September 4th, 2026 | AI Capex Bubble Bursts, A Market Like 1901, Jobs Report Beats Expectations, Sports Betting as Investing, Big Food Battles Diet Drugs, Be Your Own Bank? & More
The AI Capex Bubble Is Starting to Look Crazy I keep coming back to the same question when I look at the incredible amount of money being poured into artificial intelligence: Where is all of this capital ultimately going to earn a return? Since the beginning of 2024, roughly $500 billion has been spent on chips, $350 billion on power infrastructure, $200 billion on construction and $100 billion on networking. That's approximately $1.1 trillion of AI infrastructure spending in less than three years. For perspective, the entire S&P 500 spent roughly $575 billion on capital expenditures in 2021 right before ChatGPT even existed. And the spending is accelerating. In 2021 The four major hyperscalers—Microsoft, Amazon, Alphabet and Meta— spent about $125 billion on new plants and equipment. It’s now estimated that they will spend $1 trillion, which is about half of total capital spending for the S&P 500 and the companies could spend roughly $3.7 trillion through 2029. Add companies such as Oracle, OpenAI, SpaceX and others, and total AI spending could approach $6 trillion by the end of the decade. Those numbers are almost difficult to comprehend. And here's where I think the historical comparisons to railroads and the internet become interesting. Yes, those were enormous infrastructure buildouts too. But the economic opportunity created by those technologies was incredibly clear. The railroad connected producers with consumers, opened new markets, lowered transportation costs and allowed goods to move across the country. The internet created entirely new businesses and fundamentally changed commerce, advertising, communications and how we work. I don't see AI in quite the same light. I see enormous potential, but I don't yet see the same obvious economic expansion that will ultimately justify trillions of dollars of infrastructure spending. And now we're starting to hear another argument: "Look at the cloud. Look at how much money the cloud is generating. That's proof the AI infrastructure will earn a return." I'm not sure I buy that. That's a little like building railroads and then saying: "Look at how much money we're making selling railcars. Look at the demand for locomotives and railroad equipment. Clearly the railroad investment is paying off." The problem is that's not where the ultimate economic return came from. The return came from transporting goods and people. The railroad was valuable because businesses used it to create economic activity. The same is true of the internet. The real economic payoff wasn't simply selling servers and networking equipment. It came from everything built on top of the internet. So with AI, I think the ultimate question is not: "How much revenue are Nvidia, the cloud companies and data-center operators generating?" It's: "How much NEW economic value is being created by all of this computing capacity?" That's a much harder question. Because if we're essentially spending trillions of dollars building increasingly powerful computers, data centers and power infrastructure so companies can sell more computing capacity to other companies that are also spending billions on AI infrastructure, we need to be careful about confusing activity with economic returns. And this is where the bubble argument gets interesting. A recent Barron's article points out that historically, transformative technology booms have been able to absorb enormous amounts of capital before eventually running into trouble. Its "rule of 25" suggests that previous infrastructure booms became particularly vulnerable when investment approached roughly 25% of GDP. The railroad boom saw about $2.5 billion of rail spending before the 1873 panic and GDP was about $10 billion a year. Internet infrastructure saw about $1.5 trillion of investment before the bust and back then GDP was only about $6 trillion. For today's roughly $30 trillion U.S. economy, that would be around $7.5 trillion before we saw problems. That's being used as evidence that the AI boom has plenty of room to run. And maybe it does. But here's the funny part. We're increasingly hearing very smart people say: "Yes, this is going to end badly." "Yes, there is too much capital being deployed." "Yes, there will eventually be excess capacity." "Yes, the financing is getting complicated." But then comes the qualifier: "Just not yet." That might be the most dangerous phrase in investing. Because that's exactly how bubbles work. When I look at $1.1 trillion already spent, and potentially $6 trillion by the end of the decade, increasingly creative financing structures and companies racing to build capacity before we fully understand the ultimate demand, it starts to feel less like a normal technology cycle and more like a capital spending boom. Maybe the bubble doesn't burst this year. Maybe it doesn't burst next year. But when almost everyone agrees there is a bubble and the only disagreement is about when it ends that's