actually a better one in the sense that private credit is less levered and less systemic at risk than banks because banks are insured by the federal government.

30:45But where does that leverage crop mean? Over 80%, that's in my experience, you start seeing broadly in the market loans that are available over 80%. That's usually a red flag. and typically it's somewhere in the 70s where you start to see if you can borrow at more than 75%, it starts to get risky. And in this market where interest rates are high, what's happening is that debt service ratios or debt yields are constraining the borrowing. And so the way you would have to sort of get more borrowing if you're constrained by debt yield is start to capitalize interest rates with PIC, with accruals. And so if you start seeing pro forma returns, because if the bank's lending on a business plan where there's virtual cash flows, there's growth in those cash flows, and that's actually what their asset is, and that growth doesn't show up, then obviously the loan goes bad.

31:47And that's what happened in 2006 and 2007. Because interest rates were about comparable to where they are today. But it was just that a lot of the lending was on presumed growth in NOIs, net incomes. Yeah, let me ask you this, and to touch on something that you touched on to the top of that segment, this idea of if you have banks essentially lending to private credit institutions, don't you wind up with an absence of transparency, the potential buildup of risk in a system in a way that is not readily apparent because it doesn't show up directly on the balance sheet in a way that you can examine those ratios or understand what the levels of risk are.

32:31I mean, is that a risk that you see potentially? You know, I mean, yes and no. I think that, I mean, again, I still think private credit is less risky than bank credit. But my experience with... So banks, right now, they're safe and they're levered like nine times, right? Because that's typically the bank ratio around nine times. At the peak of the financial crisis, Lehman Brothers levered 50 times, 60 times. And the amount of leverage that the banks can take on can go to essentially infinity. You saw that in the S &L crisis, the savings and loan crisis in the 80s. private credit really can't lever.

33:19It's a private company. There's no government backstop. And so typically hard to lever more than three times. You'll see it, but the banks basically won't lever, won't loan us to it. As I said, you see 60%, 75%. That 75 % is obviously three to one. So leverage. So leverage. So if you get to four to one, that's five times leverage. So that would be high. And the banks can go to 30. I mean, so it's just the amount of leverage possible in the banking system is just infeasible in the private part of the markets. All right, this is probably a good bridge to talk a little bit about what you're doing at Fundrise.

34:05I know we did some slides. And to tie in some of the broader macro conversations that we've just had, as it relates to the inputs into the formulas that you look at at Fundrise, I know real estate is something that you think a lot about. Take us through some of your slides, Ben. Well, I have this one analysis I wanted to share. It's about the software market. because one of the questions that we frequently get is, why don't tech companies go public? And so there's a lot of information on this slide, so let me show you how I walk you through it. So on the right is the enterprise value, easy enterprise value divided by the revenue multiple.

34:43Okay, so the more the revenue multiple, right, the more valuable the company. And then we look to the revenue multiples compared to profit margin or operating margin. So if you look at the bottom right, right, there's companies with the most amount of growth and the most amount of operating margin. So if you have a growth rate of more than 20 % and you have a profit margin of more than 15%, the market on average will value at 16.4 times your revenue.

35:16You see that? Is that tracking ash? Yeah, no, I do. And let me just walk folks through this and try and explain it to them. So what you're saying is essentially, it's basically just the multiple that markets assign to the valuation is simply a function in this chart. I know this is a simplification in terms of the model, is the rate at which you're growing revenue on the one hand and the operating margin, how profitable it is to do that business per unit of economic interaction. Right. And this is probably market. And so So just to sort of stay on the right side for the moment, right? You see a company that has less than 0 % operating margin and very low growth is only trading 3.6 times revenue.

36:05So that's sort of the opposite side of that or that catty corner to that chart, right? Yeah, that's the upper left-hand corner where you see it low, which is kind of interesting also. So if you have operating margin near 0 % and growth rate of near 0%, that still sounds like a very high multiple. 3.6 cents. Yeah. Well, the market's up a lot from 2022, right? 2022, when the market plummeted, that might have been 1%. So, but the point about this is that you're obviously, it tells you the market value is growth and the market value is operating margin. And the combination of two values the most. Now, on the left side of the chart is just the number of companies in each bucket.

