Stock Talk - From Dot-Com Bubble to AI, “Situationally UnAware” Stock Talk Update August 7, 2026

Starting point is 00:00:00 Every major investment cycle creates a hero or two. It may be one person, one fund, or one company that sees the future before everyone else. They invest early, the investment rises, and for a while, they look like the smartest investor in the world. Then more investors notice the same idea, more money rushes in, prices rise faster, confidence grows. Sometimes the trade goes nearly straight up. But then the cycle changes. Cash becomes harder to find, borrowed money becomes harder. to manage, lenders ask for more collateral, investors start selling, and the same trade that

Starting point is 00:00:36 made someone look like a genius can become the trade that puts the whole fund at risk and out of business. The investor may be remembered as a one-hit wonder. In more serious cases, clients may suffer major losses, and the manager becomes a warning for the next generation. We saw this pattern during the dot-com bubble, and according to public reports, we saw a version of it again in the AI market just last month in July. So this week, we're going from the dot-com boom to the AI boom and asking one important question. What can retirees learn when the market's smartest trade suddenly breaks? This video is for education only, not personal investment tax or legal advice. No investment result was promised and every investment involves risks, including the possible loss of principle.

Starting point is 00:01:19 Now, this isn't a video claiming that AI is finished. It's not a prediction that the entire technology market's about to collapse. The real real. lesson is more useful than that. An investor can be right about the future and still lose money in the present. The idea may be right. The technology may be real. The demand may keep growing. But the price paid, the amount invested, the use of debt, and the time needed for the idea to work can still cause serious losses. That matters for every investor. But it may matter even more for retirees, people close to retirement, because they may not have decades to recover from a major decline. To understand the risk, I want to make three comparisons.

Starting point is 00:01:57 between the dot-com period, 1998 through 2000, and the AI investment cycle of today. First, we'll look at what happens when a smart idea becomes a crowded and leveraged trade. Second, we'll look at the huge physical build-out behind each technology boom. Third, we'll look at how higher interest rates, oil, inflation, and tighter money

Starting point is 00:02:17 helped break the dot-com cycle and why parts of today's market may rhyme with that period. Investors, early in my career, I lived through the dot-com bubble. I analyzed it, I invested during it, and I was fortunate enough to avoid some of the worst damage when it broke. I lived and worked in San Francisco as the technology boom grew. At first, it was business and technology stories. Then it became a stock market story.

Starting point is 00:02:41 Finally, it became a full market bubble. My first son, Kyle, was born in San Francisco in late 1997 while the boom was building. After the failure of long-term capital management in October 1998, the Federal Reserve added support to the financial system, markets recovered, technology stocks pushed higher, and excitement grew. Investors by the middle of 1999, Texas was calling my family home. I left a world where almost everything seemed to be about technology investing, and we moved back to Houston. But the years I spent in the Bay Area gave me a close view of some of the smartest technology

Starting point is 00:03:15 thinkers and investors I've ever met. One well-known firm at the time was Amarindo, led by Alberto Villar and Gary Tanaka. The firm became closely linked with the belief that the end of the industry. The internet would change business in daily life. The basic belief was correct. The internet did change the world. But being right about a technology doesn't mean every investment connected to that technology will succeed.

Starting point is 00:03:38 It also doesn't protect a concentrated portfolio when valuations fall and buyers disappear. The recent problems reported in the media about Leopold, Usherner's AI-focused hedge fund, situational awareness, brought back those memories. Before we continue, I've noticed that many people watching these videos have yet to subscribe. Please take a moment to hit the subscribe button.

Starting point is 00:03:58 You can also turn on the notification so you'll know when new educational content is posted. Here's the part investors often missed. The greatest danger may not appear when the idea is weak. It may appear when the idea is so strong, then investors stop asking what could go wrong. And almost every technology boom, people begin to confuse two very different things,

Starting point is 00:04:18 being right about the future and being protected from what can happen today. And once we compare the portfolio risk, physical buildout and the cost of money, we can see why a real technology boom and still create painful investment losses. Let's begin with the first comparison, a celebrated investor in the concentrated trade. During the dot-com era, Amarindo became one of the best-known technology investors in Silicon Valley. Its technology investments produce strong gains during the boom and more money flowed into

