The Long Run - October 2026

Many of us know the cautionary tale of Icarus flying too close to the sun. We recall him crashing to his demise as the wax in the wings that his father had crafted for him melted. Icarus gives us insight into the frailties of our human nature. The story of his father, though, offers important lessons that can help us to consider the larger picture when managing risk in the most complex environments—and during the most consequential moments.  

According to lore, Daedalus was renowned as the preeminent master of mechanics, engineering, and invention in ancient Athens. He was lured to Crete by King Minos to design an impenetrable labyrinth that could contain a horrible, man-eating beast, the Minotaur. After completing the extraordinary marvel, he was imprisoned by the king. Trapped in a tower on land—and surrounded by Minos’ enormous navy in the waters all around Crete—Daedalus conceived of an escape that no jailkeeper would suspect: the air. He fabricated two pairs of functioning wings by securing tiers of graduated feathers, with flax cord along the structural ribs, and binding the bases with beeswax—deliberately arcing the frames to replicate the aerodynamic curvature of birds. 

Daedalus designed the wings knowing that they would need to do more than just carry him and his son off Crete. They also had to deliver the two of them back to safety, and well out of the range of Minos and his navy. He explained the specific operational tolerances for the journey to Icarus. Flying too low would allow the sea spray to saturate the feathers and drag him down. Climbing too high would expose the wax core to solar heat and melt the airframe entirely. We remember what happened to Icarus, but we forget that—after he buried his son on an island near Crete, Daedalus was able to fly himself to safety, far out of the range of Minos and his navy. His knowledge of the atmosphere around him—and his understanding of the limited tolerances that his inventions could function within—allowed him to overcome the pushes and pulls of human nature to achieve his larger goal.

Lessons for Today

In the realms of capital formation, stewardship, and investing there are few minds that match the precision (or wit) of Warren Buffett’s partner, Charlie Munger. Much like Daedalus, his approach to investing focused on a comprehensive review of the internal workings of a specific investment, as well as the larger, integrated ecosystem in which it operates. His extraordinary “Multiple Mental Models” framework for gathering, processing, and acting on information called on a wide range of disciplines, including history, psychology, physiology, mathematics, engineering, biology, physics, chemistry, statistics, and economics among others. They created a framework and analytical structure that enabled him to reduce complex investment problems in chaotic market environments into a clarified set of fundamental trade-offs and opportunities. These models contributed significantly to one of the greatest performance records in investment history. 

As we take on the investment opportunities and challenges of today’s rapidly evolving and ever-more complex world, Munger might insist that we first establish the larger goal correctly. The great advantage for the investor is time in the market and the ability to compound returns over time. Munger once said, “The elementary mathematics of compound interest is one of the most important models there is on earth.” Like Daedalus needing to get all the way to safety, Munger’s goal was to maintain his positions in the market over time to benefit from compounding and create extraordinary returns while seeking to avoid taking undue risk.

We subscribe to many of Munger’s core principles—but none more than the importance of a long-term perspective and the power of compounding returns. Munger was particularly astute at making good trade-offs and avoiding the type of asymmetric risks that could set an investor back to “square one” (a reference to being sent back to Go in the game of Monopoly). His frameworks provide important insights into where larger risks might hide in our current investment atmosphere—and where the longer-term opportunities might be.

Concentration Matters 

We have spoken now for a couple of years about the fact that our major capitalization-weighted indices are as concentrated in their top ten holdings as they have been in the last 100 years. The only other times we have been this concentrated at the very top of the S&P 500 index were during the Dot-Com Bubble Peak (1999-2000) and the “Nifty Fifty” Peak (1972-1973). In their 2020 long-term study, Large and In Charge?, Dimensional Fund Advisors found that in the 10 years prior to reaching the top ten, these companies beat the market by an annualized 10%. In the subsequent 10 years after joining the top 10, their annualized excess return dropped to -1.5% relative to the broader market.

Like us, Munger was comfortable with concentrated investment portfolios. He drew a sharp line, however, between prudent capital concentration and what he called institutional herding. Munger repeatedly observed that the fatal flaw of the Nifty Fifty era was taking a fundamentally sound concept—that exceptional franchises compound capital at high rates—and driving it to speculative extremes:

“Since it’s so obvious that investing in great companies works, it gets horribly overdone from time to time. In the ‘Nifty Fifty’ days, everybody could tell which companies were the great ones. So they got up to 50, 60, and 70 times earnings. And just as IBM fell off the wave, other companies did, too. Thus, a large investment disaster resulted from too high prices. And you’ve got to be aware of that danger.”