36:47So you can see, right, there are only seven companies that have more than 20 % growth rate, software companies, more than 20 % growth rate, and more than a 15 % profit margin. Right? So not very many. Right. And so there's... And just to make sure that everybody's got the same chart. And there were four companies. There are four companies that have less than 0 % profit margin and really low growth, 0 % to 10%. So this tells you essentially, this gives you a sense of the software market, how things are priced. And there's a few things interesting about what this math is telling me. So one, and this is in here, but in 2021, there would have been dozens of companies that are growing more than 20%.

37:36I mean, dozens. Maybe half the software industry is growing more than 20 % in 2021. And now only seven are. Right? So very few companies are growing more than 20 % in the public markets. And I think there's no companies growing more than 35%. So most companies that are high growth are not public anymore. that's that's like that's with revenue growth that we're talking about right and in the private markets growth is valued much more than profits so what you're seeing here is that being profitable doubles your value approximately being profitable matters and And in the private markets, growth matters much more than being profitable.

38:34Obviously, if you're a startup and you only have 10 employees, no one's expecting it to be profitable. But even big companies, companies that have a billion dollars in revenue or 100 million dollars in revenue, if you're growing a lot, then your private market basically is not as worried about that. They're more worried about, can you take the market? Can you build an enduring product? So those are two really big differences, right? So there's very few growth companies in the public markets and that the public markets really want to see profitability from the companies that are public. And so if you think about the greatest companies that are currently private, like an OpenAI, like an Anthropic, maybe an Andrel, got on the list, those companies are focused on trying to build the future.

39:31and to try to demand profitability during that period of so much volatility, so much growth, so much change, it's really a mistake. But that's what the public market is demanding if any company goes public. And so the companies are not going public. Right. It's so interesting. Obviously, only so many dimensions you can show on a two-dimensional chart, but it is interesting because you bring up the two things that we were just talking about there, which is number one is the scale, the amount of revenue and earnings. And the second is life cycle in terms of where these companies are in terms of their life cycle.

40:10So it does get complicated quickly, but it is really interesting to look at these charts. Let me ask you this. Do you see that number when you look at that number there on the left-hand side, lower right-hand corner, seven? Is that about concentration or is that about an aggregate slowdown in the rate of growth. That's an aggregate slowdown in the rate of growth. No question. I mean, the rate of growth in the software industry went off a cliff in 2022 and really hasn't recovered. And it's kind of why I'm in real estate, I'm in tech. And both those industries are not seeing explosive growth. Now, AI is an exception to that.

40:53But if you look at the amount of SaaS companies, software as a service companies, there's thousands of them. I think there might be 14 ,000. I think there's maybe 1 ,000 that are worth more than a billion dollars. And across the SaaS industry, across the software industry, growth has just fallen probably in half or less. So that's been a big change in the software industry. It's actually arguably in a period of deflation, where if you go to re-up a contract, you might actually see the price go down and you see people shedding headcount, it's much harder to get an engineering job. So the tech industry outside AI is in a recession.

41:31Real estate industry is in a recession. And that is why growth rates have fallen so much for software industry. It's so interesting. And the conversation around AI, I mean, I'll just try and spin this thesis out there and maybe you can tell me where I'm wrong, but it seems like one of the challenges is we have this very sort of optimistic belief about what Silicon Valley can do, the idea that any small group of people can spin up a company and potentially create the next unicorn or decacorn. One of the challenges with AI, where obviously all of the attention is right now, is that you need these enormous data sets.

42:12You need enormous amounts of processing power. You need to get very sophisticated data scientists to do it. It almost seems as though that's doing a couple of things. It's raising the barriers of entry to be in those spaces, number one. And number two, it's creating a more concentrated market for this type of development. So effectively, if all the excitement is going to the notion that machines are going to be able to do the development and do all of the deep insights, does that shift the paradigm away from this very optimistic view that we have of Silicon Valley and the idea that we've seen over the last 20 years for SaaS business growth?

42:55I don't think so. I think it's actually following the pattern that existed in the past. If you compare it to the internet, well, the internet required something like half a trillion dollars of fiber put in the ground. So it needed a ton. There's no internet without fiber, right? You could have a dial-up with the AOL across your telephone line. It was 14, 4K, right? It was so small. So just like internet, you need massive infrastructure in order to connect everybody. And then look at the platforms. I mean, Google completely... And Google is like Google, Microsoft, Apple. Those are the primary platforms that control these big...