usually when I start paying very close attention. The technology can be real. The demand can be real. The companies can be profitable. And it can still be a bubble. The Stock Market Today Resembles the Stock Market of 1901 Some people believe they are witnessing something completely different in the stock market today and that what is happening now has never happened before. They believe the market will continue rising forever, and that there is simply no way they can lose. History tells us otherwise. Time and time again, we see the same patterns repeat themselves. Surprisingly, the stock market of 1901 had many of the same characteristics we are seeing today. For starters, there was a tremendous amount of trading back then like there is today. In 1901, the turnover rate on the New York Stock Exchange reached 319%, meaning stocks were changing hands roughly every 16 weeks. They also had something that resembles today's prediction markets. Back then, they were called bucket shops, where people could bet on whether a stock would move up or down. Many were led to believe they were participating in the same type of opportunity as wealthy investors. In reality, they were speculating and many people who didn't know better confused gambling with investing. Leverage was also widely used. Investors could put up as little as $10 and control as much as $300 worth of stock. That kind of leverage could produce enormous gains when markets were rising, but it could also lead to devastating losses when they turned. And this is where human psychology comes into play. People's emotions are often far stronger than their logic. The more the market rises, the more people begin to believe it will continue rising and that a crash is unlikely to happen anytime soon. When investors become excited because they are making easy money, they can lose sight of the difference between investing and gambling. The problem is that gambling can feel like investing when you're winning. The market's performance in the early 1900s is a good example. The stock market rose 19% in 1900, another 20% in 1901 and 5% in 1902. Then came 1903, when the market declined 23%. But the good times returned, and over the next three years the market gained roughly 69%. Then came the Panic of 1907, and the stock market fell roughly 30% that year. The lesson isn't that today's market will follow the exact same path. It won't. The lesson is that human behavior hasn't changed much in more than a century. Greed, fear, leverage, speculation and the belief that "this time is different" have been part of financial markets for generations. As the saying goes, history may not repeat itself, but it definitely rhymes. Investors would be wise to study those rhymes and remember that making money in a rising market doesn't necessarily mean you're investing wisely. Sometimes, it simply means you haven't experienced the other side of the cycle yet. The Jobs Report Was Much Stronger Than Expected Today’s jobs report was a big surprise. The U.S. economy added 162,000 jobs in August, well above the roughly 53,000 expected and the strongest monthly gain in five months. Even more importantly, July was revised from a loss of 23,000 jobs to a gain of 21,000. June was also revised higher, meaning the previous two months were collectively revised up by 55,000 jobs. The unemployment rate remained at 4.1%, but there was an interesting development underneath that number: the labor force increased by 683,000 people, while household employment increased by 569,000. The labor-force participation rate also rose from 61.4% to 61.6%. It is still down by 0.5% since January, but it’s a positive to see it moving in the right direction. So, we had substantially more people entering the workforce without the unemployment rate increasing. That's a pretty good sign. There was also a significant difference between industries. Food services and drinking places added 59,000 jobs, while local government education added another 42,000 and construction added about 22,000. Health care, which has been a large source of employment growth, saw a gain of just 13,000, compared with the monthly average of 32,000 over the prior 12 months. On the other hand, the information sector continued to lose jobs as information-related industries reported a loss of 23,000, putting the 12-month average at a loss of 8,000. This is worth watching given the impact of automation and AI on certain white-collar industries. Another positive: the average workweek increased to 34.4 hours, the highest level since March 2024. More hours worked can be just as important economically as more workers being hired. But there is one area that isn't quite as strong: wages. Average hourly earnings increased just 3.1% from a year ago. That's a healthy increase, but wage growth continues to moderate, and this marked the lowest growth in 5 years. And then we have the JOLTS data. The latest report showed 7.27 million job openings in July, that's approximately 1.1 job openings for every unemployed person. That is an important distinction. The labor market is clearly cooler than it was a few years