Starting point is 00:04:47 the firm. Vila argued that the internet would transform companies, communication, shopping, entertainment, and daily life. On that big idea, he was early. History showed that internet effects may have been even greater than many early supporters expected. But the portfolio was heavily concentrated in technology shares near the time the NASDAQ peaked in March of 2000. When money became tighter, investors became less willing to pay extreme prices and technology valuations fell. Concentrated technology portfolios suffered major losses. The internet did not stop growing. Investment cycle changed. The difference is the center of this story. More than two decades later, a similar pattern

Starting point is 00:05:27 developed around AI, although the company's market structure and amount of capital are very different. Leopold Oshenbrenner, a young former Open AI researcher, launched a hedge fund called Situational Awareness in June of 2024. He published a long paper called Situational Awareness the decade ahead. I'll place a link to the paper in the description for viewers who want to read it. The paper made a strong case that advanced AI would require enormous amounts of computing power, semiconductors, memory, networking equipment, electricity, land, cooling systems, and investment capital. That broad direction has been supported by the rapid growth in AI infrastructure spending. The fund grew quickly as its AI-related investments rose.

Starting point is 00:06:12 Public reports in June of this year said the firm had grown to over 20 billion in assets under management, but the portfolio also used leverage. Leverage means borrowing money to increase the size of investment. For example, an investor may have $1 of client capital, a control, more than $1 of investments, because part of the position is funded with borrowed money. This can increase gains when the price rises, but it also can increase losses when prices fall.

Starting point is 00:06:38 And when lenders become worried, they may demand more cash or collateral. This is called a margin call. In July of this year, many AI-related and semiconductor stocks fell sharply. According to an investor letter reported by a. Reuters, the value of situational awareness portfolio fell by 67% during July. The fund sold most of public stock portfolio and removed its leverage.

Starting point is 00:07:01 The same lever said that the fund remained about up 80% for the year because of its earlier gains. Those figures come from third-party public reporting and haven't been independently verified by us. This full context matters. The fund didn't simply place a foolish bet on fake technology. fake technology, the manager had identified a real long-term trend and had produced major gains before the decline. However, concentration, leverage, falling stock prices, and weaker market liquidity created a serious short-term problem. The manager reportedly told investors that the fund had come closer to permanent damage than the firm believed was acceptable. That's the lesson.

Starting point is 00:07:42 The AI investment cycle doesn't have to end for an AI-focused investor to face major losses. A market can force an investor with a sound long-term idea to sell during a terrible short-term moment. For retirees, that risk deserves special attention. You can be right about a technology and still lose money because you paid too much, invested too much at one theme, used borrow money, or didn't have enough time to wait for the idea to recover. Now let's move on to a second comparison, the physical buildout. The Internet wasn't built with investment ideas alone. It required fiber optic cable, routers, switches, servers,

Starting point is 00:08:18 telecom equipment, data centers, and semiconductor factories. By the late 1990s, companies were spending based on what they believed future demand would become. The equipment was real, the networks were real, the future uses of the internet were real. But in many cases, too much money arrived before customers and profits were ready. Companies installed huge amounts of fiber optic cable. Some of that cable was called dark fiber because of the that had been placed in the ground but wasn't yet being used. Investors expected demand to arrive quickly. Instead, full use took years, if not decades. Consumers had to move from slow dial-up

Starting point is 00:08:55 service to faster internet. Entertainment had to move from video stores to DVDs sent through the mail, and then to streaming. The internet moved from simple websites and email to photo sharing, music, full video, live broadcasts, cloud software, and mobile apps. The infrastructure was useful, But the money spent on it didn't always earn a good return at the time it was built. The AI cycle is following a similar path, but with different limits, faster adoption, and a much larger amount of money. Major technology companies are spending hundreds of billions of dollars on data centers and other AI systems. The exact forecasts differ and future spending can change. Still, the direction is clear.

Starting point is 00:09:35 AI infrastructure has become one of the largest areas of business investment in the United States. First, companies need land and buildings. Then they need power, water, cooling system, backup systems, and connection to the electrical grid. After that, they need equipment inside the building. That includes graphics processing units called GPUs and other AI accelerators. It includes high bandwidth memory, advanced chip packaging, semiconductor factories, lithography tools, networking chips, optical connections, and power equipment. This semiconductor industry isn't just watching the AI boom. It's one of the main roads the boom has traveled on.