43:37Whether internet or mobile. And it's super heavily concentrated. And then on top of those platforms are the applications. applications might be Salesforce, might be Airbnb, might be Uber. And so I think what you're seeing is very similar. You're going to have a massive amount of infrastructure needs to be put on the ground, which is going to be data centers, compute, electricity, data, I mean, power. There's going to be a handful of platforms like OpenAI, maybe Anthropic, Google Gemini. And then there's going to be applications. and those applications really don't exist yet. I mean, they will.

44:17I mean, we're building one. Lots of people are building them. They probably... You have some small ones, but mostly they're still coming. Just like if you remember in 2011, 2010, the App Store had like... There's basically no apps that was popular in 2010 that ended up being a dominant company. You know, most of the great apps would have invented later. So it takes a while for it to all mature. And those applications are being funded and built in the private markets. So I think it's actually totally true to form and that people just can't remember how long the lags are. Right. It's almost like long and variable lags.

45:04But for software development. Yeah, it's a funny metaphor. But let me ask you this. Is it different in the sense that when you talk about the fiber infrastructure in the ground that was necessary for the internet backbone to be built and the ability to build these SaaS platforms on top of it? It's interesting because that obviously was being done through government investment. This is something that you see obviously happening in the private sector. So I guess the question is, are there going to be application layers that are going to be built on top of the AI? And will there then be this potential or opportunity to facilitate growth for new companies that use this sort of softer infrastructure in AI to develop the next generation of applications and therefore the next generation of great companies?

45:51Yeah, I mean, that's what I think is happening. that that's what's actually good. If you go back to the mainframe and then the PC, whether internet, mobile, cloud, I mean, the companies that become the platforms, they want to encourage applications built on top of them. The Android, iPhone war, right? You want the application built on top of your platform. Ultimately, your customer is the application, right? So like, for example, Fundrise, we're a customer of OpenAI. We pay them. We have access to their API. And if we build an application, an AI application, or whatever, Anthropic. Anthropic is probably a better example.

46:36Anthropic is mostly an enterprise product. And Anthropic wants us to build on Anthropic, not on OpenAI. And so they do all sorts of things that encourage that type of sort of competition for customers. and and I mean only in the last few months right have we seen sort of like the models get good enough where you can really build applications that make sense I mean the the the breakthrough with 01 or with inference which is you know really really recent that it has made the products you know much much better it's just there were there were serious limitations to what the model could do to solve problems.

47:22But the types of problems that it's going to be able to do most accounting. I mean, Fundrise uses it for customer service. It resolves half our customer service outreach. We get, I think, 30 ,000 customer service tickets a month. And we work with an AI product called Intercom. And so it's happening and it's going to be revolutionary. It's just, unfortunately, it takes a long time to build because it's like building a building, right? It takes two years to build. But when you're done, that building is useful for 100 years. So I think that the kind of product development is happening with AI. I mean, I know it is.

48:05We're invested in some of the companies. We're building with it. But what's the breakout product going to be that everyone wakes up to overnight and starts using? It's like Uber, right? We don't know yet, but it's imminent. Yeah, Uber is an interesting metaphor because it would have been very difficult to predict. And obviously, you needed all of those APIs. You needed all of those services in terms of GPS payment processing, the ability to rapidly do data management. I mean, a whole suite of different integrated applications that went into that, the infrastructure to do it. Let me ask you this, particularly because you're doing this today.

48:42Talk about what some of those application layer development projects that you're working on there on top of AI, because I think it's hard for people sometimes to get their head around what that application development looks like on top of the AI infrastructure. Yeah, I mean, it's hard because you're getting into the nitty gritty. It's like, what does it take to put plumbing in through a building? You have to stack and pack the toilets and stuff. Like it's, you know, like open up the walls and talk about, you know, how you actually build these things. Well, let me take it a little more generally than that.

49:16Just give us a sense of what the opportunity set looks like, what the end user functionality would be, what the promise would be. I don't think we have to go into the nitty gritty about how it gets built out. But when you look at this, you go, okay, so I go on to ChatGPT and I spend an inordinate amount of time having these conversations with ChatGPT. And, you know, I ask of things and do research and that sort of thing. But how does that translate into application development and what some of the opportunities could be in terms of the value that it would provide to end users? Yeah, I mean, there's probably like, I'll say three main challenges, but I'm sure there's more.