ago, but there are still more available jobs than unemployed workers. Put it all together and I think today's report tells us something pretty simple: The labor market is still healthy. Job growth has cooled considerably from the boom years, but unemployment remains low, the labor force is expanding, job openings remain above the number of unemployed workers, and today's payroll number was substantially stronger than expected. This also makes the Federal Reserve's decision much more difficult. If the Fed's primary concern is a rapidly deteriorating labor market, today's report doesn't provide much evidence for that argument. Now the focus shifts back to inflation. If inflation remains sticky while employment is holding up this well, the argument for aggressive rate cuts becomes much harder to make. The next big test for the Fed is going to be the inflation data. Sports betting as an investment strategy? This is crazy. According to a Siena Poll, more than a quarter (27%) of Americans and over half (52%) of men aged 18 to 49 say they have an active online sportsbook account. That’s not a problem to me if you view sports gambling for what it is…. Which is gambling. The bigger problem I see is another recent survey from Betterment showed 52% of Gen Z investors (those born between 1997 and 2007) have redirected money intended for investing to sports bets. Think about that. We're not talking about occasionally putting $20 on a football game for fun. Some people are actually incorporating sports betting into their financial plans, viewing it as a way to build wealth, pay off debt, buy a home or reach other financial goals. People need to understand that gambling is a losing strategy in the long run. Let's say you have a 50/50 bet, essentially a coin flip. You might think that means you have an equal chance of winning or losing your money. Not quite. To win $100, you have to bet $110. If you win, you make $100. If you lose, you lose the entire $110. So even though the underlying event might seem like a 50/50 proposition, the sportsbook has built in an advantage. That's not investing. When you buy a stock, you're buying an ownership stake in a business. The company can generate profits, grow its earnings, reinvest in the business and potentially pay dividends. When you make a sports bet, you're putting money at risk on an outcome where the odds are designed to give the sportsbook an edge. The consequences of legalized sports betting may go far beyond losing a bet. Research from the New York Federal Reserve has found that the expansion of legal sports betting has coincided with rising rates of delinquency and bankruptcy. And the personal financial impact can be even more alarming. A 2025 U.S. News & World Report survey found that 25% of sports bettors said they had missed a bill because of their wagers, while 30% said they had taken on debt because of their betting. When people start borrowing money, missing bills and taking on debt to place bets, sports betting can become a serious financial problem. I understand why this mindset is developing. Younger people are dealing with expensive housing, high living costs and the frustration that traditional investing can take decades to build significant wealth. Sports betting offers something investing doesn't: the possibility of making a lot of money very quickly. But there's a catch. You can also lose a lot of money very quickly. And that's a terrible foundation for a long-term financial plan. Think about what young investors are seeing every day on social media. One video might explain the benefits of starting early, investing in a diversified portfolio and letting compound interest work for decades. Then, the very next video might show someone claiming you can make all of this money in a single football game by placing bets on a sportsbook. Which one sounds more exciting? Sports betting can also create an illusion of control. You may know a lot about football, basketball or baseball and feel like that knowledge gives you an advantage. You follow the teams, know the players, understand the matchups and watch every game. It can make you feel like you're making an informed investment decision. But knowing a lot about sports doesn't change the fact that the sportsbook sets the odds and builds in an advantage for itself. You might think, "I know more about this team than I know about the stock market, so I have a better chance of making money betting on them." That's a dangerous way to think about building wealth. If you want to build wealth, there's no substitute for saving, investing, compounding and time. Investing can feel slow. But slow is exactly what you want when you're building wealth. You don't need to hit a parlay to retire. How the Big Food Companies Are Battling Diet Drugs It is estimated that by 2035, 15% of the American population will be using or will have used GLP-1 drugs. No surprise, this is a potential problem for the big food companies, which have historically benefited from consumers eating more. We are still in the early stages of the diet-drug revolution, and some of the downsides are