TIP823: From Railroads to AI: The Timeless Patterns Behind Market Bubbles w/ Kyle Grieve

23:26Preston Pysh:But as the bubble grew, investors started wondering just where the return was, and the businesses weren't really showing that they were able to validate the marketing from their pamphlets. The roads ended up breaking early and required very costly work and supplies just to maintain. A great case study of the plank road investment comes from an investor named John Taylor. So he invested$900 in three local plank companies in the Albany, New York area. Over a 12-year period, his combined dividends totaled less than$80. This implies a dividend yield of only 0.7 % per year, a far cry from that 10 % to 40 % that were being promised by the promoters.

24:01Preston Pysh:One$250 investment that he had into Albany and Renssel Aerville Plank Company was sold for only$25. And another of his investments just never paid a dividend simply because they had to spend too much money on repairs. Taylor writes that by 1865, the majority of plank roads were either abandoned or connected to turnpikes. This example shows the dangers of combining hype, oversupply, and poor economics. It's also a case study showing that these three ingredients just tend to completely wipe out investors. And it's a solid example of Kindleberger's framework. The first one being displacement. So the reduced prices of transportation was obviously seen as this kind of new revolution for commerce.

24:39Preston Pysh:And this resulted in thousands of miles of plank roads being proposed, funded, and built. And then you look at over trading. As the bubble built, more businesses traded for higher and higher evaluations, even though the business plan didn't really support any value creation. Third is monetary expansion. As the plank roads were built, they received funding from investors to continue building more. Fourth is revulsion. As people like John Taylor showed, they eventually just left the idea simply because it wasn't really offering any real returns, just a broken promise. And five is discredit. These businesses were rightfully, I think, discredited because they just completely eviscerated capital, which is why they were eventually abandoned.

25:17Preston Pysh:Now, I think it's obvious that hype cycles like this are being seen today. Quantum computing is another area today that I think is completely ripe for a bubble. Rigetti computing at one time in 2025 was up nearly 12x in under a year. And what fundamental results from the company precipitated this massive increase? My only answer? Hype. Since 2022, revenue and EPS have decreased while margins continue to decline. The only answer for a 12x in its share price is that investors are relying on hope and the greater fool theory just to make their return. Just like the plank roads of the 1850s, investors are putting money into a capital destroyer simply because it tells an interesting story.

25:55Preston Pysh:Now, the plank road bubble, I think is a pretty good example of what Ron calls tiny bubbles. These are bubbles that will only affect a pretty small number of the population and won't cause Jerome Powell to lose any sleep. In other words, tiny bubbles aren't system-wide bubbles. A$200 million micro cap business trading at maybe 300 times revenue might harm a few investors once that bubble pops, but it's not going to cause panic in the global financial system. But tiny bubbles still obviously matter a lot to investors because while they may not ruin the global financial system, they can easily ruin your portfolio.

26:30Preston Pysh:If a single investor has exposure to a business that is part of a tiny bubble, when it pops, the market may not notice at all, but I guarantee you that the investor will. So what are some good examples of these mini bubbles? Insana goes over things such as closed-end funds, which I think are a great example. So first thing first, what is a closed-end fund? It's basically a mutual fund that purchases a basket of stocks representative of the stocks in a given market. So he goes over European and emerging market closed-end funds. It's important to recognize that closed-end funds are not ETFs. Therefore, the prices of closed-end funds can often be completely disassociated from the net asset value of the asset.

27:08Preston Pysh:So what happened in Europe to create a bubble in closed-end funds? Countries like Germany, Austria, Spain, and Italy all saw very large spikes in the prices of their closed-end funds peaking in 1989, which coincided with the fall of the Berlin Wall, which speculators thought would maybe create some sort of peacetime dividend. And these funds attracted many foreign investors who wanted exposure to a country's corporate world. The problem was that many of these funds were trading at over 150 % of their net asset value. So if you simply had just bought the stocks on the open market, you would have received a massive discount compared to that country's closed-end fund.

27:43Preston Pysh:But the closed-end fund allowed investors to just have exposure to these markets without really having to do any due diligence on specific companies. Now, when the bubble popped in the early 90s, many of these closed-end funds dropped significantly in price. German closed-end funds declined by 75%, Austria 80%, Spain 60%, and Italy 65%. Now, another example from the book was the Cuba Fund. This was a closed-end fund in the US meant to give American investors exposure to Cuba once it reopened for business to Americans. Since Cuba was illegal for Americans to invest in, the fund came up with very, very creative ways to get exposure.

28:18Preston Pysh:These included things such as investing in Florida-based and Caribbean-based companies that maybe had a chance of benefiting from doing business in Cuba if it was eventually open for business. And apparently a lot of the capital in the Cuba fund was actually just in T-bills, which were accumulating interest in anticipation to invest at a later time. So tiny bubbles don't even actually have to happen in public markets. Insana points out that collectibles are another area where tiny bubbles have formed in the past. Beanie Baby dolls are a great example of this. So Beanie Babies had some of the best marketing gimmicks that I've ever seen.