Starting point is 00:10:11 That created major demand and strong earnings opportunities, but it also creates cycle risk. The semiconductor factory can take years to plan and build. It may cost tens of billions of dollars. When chips are hard to get, customers may place large orders to make sure they receive enough supply. Some customers may place orders with more than one supplier or distributor. Suppliers see the strong demand and build more capacity. But new factories and equipment take time to come online, then the market can change.

Starting point is 00:10:38 Customers may decide, I've got enough inventory. They may slow spending. Their own customers may delay orders, or credit markets may tighten, making it harder for companies to borrow the money they need to build more data centers. When that happens, chip orders can fall faster than the final demand for AI services. That's how a long-term growth story can still create a short-term industry downturn. The important question is no longer just when AI demand is real. It clearly appears real today.

Starting point is 00:11:07 The harder question is whether it is, each new dollar spent on AI equipment will produce enough new revenue, new cash flow, and profit to justify the cost. That's the tipping point investors need to watch for. Many large technology companies were once known as Asset Light businesses. Asset Light meant that you could grow without spending huge amounts on factories, buildings, and equipment. AI is pushing some of those companies towards more asset-heavy model.

Starting point is 00:11:32 More of their operating cash may need to be used for data centers, chips, electricity, and long-term contracts. That doesn't mean spending is bad. The spending may create future growth, but it does mean less cash is available for stock buybacks, dividends, debt reduction, and other uses. Investors must watch return on that spending, not just the size of the spending.

Starting point is 00:11:53 Now we reach the third comparison, how the dot-com bubble broke and why today may rhyme. The dot-com bubble didn't break because the internet was fake. It broke because investors placed a very high price on the future growth at the same time that money became more expensive and market liquidity. became tighter. After long-term capital management failed in October of 1998, the Federal Reserve lowered rates and worked to support the financial system. Markets moved higher. Then, near the end of

Starting point is 00:12:20 1999, the financial system received extra cash because of fears that the Y2K computer problem might disrupt banks and business systems. When the year 2000 arrived without the fear of shutdown, that special support was no longer needed. The Federal Reserve tightened policy. Short-term interest rates rose, and dollars stayed strong, oil prices rose from their low levels in 1998, inflation pressure began to return. At the same time, many technology evaluations had moved far beyond what normal business profits could support. Cisco systems traded at the extremely high earnings multiple of around 100. Many public technology companies had little revenue and no profit at all. When investors finally began demanding a clearer path to earnings, the next buyer disappeared. New stock offerings became

Starting point is 00:13:05 harder to sell, funding became difficult. The system that had pushed technology stocks higher began working in reverse. Now look at the markets this year in 2006. The economy is growing. AI demand is real. Business investment is very strong. And inflation has remained a concern and interest rates are still higher compared to much of the period after 2008 financial crisis. As of this writing, the federal funds target range of three and a half to three and three quarters percent, the two-year Treasury yields about four and a quarter percent, and the 10-year Treasury yield was about 4.68 percent. These rates change each day, so viewers should check current information before making any decisions. Oil prices and

Starting point is 00:13:47 conflict in the Middle East have also added to uncertainty to energy costs and inflation. The dollar has at times gained support from high US interest rates and demand for safer assets. Higher long-term rates and a firm dollar and certain energy prices and less room for the Federal Reserve to lower rates and place pressure on growth investments. That's one important way the second half of 2006 may rhyme with 1999 and early 2000. But the differences are just as important. Many of today's largest AI companies produce real revenue, real profits, and very large amounts of operating cash flow. That's not the same as dot-com period when many public companies had weak business models and little or no revenue.

Starting point is 00:14:27 Large technology companies can fund a meaningful part of their AI investments from their own business operations. Some of the most closely watched AI companies also remain private, which helps part of the market's risk being pushed outside normal public stock indexes. However, strong cash flow doesn't make price or valuation unimportant. It changes the type of risk. The biggest danger today may not be the hundreds of public companies with no real businesses. The danger may be that profitable companies spend so much on AI infrastructure, the free cash flow. falls, debt rises, and the next data center earns less than investors expected. Debt also is becoming more common in the market for common stockholders in this financial

Starting point is 00:15:09 structure. Interest in principal payments must be made before value can flow to common equity shareholders. The dot-com cycle broke when money became more expensive and investors stopped rewarding growth at any price. The AI cycle could face similar pressure from the same basic forces. Their inflation adjusted interest rates, higher energy costs, a stronger dollar, slower profit forecasts, weaker semiconductor orders, more debt, and falling confidence in the return earned on AI spending.