becoming more apparent. Some users report that food doesn't taste as good, sometimes describing it as tasting like Styrofoam. There are also concerns about muscle loss and, perhaps most importantly, the simple pleasure of eating for enjoyment. For decades, food companies have catered to consumers' taste buds with sugar, salt and an endless variety of flavors. But that strategy may not work as well for people taking GLP-1 drugs, whose appetites and food preferences can change dramatically. At the same time, there is a broader movement toward healthier eating, which creates another challenge for traditional food companies. So how are the big food companies fighting back? They're giving consumers what they want. One of the biggest concerns with GLP-1 drugs is muscle loss. Food companies see an opportunity here by developing products with more protein and fiber. For example, companies are introducing meals such as buffalo mac and cheese with 40 grams of protein. Another example is a chewy fudge brownie mix made with cottage cheese and a peanut-butter swirl. It not only looks appealing, but also offers significantly more protein. And food companies know something else about consumers: we eat with our eyes first. Packaging and presentation matter. Research has shown that phrases such as "good source of fiber" and "high in protein" resonate with consumers, particularly those who are trying to make healthier choices. At the same time, companies are tapping into something that never seems to go out of style: comfort and nostalgia. Phrases such as "Mom's meatloaf" or "Grandma's roast chicken" immediately create an emotional connection. One company has even developed a marinade and added grill marks to chicken breasts to make them look more appetizing. Smaller portions and convenience are also becoming increasingly important. Even if people want to eat healthier, they still have busy lives. They're working, socializing and taking care of their kids. Most people don't have the time or the desire to spend two hours preparing a healthy meal every night. And while the number of people taking GLP-1 drugs will likely continue to grow, I also think we'll see some people eventually stop taking them. Over time, some may decide the drugs don't work quite as well as they had hoped, while others may become frustrated with side effects, changes in how food tastes or the loss of muscle. When looking at themselves in the mirror one might think they look too skinny and rather frail because of muscle loss. There is also a bigger question: How much are people willing to sacrifice the pleasure of eating? Food has always been one of life's simple pleasures. For some people, after months or years of reduced appetite and diminished enjoyment from food, the desire to sit down and truly enjoy a great meal may eventually outweigh the benefits of staying on the medication. That creates an interesting challenge and opportunity for the food industry. The companies that succeed may not be the ones selling the most food. They may be the ones figuring out how to make healthier, higher-protein, higher-fiber foods that still look, smell and taste great. Because even in the age of diet drugs, people still want to enjoy their food. Financial Planning: What It Means to “Be Your Own Bank” Sometimes phrases like “be your own bank” or “borrow from yourself” are presented as sophisticated ways to access capital without being taken advantage of by a lending institution. But the truth is, it is impossible to literally “borrow from yourself.” You either use your own money, or you borrow someone else’s money. When you take a loan against a life insurance policy, use a HELOC, or establish a securities-backed line of credit (SBLOC), you are not borrowing from yourself. You are using your assets as collateral to obtain a loan from a bank or insurance company, which you must repay with interest just like any other loan. There is nothing inherently wrong with borrowing money, and using an asset as collateral can be a perfectly reasonable financial strategy. The problem arises when the ability to borrow against an asset becomes the justification for owning the asset in the first place. Phrases like “borrow from yourself” and “be your own bank” are marketing and sales tactics that can make a financial product sound more attractive than it actually is. For example, the fact that you can borrow against the cash value of a permanent life insurance policy does not, by itself, make permanent life insurance a good investment. The financial product should first stand on its own merits considering its costs, risks, returns, liquidity, and whether it actually meets your financial objectives. The ability to borrow against an asset should be viewed as a financing feature, not a reason to purchase the product. Borrowing can certainly be a useful financial tool, but the promise of being able to “borrow from yourself” should never be the primary justification for putting your money into an asset or financial product that you otherwise would not want to own. Company Discussed: DICK'S Sporting Goods, Inc. (Ticker: DKS)