28:50Preston Pysh:And the mastermind behind the marketing strategy was this gentleman named Ty Warner. So he created scarcity by retiring certain models of these Beanie Babies each year, effectively decreasing the supply, which created this supply and demand imbalance. To help drive interest in Beanie Babies, he'd do kind of these product tie-ins with businesses such as McDonald's or have these giveaway nights at professional sporting events. And one of his biggest schemes was to announce that Beanie Baby production would completely cease at the end of the 20th century, which further increased the buying pressure.

29:20Preston Pysh:The book outlines that the price of Beanie Babies went up a thousand times its face value at its apex and have now stabilized for decades. But I'm actually not sure that the bubble ever fully popped. A quick search on eBay shows a number of Beanie Babies still selling for over$10 ,000. For instance, the Princess Diana Beanie baby is listed for$19 ,500 today. The key lesson here regarding big bubbles or tiny bubbles is to think about whether you are truly investing, which is based on capturing a gap between price and value, or if you're just speculating and focusing purely on price. Buying a doll for$20 ,000 seems pretty obvious that you're probably just focusing on price.

29:59Preston Pysh:You want to make sure that you are focusing on the fundamentals and not necessarily purely on the narratives that people are telling you. Just like stocks, any asset can rise and quickly collapse in price. So make sure you're allowing Benjamin Graham to whisper interior about the importance of investing versus speculation. Unfortunately, tiny bubbles are an excellent trap for retail investors. And that's because participants feel early, they feel highly intelligent. And that unfortunately creates this velocity in price increase that only reinforces their illusion. But if you partake in too many tiny bubbles, you can easily end your career as an investor.

30:35Preston Pysh:Now, as a concentrated investor myself, I'm fine having positions that go up and make up a larger and larger percentage of my total portfolio. But it's also essential to have some sort of diversification embedded in the portfolio. Don't be the guy who loses a bunch of capital, then goes all in on one asset, hoping to make up your loss. That's a great way to just destroy wealth. So even if you are concentrated and focus on the fundamentals, make sure that you have a few positions to spread your risk into. And it's always important to remember that there's no such thing as a risk-free investment.

31:05Preston Pysh:The next bubble is arguably the largest in modern history, and that's the tech bubble of the 1990s. And what stands out to me most about this bubble is the massive participation of the American public. This bubble wasn't just professional investors bidding up railways. It was your average household pouring money into the market via 401ks, mutual funds, and just day trading. Another good indicator was to look at bank deposits. When bank deposits tend to be high, it generally tells you that the market is in no rush to spend or invest money. But during the tech bubble, bank deposits fell to 50-year lows, all while stocks grew to 58 % of household financial holdings.

31:45Preston Pysh:The widespread participation injected a massive amount of liquidity into tech stocks and made the eventual collapse of the market all the more painful for everyday people. Now, I don't really believe the narrative of the tech bubble was much different from other past manias. Whether you look at things like railroads, automotive industries, cars, turnpikes, canals, or even Beanie Babies, where the big difference was in this mania was just how early the market was. The tech bubble was built around the narrative that the internet would change the world. And it did. But that value creation unfolded over many decades and not over quarters.

32:20Preston Pysh:But the market priced those expectations as if they were just around the corner. Now, the extreme inclines in valuation do a very good job of underlying the disconnect between price and value. So in March of 2001, before the bubble burst, the Nasdaq was trading at 246 times earnings compared to a historical range that was close to only 40 times. Now, a six times premium to historical averages isn't just a minor stretch. It's basically impossible to justify a premium that high for an entire index. And here's the thing, there's nothing wrong with a company getting a multiple re-rating once the market verifies that the business is improving, but it's unheard of for a market to improve at that rate across a very broad number of companies.

33:01Preston Pysh:Then when you consider that many companies in the index just didn't even have working products, you begin to see a picture where the multiples were going up while the average quality of the businesses inside of that index were actually declining. So in 1999, economist Robert Samuelson concluded that 77 % of IPOs had no profits. The speculative energy in the market just was inflating valuations across the entire NASDAQ. But there are other forces at play here as well. The Fed reduced interest rates to help troubled banks recapitalize themselves and pull themselves out of an economic slump. And this created reduced borrowing costs, which are obviously key ingredients to a bubble.

33:38Preston Pysh:Energy costs had also come down, freeing up money to be spent elsewhere, such as gambling and stocks. Wall Street had its part in the euphoria as well. Since many of these dot-com businesses couldn't be evaluated using traditional value-creating metrics such as free cash flow or profits because they just didn't have any. So investors on Wall Street created new metrics. Instead of analyzing the future cash flow of these businesses, they looked at things such as page views and unique visitors. And there wasn't any shortage of new opportunities for investors to look at either. Another key sign that markets are in bull mode is to look at IPOs.

34:11Preston Pysh:So IPOs from 1980 to 1989 average about 311 IPOs per year. Now from 1990 to 2001, that number jumped to 380 per year. So after the tech bubble exploded, the average dropped closer to 200. Now one of Monish Pruvai's biggest multi-baggers was explained as a case study in this book. And this is CMGI. Where Insana saw a failed internet incubator, after the fact, Monish saw an opportunity before the fact. Now I'm not sure how much of a multi-bagger CMGI was for Monish. But what the business was, was basically an internet holding company. So it would basically supply internet companies with capital and then take them public.