Starting point is 00:15:38 None of these signs alone prove that AI downturn is coming, but together they can increase risks. This brings us back to the two investors at the center of the story. Alberto Villar saw the power of the internet early. Leopold Usherner saw the scale of the AI infrastructure build out early. main technology ideas weren't necessarily the part that failed. The greater problem was the investment structure built around those ideas. Strong vision can still be damaged by concentration. A correct forecast can still be damaged by leverage.

Starting point is 00:16:09 A great company can still be a poor investment if the price is too high. And a long-term opportunity can still create short-term losses that investor can't afford to wait through. That's especially important in retirement. A younger investor who has suffered a major decline may have years or decades of new income, and savings ahead. Retiree may be withdrawing money from the same portfolio at the time prices are falling. That can make recovery harder. The goal isn't to avoid every falling market. That's not realistic at all. The goal is to avoid building a financial plan that depends on one company,

Starting point is 00:16:42 one theme, or one forecast working perfectly. AI may change the world. The internet certainly did, but create technology doesn't cancel the basic rules of finance. Price still matters. Cash flow still matters, debt still matters, diversification still matters, and the amount of time an investor has to recover still matters. Retirees, near retirees, and other long-term investors may want to watch four broad areas. First, watch semiconductor orders and inventory. Stocks often react before factories and company reports fully show that demand has changed. Rising inventory, shorter delivery times, canceled orders, or slowing equipment demand may be an early sign that the cycle is cooling. Second, watch the free cash flow of the largest AI spenders. Revenue can rise while the

Starting point is 00:17:27 cash left for shareholders fall. The key question isn't only how much a company spends, it's whether that spending produces enough future profit in cash to justify the cost. Third, watch 10-year treasury yields and the US dollar. When both move higher, financial conditions may become tighter for global businesses and for technology stocks whose expected profits are far in the future. Higher rates can also lead investors to pay lower price to earnings ratios for stocks. Fourth and finally, watch leverage in credit conditions. A normal stock market correction is common, but forced selling by a leveraged investor can turn a normal decline into a larger liquidity event.

Starting point is 00:18:05 According to public reports, that's part of what happened with situational awareness just last month. A different form of fast-moving liquidity crisis appeared during the failure of Silicon Valley Bank in March of 2023. The details were different, but both events showed how quickly. Pressure can grow when confidence falls and cash is needed all at once. The Internet changed the world. AI may change it in many ways we can't fully see today,

Starting point is 00:18:29 but investors don't need to choose between ignoring the opportunity and betting their financial future on it. They can take part while still respecting the trade. They can watch valuation. They can follow cash flow in debt. They can limit concentration. And they can make sure their investment mix fits their own goals, income needs, time horizon, and ability to handle loss. The final lesson is simple. Whatever your investment style or view of the economy, stay aware of the risks inside your own financial plan. Don't let the excitement, fear, or greed make the decision for you.

Starting point is 00:19:01 And never let one theme, no matter how smart, powerful, or exciting it appears, control your retirement future. In other words, every investor needs to remain situationally aware. If you're retired or close to retirement, your plan shouldn't depend on one market trend working perfectly. Give us a call here at Oak Carbis Financial Group to review. your retirement plan and make sure it's built around your income needs, goals, and risk tolerance. Stay situationally aware, not just of the market, but of your retirement. All content contained with an Oak Harvest podcast expresses the views of the speaker and is for informational purposes only. It is based on information believed to be reliable when created,

Starting point is 00:19:40 but any cited data, indicators, statistics, or other sources are not guaranteed. The views and opinions expressed herein may change without notice. Strategies and ideas discussed may not be right for you, and nothing in this podcast should be considered as personalized investment, tax or legal advice, or an offer or solicitation to buy or sell securities. Indexes such as the S&P 500 are not available for direct investment, and your investment results may differ when compared to an index. Specific portfolio actions or strategies discussed will not apply to all client portfolios. Investing involves the risk of loss and past performance is not indicative of future results.

Stock Talk - From Dot-Com Bubble to AI, “Situationally UnAware” Stock Talk Update August 7, 2026

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.