49:49The three that come to mind immediately for me are most data that you want it to access, it's not in the model. So if I'm a small business and I have all this accounting I'm doing on my AP, my account's payable, my account's receivable, and I have all these processes I wanted to do, it needs to be able to access all that stuff. Today, it really can't, right? You sort of see it can surf the internet, but it can't access your data. People, big companies, especially are worried about accessing the data. So there's this whole challenge of how do you get it to have access to it? There's a sort of, if you flip it, the other side is it can't go out into the world and do anything, right?

50:30It can't buy you shoes. It can't shop. It can't book travel tickets, right? So there's just like, this access is two-side access is a challenge that hasn't really been solved yet. So you can have an agent, but if it doesn't really have the ability to read the data or write the data, right? write a transaction, it's really hard for it to be effective in the world. And the third thing is that you can have the ability to read or write, but it actually then also has the ability to understand. And most knowledge that we want to understand is specialized, like if you took accounting or real estate, any kind of really narrow vertical, they call it, or field.

51:15And so there's all this specialized knowledge that's not in the general model. And so you have to then take a team of people and, and either teach it, train, teach it, train it, or build these are like almost like minimum, minimum roads or like guardrails and say, okay, here's how you do this one thing. Here's how you balance a trial balance, right? Here's how you reconcile, reconcile some, you know, when you're, you know, you're so much of, I'm picking accounting because I think people can imagine, you know, you have, you need to reconcile your bank statement. with your accounting statements.

51:51There's a sort of, you wreck it out, right? So you have to, and so there's like, you have to read the bank statements and then you got to go put those banks over there. Well, who's going to let AI read their bank statements? So there's all this sort of challenge around reading information and writing information and then making sure it does it correctly in a specialized area that the model doesn't have like specialized knowledge. It has just general knowledge. And so that's the kind of stuff that people are having to do in probably thousands of little sectors, thousands of little problems like the apps.

52:24I think there's 2 ,000 apps on the iPhone.

52:31And they do all sorts of things, right? Teaching you Duolingo, teaching you language. All these things are happening today. And it is challenging because, you know, who's going to let an AI write to the database? Like most engineers would try to strangle me if I try to let AI write to our database. And not to mention the CFOs who would say, how do we know that this is right? And today the people are happy to pay individuals to do that because they have the accuracy and accountability. And one of the other challenges about AI is that you get the input and you get the output, but the process that happens in the middle is a total black box, right?

53:10You don't really know how it comes to those conclusions. It just does. And that's something that as a CFO of a company, I wouldn't certainly be comfortable with, okay, hey, here's all of our bank statements. Here's all of our accounting data. Reconcile it. Yeah. Yeah. And don't publish it to the internet. Right. Right. Yeah. And we're, as you say, guardrails for knowing what that processing looks like and how you can make sure that this is all happening in a secure enclave of understanding what the input and output is and where the potential for leakage is for data. Yeah. Or being hacked, right?

53:45So that's all happening. All those things, all those little nitty-gritty complexities are being... I'm in the process of doing it. And it's not very glamorous, but it's definitely underway. And so far, it's been small things. Our IT now, we use a software called Risotto to do our IT work. So people say, oh, I can't get my Adobe to work and my Adobe's not working. Deal with an AI bot rather than dealing with one of our IT professionals unless the problem has to be escalated. This is an agent, a chat that essentially informs the user, okay, have you checked this? Have you gone and checked to see if these files are there?

54:32Have you gone and checked this aspect of settings? Have you checked these registry keys if you're in Windows? Right, exactly. IT is a lot like customer service. The same with HR. And this company, we test drove a bunch of them. We picked this going with Risotto. And then I'm trying to get to invest in the company because it's great. And then as you imagine, as it does IT for more and more companies, it gets better and better at IT. But it still can only, you know, it's stuck. Its medium is a person, right? It can interact with people, but it can't really interact with our systems yet. Our IT team would never give it access to our systems because it's worried what it might do.