34:49Preston Pysh:It was essentially just a public venture capital company. Monish wrote this idea at the exact right time and got out before it cratered from around$140 to just$5 in early 2001. Now the problem with CMGI was that it was just not a sustainable business model. It required a bubble in internet stocks to function. And any business that requires an eternal bull market is just not going to last very long. Now, while Insana liked Kindleberger's framework about bubbles, he actually had his own spin on it, which I think is actually an improvement. If an investor has this framework, they could theoretically speculate on a bubble and try to get out before everyone else does.

35:25Preston Pysh:I think that this is maybe what Monish was trying to do with the CMGI bet. Now, to help you better understand how to deal with bubbles, let's go over Insana's updated version of Kindleberger's framework. So Insana's framework differs from Kindleberger's as it focuses much more on the rise of the bubbles, whereas Kindleberger's framework focuses on both the rise and the fall. So here are the five ingredients of Insana's bubble. The first one's called the Eureka moment, which is when the world makes an exciting discovery or invention. Two is easy money. This is when there's a low cost of money and high availability of cash and credit.

35:58Preston Pysh:Third is government largesse. This is where there's favorable economic conditions or maybe tax incentives that are put in place by the government. Fourth is auspicious economic conditions. And fifth is an external stimulant. Now, we've already covered many discoveries and inventions such as plank roads, but you can also include other things, of course, such as railroads, cars, the internet, or even mortgage-backed securities. Next is easy money. So when there's cheap credit, corporations can fund growth at a cheaper cost. This means they're more likely to borrow and spend even more. When a business can spend more on investments, there's a chance that they can create additional shareholder value, which of course, shareholders love.

36:37Preston Pysh:When investors show a disposition to deploy their savings, more money comes into the market, which can further improve deal making. So if a company has a high valuation, it can then use its stock as currency to make additional deals or to invest in itself. Now, government largesse includes things such as tax incentives, subsidies, and new policies that can supercharge an industry. For instance, during the railroad boom, there were these land grants that were granted to railroad builders. During the internet boom, there was a tax moratorium. So these seemingly small events can have very outsized effects down the road when other entrepreneurs come in and try to take advantage of them.

37:13Preston Pysh:Next are auspicious economic conditions. Look, you know, bubbles don't happen when a country is in a recession. They happen when things are going well, when GDP growth is strong, when unemployment levels are low, and when consumers are optimistic and willing to open their wallets and purse strings. If unemployment levels are high, chances are a bubble isn't going to happen simply because money isn't plentiful. And the last part here is about external stimulants. This can include macro factors such as war, regulatory changes, demographic shifts, or even crisis. These stimulants can accelerate demand and attention.

37:46Preston Pysh:What I like to look for are what I call micro stimulants. So in one of my investments that didn't quite work out for me, called BQE Water, the thesis centered on their selenium product, which could reduce selenium levels in effluent water. So the stimulant for this was simply that North America would no longer allow selenium to be deposited into waterways at high levels. And BQE had an excellent product that would directly help with this issue, and it was backed by patents. I like these kinds of stimulants. Now, just like Howard Marks, who I'm going to discuss in a little more detail here shortly, Insana believes that we can detect bubble-like events through pattern recognition.

38:22Preston Pysh:His pattern recognition framework is what we just covered. It's important here to know that we can never predict things with 100 % certainty. The best hope we can have is to recognize the recurring conditions in which bubbles form. So the first three ingredients necessary for an asset bubble to form are technological innovation, easy money, and government largesse. To identify the things, you just simply have to observe the world, you know, read newspapers, look at industries and individual companies, analyze economic indicators, and maybe consult with experts. Insana says that if you can identify those three ingredients, you can be ahead of the curve in finding high potential investments.

38:58Preston Pysh:But if you notice an industry with these three ingredients that are truly creating value, how do you make sure you protect yourself once the bubble pops? While I have no desire to participate in bubbles, I'd actually just prefer that my business's public market valuation simply increase at the same rate as their intrinsic value. But in reality, that's just not what happens. Stocks go up and down with nearly 50 % variance on average each year. And sometimes they can obviously go up much, much higher than that. So with that in mind, investors must always watch to make sure an investment isn't in bubble territory.

39:30Preston Pysh:What does Insana say to look out for when a bubble is getting into this terminal phase? So he notes a few key indicators to watch for. Money turns tight. Fiscal policies that once encouraged investments are repealed, innovations, inventions, and discoveries fail to deliver on their initial promise, underlying economic conditions deteriorate, and public participation peaks, which is a signal that the bubble is about to pop. Now, the easiest way for money to go from easy to tight is just when interest rates rise. Rising interest rates means higher borrowing costs and less corporate profits. So that's an easy signal that any investor can really look for.

40:06Preston Pysh:But it's also important not to get scared out of a position that's maybe not in bubble-like territory just because you see rising interest rates. So while I think rising interest rates weren't a cursory glance, they need to be considered alongside other signals to be valid. Next is fiscal policies. So Insana argues that fiscal policy that was released before Black Monday in 1987 was actually one of the reasons for that crash. The story here is that much of the boom leading up to Black Monday was actually fueled by the leveraged buyout boom. And these LBOs were funded by something called junk bonds.

40:41Preston Pysh:A bill was then introduced that actually repealed the junk bond deductibility, which sent buyout firms down. Now, when we analyze the failure of delivery of the benefits of new technologies, we don't have to look much further back than the example which I already gave, which was the tech bubble. So many investors were drawn to the potential profits that were promised in the narratives of these businesses. But once it became completely apparent that these narratives were more dreams than reality, investors began exiting in droves. Now, the breakdown of economic conditions can be seen in a more recent event, such as the Great Financial Crisis.

41:15Preston Pysh:During that time, credit froze, asset prices declined, the housing market fell, global GDP declined, and US unemployment rose to 10%. The bubble in mortgage-backed securities was no longer sustainable, so it eventually burst. During the great financial crisis, there was no obvious bubbles that formed during that time, simply because economic conditions were not set up to create one. Finally, we reached the final signal, which is public participation. One of my favorite anecdotes on this came from Peter Lynch. He said he was surrounded by people at a dinner party. In stage four, once again, they're crowded around me.

41:50But this time, it's to tell me what socks I should buy.

41:53Preston Pysh:Even the dentist has three or four tips. And the next few days, I look up his recommendations in the newspaper, and they have all gone up. When the neighbors tell me what to buy, and that I wish I'd taken their advice, it's a sure sign that the market has reached the top and is due for a tumble. There isn't much more to add to this. Now, I do think that all these signals are important to look for. But just because a segment of the market is in a bubble does not mean that the entire market is. When the market is expensive, I like to look at my own portfolio and see which of my assets have risen the most in price.

42:24Preston Pysh:Then I compare it to the increases in the intrinsic value of those businesses. I have a few positions, for instance, that have doubled in 2025, but does that mean that they're in a bubble? I don't think so. I think it just means that they're probably a little bit overpriced. Now, in this case, some investors might take profits and some, like me, just do nothing. When you have an asset that wasn't cheap to begin with and maybe went up five times in price and has only increased its intrinsic value by, let's say 10%, then you need to get very, very worried. One experience I had with this exact kind of scenario was a business called InMode, which I no longer own.

42:57Preston Pysh:So this business makes minimally invasive aesthetic devices. Now, from the depths of COVID when I initially bought it until just late 2021, less than two years later, the company was doing exceptionally well. It doubled its EPS in that time. Let's take a quick break and hear from today's sponsors. Curious about online trading, but haven't taken the first step yet. You're not alone. And Plus 500 Futures is a great place to start. The futures markets are moving fast. And with Plus 500, you can explore popular assets like oil, gold, S &P 500, Bitcoin, and more. From crypto to commodities, there's always something happening.

43:33The platform is super easy to use. So you can trade on the go right from your phone. You can get started with just$100 and jump into the action. See something interesting? Once your account is open, you can trade it in just a couple of clicks. And if you're not quite ready yet, you can practice with a free demo account. No risk, no pressure. With 20 years of experience, Plus 500 makes trading more accessible than ever. Check it out at plus500.com. Trading and futures involves risks of loss and is not suitable for everyone. Not all applicants will qualify. Plus 500. It's trading with a plus. Every business is asking the same question.

44:10How do we make AI work for us? Sitting on the sidelines is of course not an option. Your competitors are already making their move. But with NetSuite by Oracle, you can put AI to work today. NetSuite is the number one AI cloud ERP trusted by over 43 ,000 businesses. It unifies your financials, inventory, commerce, HR, and CRM into a single source of truth. And that connected data is what makes the AI smarter. It doesn't guess, it knows. Automating routine tasks, surfacing actionable insights, and helping you cut costs and make fast, confident decisions. From software and IT services to healthcare, equipment manufacturing, financial services, and many other great American industries, NetSuite delivers a customized solution for your business.

44:51This is not a bolted on tool. It's AI built into the system that runs your business. And if I hadn't needed this product, it is exactly what I'd use. If your revenues are at least in the seven figures, get their free business guide, Demystifying AI at netsuite.com slash TIP. The guide is free to you at netsuite.com slash TIP. That's netsuite.com slash TIP. Before I joined the Investors Podcast, every what if you can imagine was running through my head. What if I'm not the right fit for this audience? What if I freeze up on camera? What if I can't keep up with the level of analysis people here expect?

45:26Well, betting on myself anyways turned out to be one of the best decisions I've ever made. And I'll tell you, those early days would have been a lot less stressful with a partner in my corner the way Shopify is, the businesses that work with them. Shopify is the e-commerce platform behind millions of businesses around the world and 10 % of all e-commerce in the US, from household names like Allbirds and Gymshark to brands just getting off the ground. You get started with their design studio, hundreds of ready-to-use templates that help you build a beautiful online store that actually matches your brand with no design background required.

46:03It's also packed with AI tools that handle the work you didn't realize you'd have to do, writing product descriptions, headlines, even enhancing your product photography. And what if you get stuck? Well, Shopify is around 24-7 with award-winning customer support to walk you through it. It's time to turn those what-ifs into with Shopify today. Sign up for your $1 per month trial at shopify.com slash T-I-P. Go to shopify.com slash T-I-P. That's shopify.com slash T-I-P.