55:15In other words, autonomously, hey, just go into the system, go to our network, fix it. Yeah, go into my computer and turn it off and reset the settings or something. That's something people would be terrified to do. So that's the... How do you build those types of guardrails? So there's all these things like that happening and at the same time, the models get better and better. And so, yeah, it's a productivity revolution. And it's really, I mean, it's like at this point, if you're in the work, it's like when I build a multifamily building, I'll look at it and it'll be a piece of dirt, but I'll have the permanent hand.

55:56I'm like, that building's done. Right? Okay. Two years before it's actually opened, but essentially it's fully funded as permitted and now it's just like a construction project. But it's like, at that point, it might as well be done from a net present value point of view, right? From a financial forecast. And so maybe there's some budget busts, but it's like, that's how people who are in the industry look at it. They see a piece of dirt and most people see a piece of dirt and I see a building. But that's where AI is. And, you know, yeah, 24 months at most It's where everybody will see the buildings that AI will have built.

56:39Ben, you are an extremely interesting guy. I feel like you're the only person I can talk to about building from plumbing to cap rates and then talk about agentic AI, the opportunities that the risks and the guardrails. It's just always interesting to have you on the show because you've such a diverse way of looking at the world. And so that drives your metaphors and the way you see and think about this. yeah well we sit at the intersection of real estate and tech and and i think to be successful either you're a specialist or you sit at a nexus and and so where we sit at a nexus and and uh i think application of ai real estate is going to be revolutionary um and and um you know it's like it's gonna it's but you know what's it gonna do to cap rates that's what my real estate team will say, okay, but it's like, is AI going to drive cap rates up?

57:29Is it inflationary or deflationary? Because it's like, there's this great guy who wrote this thing called Situational Awareness, and he predicted that AI would drive real interest rates up because it would demand so much capital. And no one would put any money into anything but AI because the returns are so high. And so drive interest rates up. Typically, technology drives interest rates down because is deflationary. So anyways, there are big open questions here that are exciting to try to speculate about. Yeah, it's funny when you say you can either be an expert or sit in an axis. It's always more interesting and more fun for people who sit in an axis, right?

58:09I mean, it's just like there's this weird cross-fertilization of ideas that just makes it such an interesting conversation, as it always is when you join us right here at Real Vision. I think we're coming up on the hour here, but Ben, let me ask you this. We've a wide-ranging conversation, everything from policy, macro, AI, and real estate. Final thoughts, key takeaways that you'd like to leave our listeners and our viewers with. Yeah. I mean, the only other thing that I think I've thought a lot about is how real estate has been beaten down over the last couple years because the interest rates went up and stock market has gone up.

58:42And typically, the interest rates in real, typically real estate and stocks actually move together because both are economic assets and they've moved now. Real estate's really, really cheap and stocked really, really expensive. And so that may continue, but it's really unusual for that to be the case. It hasn't been the case in 15 years, maybe longer. So that's something I plan to talk about because it's such a strange situation, but it's, I guess, for another time. Well, come back and join us again. That would be a great conversation to have. Yeah, maybe by that point, something's changed. Indeed.

59:19Ben Miller, thank you so much for joining us. Yeah, thanks for having me. Thanks for listening. Thanks for watching. See you in Miami.

1:00:16Transcription by CastingWords

1:00:20If you liked this episode, I'd love for you to head over to realvision.com forward slash join for a free membership. Start your journey today to unfuck your future. Just one click away. Have you ever wanted to trade Bitcoin but haven't dared try? With Plus500 Futures, you can trade crypto without the hassle of opening a wallet. With just a few clicks, you can register and start practicing with their free and unlimited demo. See a trading opportunity? You'll be able to trade it in just two clicks. Feel ready? You can move to real money with as little as$100 once your account is approved. And the great thing is that in addition to crypto, Plus500 gives you access to a wide range of instruments.

1:01:00S &P 500, NASDAQ, gas, and much more. Explore equity indices, energy, metals, Forex, and beyond. With a simple and intuitive platform, you can trade anytime, anywhere. Experience the fast, accessible futures trading you've been waiting for with Plus500. With over 20 years of experience, Plus500 is your gateway to the markets. Visit us.plus500.com to learn more. Trading in futures involves the risk of loss and is not suitable for everyone. Not all applicants will qualify. Plus500, it's trading with a plus.