46:36Preston Pysh:All right, back to the show. So you'd expect the stock price to at least double during the time, but in reality, it actually went up 7x, which looking back now should have been a very good signal that this stock was probably in bubble territory. Now, even today, five years later, the stock price is only a fraction of what it was at that all-time high. So the returns, if I had held through the entire bubble would have been very, very unattractive. I unfortunately carried it through and didn't sell anywhere close to the top of the bubble, but luckily I still made a good return simply because I got out before the business really started to stagnate.

47:10Preston Pysh:Now, let's continue with the point here that tiny bubbles can happen, but sometimes they are precursors to maybe some hidden dangers. And that danger has to do with fraud. The unfortunate part about bubbles is that fraud can flourish when oversight tends to maybe be a little more lax and when investors' euphoria is intense. So the late 90s were a time of rapid growth and renewed optimism based on tech. These rising markets helped weaken skepticism among key parties, including auditors, analysts, banks, and investors. At this time, Enron was just thriving because the increasing complexity of the business was mistaken for intelligence.

47:45Preston Pysh:And since the stock price continued to rise, investors just threw caution to the wind. Now, one of the problems with Enron, according to investor perception, was that its complexity was seen as a feature and not a bug. Enron positioned itself as this innovator in energy trading and broadband. And even though the business was really peddling commodities, it created a narrative based around how it was transforming itself into a tech company. Now, during this time, one of its key markets was deregulated. California recently deregulated its market, allowing Enron to sell power and trade electricity in one of the largest markets in the US.

48:20Preston Pysh:Now, unbeknownst to regulars at this time, Enron was just manipulating energy prices in California, which ended up costing the state about$11 billion. But the regulators didn't even catch on to this until after Enron went bankrupt. Now, the issue here with Enron was that they were really, really good at hiding its fraud. They hid it from regulators and auditors. So the average investor just didn't have a chance. So what exactly was Enron doing? They did things such as booking imagined profits as current income. They used off-balance sheet debt to make their balance sheet look healthier than it was in reality.

48:52Preston Pysh:And then they had off-balance sheet partnerships, which Enron Insider secretly ran. And they counted trading transactions within their own company, and they allowed those fees on those transactions to count as their own revenue. So in a typical environment when everyone wasn't getting rich, this may have raised questions. But in the increasing bubble at that time, these issues were just kind of swept under the rug. So my big lesson here is to just not get complacent in bull markets. Fraud happens all the time in public markets. The key is to remember to stay vigilant. It's when emotions are running high and your intuition tells you to just be lax is probably the time when you should actually be the most vigilant.

49:29Preston Pysh:Now, the final chapter of trend watching deals with potential bubbles in the future. The book was published in 2002 before the great financial crisis. But interestingly, Ron directly discussed real estate as a possible bubble. So he created a checklist with his own framework. External shock? Yes. Monetary stimulus? Yes. Fiscal stimulus? Yes. Economic conditions? Favorable. Speculative public participation? not quite yet. So he was already starting to see some of the writing on the wall that perhaps inside of real estate, a bubble was beginning to form. And that final part, speculative public participation needed just a few more years to fester before things ended up blowing up.

50:09Preston Pysh:So according to Y charts, US housing homes sales didn't peak until around 2006 at 7 million homes sold. That was up from just 5 million in 2000. And after the bubble popped, sold homes normalized back into about a 4 million range. So I would say public participation indeed reached a very speculative fervor, but not until 2006. And even after that data was public, it still had a lag time before fear gripped the market as a great financial crisis didn't actually start until mid 2007. Now, if I'm being completely honest, I don't think about bubbles much using Ron's specific framework simply because I subscribe to Peter Lynch's notion that if you spend more than 13 minutes analyzing economic and market forecasts, you wasted 10 minutes.

50:53Preston Pysh:So I tend not to spend too much time on this at a macroeconomic level, because I think my time is much, much better spent looking at the impacts on each of my individual businesses. I have a grocer in Poland, and I have a trust builder in Canada, and they're just not going to have much macroeconomic concerns in common. Instead, I look at what concerns in each of those countries can contribute to the businesses booming or declining in the future. Now, I think it's definitely worth using in Sauna's framework here in businesses that you own, where maybe the businesses themselves, or maybe the businesses that are associated with them begin increasing in price very, very rapidly.

51:30Preston Pysh:But if they aren't really showing a rapid increase in price, there's no real need to use this framework because you're probably just going to scare yourself and there's much, much better uses of your time. So a potential bubble is forming today that I think is very fascinating. Now, while I don't have direct exposure to it, I do have indirect exposure. So it does take some of my thinking time. And the rest of this episode will be devoted to discussing AI and whether or not it's in a bubble using Insana's framework. So throughout this episode, we've gone through a number of different bubbles and we've traced the recurring patterns that happen in each of these bubbles.

52:05Preston Pysh:So nearly every bubble is driven by things like human behavior, technological innovations, access to easy money, which all ends up culminating in speculative of excess. And while AI ticks a few of these boxes, I'm not sure we're all the way there yet. Now, there is no way that I can know AI is in a bubble with 100 % certainty. All I can do is use the tools available to observe where we are today, which at least will inform me where we could be headed in the future. But as Insana's call and the great financial crisis showed, you can still be mostly right, and yet the bubble won't pop for many years ahead.

52:37Preston Pysh:Now, where AI is strongest is in the general interest from the media and from investors. The fascinating thing about AI is that nearly anybody can touch it. It's not some sort of abstract thought or idea just floating around. It's tangible. You can easily go and use AI right now to improve your efficiency, learn something new, or automate a process. One interesting area of AI to ponder is just what kind of bubble it could be. Howard Marks outlined two types of bubbles. The first is inflection bubbles, which are in some ways good bubbles. These are bubbles that deliver truly transformative technology to humankind.

53:12Preston Pysh:So whether or not investors win or lose money is another story, but the underlying technology remains a positive for mankind. Railroads and the internet are great examples of inflection bubbles. The second is mean reversion bubbles, or just simply bad bubbles. These are the types of bubbles where the markets tend to rise and drop precipitously, but there's actually no added benefit to society. So I think it's pretty evident that AI probably falls into the inflection bubble category. And while you can consider inflection bubbles to be good, that definitely comes with a caveat that I already mentioned here.

53:43Preston Pysh:And that's that you can still lose money investing into transformative technology that is actually transformative. But the reason that inflection bubbles can be considered good is just because of all the interest in that technology that it creates. Scientific progress ends up being compressed from decades to just years. Now, the thing with AI, like many other technological marvels from the past, is that there's just no simple historical benchmark to try and draw wisdom from. Is AI like railways or the internet? There's no way to know unless you can put yourself in a time machine and go into the future.

54:14Preston Pysh:The uncertainties in AI involve a few questions. Things like who's going to be the biggest beneficiary of AI? Which of today's leaders in AI infrastructure build out will be overturned like the early social media companies such as MySpace? Will AI profits go to vendors or will they be competed away by price wars? And will AI products be these specialized products or will they just be treated as commodities? The next area of exploration is regarding monetary policy. The thing about economic policy is that the past monetary policy can have second order effects that happen years down the road. So if we rewind back to 2020, 2021, when interest rates were rock bottom, this had several effects.

54:54Preston Pysh:Money was easy to borrow and retail investors flush with cash could invest it into the market. So today's AI buildout can be partially explained by the lagging effects of that added liquidity that we ended up getting post-COVID. And part of that is that companies funding it are also just wide-mode companies that are flush with cash on their balance sheets and generate cash from their operating business. But the buildout still can't be solely financed with cash. So JP Morgan analysts believe that the current cash outlay for AI infrastructure will cost somewhere around$5 trillion. But companies like Microsoft, Alphabet, Amazon, Meta, and Oracle only have about$350 billion total on their balance sheets.

55:34Preston Pysh:Speaker 1 So while part of it can definitely be funded with cash on hand, there's going to be significant amounts of leverage that are going to be required to build it out here. Oracle, Meta, and Alphabet just issued 30-year notes with average coupon rates around 5.7%. Current 30-year treasury yields are around 4.7%. So investors in those bonds are only getting 1 % returns above treasury yields, which is a pretty low spread given just how uncertain the results will be from these investments. Junk bonds right now yield around 7 % to 8%, but don't have these 30-year terms. So money right now isn't the cheapest it's ever been.

56:08Preston Pysh:We aren't in that zero interest rate world anymore. Although you can argue that the Magnificent Seven's cost of equity has gone down as those businesses tend to trade at premiums. But there are other deals out there that maybe do make it appear that money is easy to get a hold of. So Marks highlighted a deal in his recent memo. So a company named Thinking Machines, which is an AI startup by former OpenAI executive Mira Mirati, just raised about$2 billion in its seed round at about a$10 billion valuation. Yet the company doesn't have a product, nor do they even tell investors what plans they have for a product.

56:44Preston Pysh:So one investor went to a pitch with Mirati and Mirati said, we're doing an AI company with the best AI people, but we can't answer any questions. And if you think that's completely wild, with this first seed round taking place sometime around October 2025, there's actually news today that the same startup is now seeking another funding round, valuing the business at $50 billion. Another former open AI scientist raised$2 billion for his company, again, with no product, valuing it at$32 billion. So perhaps money is cheaper than what treasury yields are telling us. Now let's turn our attention to government largesse, or fiscal stimulus.

57:20Preston Pysh:Here, I think it's pretty clear that the government is indirectly subsidizing the AL buildout via things such as semiconductor incentives and cloud infrastructure spending. So in the US, the CHIPS Act authorized nearly$53 billion in federal spending to support US semiconductor manufacturing and research capacity. This has helped chip companies like Micron, Intel, TSMC, Samsung, and Texas Instruments fund US-based fabs and R &D labs. The EU has a similar CHIPS Act with$16.5 billion in funding. China has spent an undisclosed amount on shoring semiconductor manufacturing and on acquiring new technology.

57:57Preston Pysh:So, you know, around the world, governments are fighting to get ahead in this game. And while these investments aren't directly tied into AI, they are indirectly as the AI build-out requires these advanced chips. Where the government is spending money, investors will continue to flock to. So now we get to two areas where I think AI is a little weaker as viewed through Insana's framework. So the first is that bubbles are most likely to form under favorable economic conditions. So what are economic conditions like today compared to five years ago? I would say they are probably less favorable. Interest rates are higher, which increases borrowing costs.

58:31Preston Pysh:The US government hasn't distributed handouts lately, like it did to address the pandemic. So a lot of the fiscal stimulus has kind of faded away. The GDP growth in the US in 2025 was just 2.1%, which is the lowest since 2019 except during the COVID recession. Unemployment rates are currently around 4%, which is higher than any year since 2019 other than the two years immediately after COVID-19. So I wouldn't say economic conditions today are the best they've been. The last factor is public participation and speculative behavior. I would say with some of the private deals I've already highlighted along with these 30-year bonds that I spoke about, there is undoubtedly some speculative behavior going on.

59:09Preston Pysh:Firms with no products being valued at$30 to$50 billion within months definitely strikes a speculative note. But looking at IPO markets, which is an excellent indicator of bubble-like behavior among the masses, doesn't really show much evidence of a market-wide bubble. 2025 IPO proceeds were$38 billion. This pales in comparison to 2021's$142 billion. And while the IPO market has heated up since 2022, it still lags the years preceding COVID. And the absolute number of IPOs is just slightly above the average for the decade, but nothing too alarmingly high. Now, you could argue that the increase in the mag seven prices is part of a public participation in the bubble.

59:47Preston Pysh:After all, these are companies that are just gushing cash and they're investing in the future of AI. The S &P 500's results look much different from those of its equal weighted counterpart. So the total returns of the S &P 500 has been around 19 % per year compared with just 11 % for the equal weighted S &P 500 index. But even when you look at the evaluation of the MAG7, I would say as a whole, yes, it's expensive, but I would be very, very hesitant to say that it's in bubble-like territory. Sure, you're paying a premium to the index, but these businesses, they have better quality, they have wider moats, and they're growing faster than the average American business as well.

1:00:22Preston Pysh:So they deserve to have a premium. A company like NVIDIA, which trades at a trailing 12 months price earnings multiple of 44 times, is expensive, sure. But analysts estimate it's going to grow EPS nearly 60 % next year, which helps explain that high evaluation. Now, it's important to remember that expensive doesn't mean bubble. If you look forward to a business like NVIDIA, analysts actually believe that it will continue to grow EPS well above 50 % compounded into 2027. So if I look at the forward PE of NVIDIA, it's around 24 times, which is actually lower than the 27 times of the S &P 500. This just doesn't really scream bubble to me.

1:00:58Preston Pysh:But here's the thing. Bubbles can pick up momentum very quickly. And if history is a precedent, there are plenty of businesses inside the S &P 500, or that are even currently private, that can create a narrative based around it transforming itself into some sort of AI play. Now, some of these businesses maybe are actual AI plays, but some of them will probably try to take advantage of the speculative markets by claiming they're an AI business, when in reality, they're just putting lipstick on a pig. So how can we protect ourselves if AI remains top of mind for the market, and as the economy improves and mass participation becomes a reality?

1:01:34Preston Pysh:First, we must recognize that bubbles are simply a part of investor psychology. Whether you want to label AI as being in a bubble or not is much less important than understanding just how investors are behaving around it. So pay attention, read the news, talk to other investors, and observe how investors are behaving towards AI and AI stocks. If you think that AI is headed towards a bubble, some investors might see that as an opportunity to maybe try and get involved and try to time the market. They may tell themselves that they can get out of a position that begins getting into bubble-like territory before the market does.

1:02:07Preston Pysh:But this is a pretty challenging game to play. Even if you're right on the bubble, there's no way of knowing just when it's going to form or how your own perception of it will change over time. So make sure you size your positions right, stay disciplined on your evaluations, and don't allow your ego to inflate, which can cause several problems down the road. Next, you need to separate three layers of risk. Layer one is a technology. AI will truly be a transformer of technology, maybe the biggest technological breakthrough of my lifetime. But it's vital to remember that technological success does not equal investor success.

1:02:41Preston Pysh:My general strategy is to avoid investments that are spending money on the AI buildup. To me, that's just pure speculation. I'd rather invest in businesses that are already leveraging AI capabilities today. The second layer is the business. When it comes to who's going to win the AI race, anyone who claims to know the answer, I think is probably just a liar. In the big bubbles of the past, many entrants tried to win the market, but very few actually ended up surviving over time. Let's just look at the automotive sector in the US. So in the early 1900s, the US market had hundreds of car manufacturers.

1:03:15Preston Pysh:And that's because there were just no clear winners, just like AI today. In cars, there are no barriers to entry and there was rapid innovation. Sounds an awful lot like these AI startups that are starting to pop up everywhere. But today, there are only three car manufacturers, Ford, General Motors, and Chrysler. The survival rate here was less than 1%. So if you're looking at an AI investment, you have to ask yourself whether the business that you're investing in is truly a business that has some sort of competitive advantage, or if it's just purely based on hype. And my guess is the majority of these new businesses popping up are going to be based on hype.

1:03:47Preston Pysh:The third layer is to analyze value. This is where bubbles form. When investors are willing to pay higher and higher prices for an asset, bubbles form because new investors come in thinking someone else will buy it from them at an even higher price. The stock price embeds things like optimism, which is why it's imperative to understand it deeply. The businesses that I own are exposed to AI, such as Luma and Topicus are already businesses that are really profitable. They have recurring revenue engines. So I'm very comfortable holding them. They're not some sort of speculative pre-revenue company, which could just as easily go to zero as it could go to 100X.

1:04:24Preston Pysh:So I have to think long and hard about what they're valued at. And if there's some sort of mispricing that I can take advantage of. One strategy that I like to use here is to just view data from a business or industry's historical averages. We did this exercise already on the NASDAQ earlier, but if you're looking at AI companies, consider the historical averages for their earnings and cash flow multiples, and then just compare them to today. Are they below, in line, or above historical averages? Beware when the multiple is double its historical average. The next point of concern is once again related to value.

1:04:53Preston Pysh:You may have found a business that is utilizing AI to turn itself around. As a result, profit margins are set to explode, And the historical PE numbers maybe just aren't as helpful as they once were simply because the business model has fundamentally changed. Now you need to give yourself a terminal value reality check. So I like doing this when I'm looking at basically any business that I own. So if I have a business that I think can grow at 15 % per year, what does that company look like in five years? So let's use Lumine as a very simple example. In order for them to continue growing their intrinsic value around 15%, they're probably going to need to add something like three to five new acquisitions per year.

1:05:31Preston Pysh:Maybe that scales up over time. But let's just use that three to five year number here for simplicity's sake. So in five years, that would be a total of 15 to 25 acquisitions. Now, when I think hard about that, and I think, is that possible for Lumine to do? The answer is very easy. Yes. I think they could do that quite easily. So if we contrast that now to a business such as Thinking Machines and its$50 billion valuation, what really needs to happen for the company to grow into that valuation? Even though I'm not a fan of valuing businesses on revenue multiples, my guess is that a business like this is probably going to be evaluated on that while they figure out how to turn a profit.

1:06:08Preston Pysh:So an ARR business growing, let's say 20 % per year could easily fetch a 10 times ARR multiple, especially in an industry with tailwinds and a lot of media coverage. So in that case, it needs to generate$5 billion in revenue over the next few years. Is that possible? I have zero conviction of my ability to make that decision, which is why I have no skin in the game. So whenever you're looking at any business, whether it's AI or not, I would recommend sense checking what you think it can be worth in the future. If it makes no sense, such as that Yahoo example that I gave earlier, where it required 18 billion users when the world's population was just a fraction of that, then you know that the evaluation has become incredibly stretched and may be in bubble territory.

1:06:47Preston Pysh:The next thing you can do is try to identify bubble behavior. Howard Marks mentions that he's no AI expert in his latest memo, which I'll link to in the show notes. But he is a master at observation. He's someone who likes making these kind of deal making vignettes to help him better understand investor psychology. This allows him to determine if he should be aggressive or conservative. Now, given the AI examples that I've covered here, I would say that being overly conservative towards AI businesses today is probably an excellent idea. Now, if you want to move away from more subjective observations, such as looking at bubble-like behavior, then just look at objective measures.

1:07:21Preston Pysh:So I already went through some of the deals involving debt for AI companies in public and private sectors. So look at other deals and see what other businesses are paying for debt and for mergers and acquisitions. Does the price that they're paying defy reality? Then that's a great signal to just take a pass. So what does all this really mean for how I actually manage my portfolios? It's quite simple. The first thing is that I cap my exposure to narratives and I focused as exclusively as I can on the fundamentals. Now, I know I'll screw up on that because it's impossible not to get caught up in narratives at some degree, but you have to try hardest to focus on the fundamentals of the business.

1:07:58Preston Pysh:The second is to ensure that my winners are growing specifically because the intrinsic value of those businesses are compounding and not purely because of multiple expansion. If I have a business that I held over five years and intrinsic value didn't grow at all, the share price went 10X or 5X even, that's pretty scary because basically that just means that there's a narrative around the business and more and more investor interest is being generated by that narrative and not by the changing fundamentals of the business. And the third one here is to just simply observe deals that are going on in your company's industries.

1:08:29Preston Pysh:Do they seem reasonable or are they based on dreamy narratives? If narratives are propping up a business or an industry that your business is involved in, be extremely cautious. Selling might be the best idea, even if it's a company that you really, really like. Thanks so much for spending time with me today. If you'd like to continue the conversation, please follow me on Twitter at IrrationalMRKTS or connect with me on LinkedIn. Just search for Kyle Grief. I'm always open to feedback, so please feel free to share with me how I can make this a better experience for you. Thanks for listening and see you next time.

1:09:13Kyle Grieve:This podcast is for informational and entertainment purposes only and does not provide financial, investment, tax or legal advice. The content is impersonal and does not consider your objectives, financial situation or needs. Investing involves risk, including possible loss of principle and past performance is not a guarantee of future results. Listeners should do their own research and consult a qualified professional before making any financial decisions. Nothing on this show is a recommendation or solicitation to buy or sell any security or other financial product. hosts guests and the investors podcast network may hold positions in securities discussed and may change those positions at any time without notice references to any third-party products services or advertisers do not constitute endorsements and the investors podcast network is not responsible for any claims made by them copyright by the investors podcast network all rights reserved