Today, dominant capitalization-weighted indices like the S&P 500 reinforce the idea to investors that the largest companies are currently driving most of the market’s returns. And, Wall Street is wasting no time herding as many investors as they can into the winning trade. At a time when single-day option trades and triple-levered ETFs based on a single company are common offerings, it makes sense to think about what kind of concentration we are seeing today.

Capital-Intensive Businesses

Much of the recent acceleration in earnings of the S&P 500 is due to one of the largest infrastructure spends in U.S. history. A recent study from Columbia Business School models that domestic capital expenditure on AI data centers, associated power generation, and specialized utility grids will reach $10.3 trillion between 2025 and 2032. This is larger than the Transcontinental Railway buildout (1860s-1880s), the Electrification and Grid buildout (1920s), the Interstate Highway System buildout (1950s-1960s), or the Telecom and Fiber Internet buildout (1990s). Another study from FactSet shows that capital expenditures across top hyperscalers are expected to reach over $690 billion in fiscal year 2026. This surpasses their organic operational cash flow, forcing legacy cash-rich tech companies into becoming major corporate bond issuers to fund data centers.

Munger viewed massive, competitive capital buildouts with profound skepticism. In capital-intensive industries, operating cash flow is an illusion because maintenance capital expenditure eats the economic surplus. If a company reports $5 billion in profit, but must immediately spend $6 billion on next-generation silicon, power substations, and cooling towers simply to remain competitive, the shareholder owns a depreciating machine, not a cash-compounding asset.

“There are two kinds of businesses: The first earns 12%, and you can take it out at the end of the year. The second earns 12%, but all the excess cash must be reinvested—there's never any cash. It reminds me of the guy who looks at all of his equipment and says, 'There's all of my profit.' We hate that kind of business.”

Munger also pointed out that when multiple well-funded competitors race for the same prize, they can create structural collective ruin: “When someone takes their existing business and tries to transform it into something else—or races into technology—they often fail.” He observed that when demand surges early in a buildout, shortages present themselves, prices surge, and margins look extraordinary. This leads to a race that often turns into a capital avalanche. In the telecom boom of the late 1990s, dozens of carriers simultaneously laid transcontinental and subsea fiberoptic cable, and each rationalized their billions in capex by assuming they would capture the dominant monopoly position. Much of the capacity they built was never used, and many of the carriers in the center of the fiberoptic buildup wound up bankrupt. The carriers that survived saw bandwidth prices drop by 90%. With so many hyperscalers investing to build out their own full-service AI platforms today, it is fair to wonder if we might be creating more capacity than we will ultimately need in this 
datacenter buildout—particularly considering how these technologies are very likely to become increasingly efficient over time.

Importantly, Munger saw infrastructure booms not as engines to drive investor wealth, but as engines of wealth transfer from capital providers to society at large. The racing and inefficient capital allocation leads to pricing shocks that ultimately drive broader market adoption, creating an economic benefit for a larger set of end customers.

The "Sudden Stop" 

While many on Wall Street today see leverage as a tool for financial optimization, Munger saw it more as an existential hazard. For him, the presence of debt strips away the single most valuable asset an investor or enterprise can possess: the luxury of time. An unleveraged company can watch its stock or its sales drop by half during a recession, freeze discretionary spending, and patiently wait for the cycle to turn. A company that has issued tens of billions in bonds to fund long-dated physical infrastructure must service the coupon and roll over the maturities—regardless of where revenues or interest rates stand.

The massive capital demands of the AI buildout have outpaced what hyperscalers can comfortably absorb on their primary balance sheets, driving credit risk into structured finance. Companies are, therefore, increasingly reliant on off-balance-sheet Special Purpose Vehicles (SPVs) that leverage project debt up to 90%. Meanwhile, specialty lenders issue billions in high-risk loans collateralized directly by rapidly depreciating GPUs. This systemic risk is further compounded by circular financing—where chip designers and cloud providers supply the very prepayments and capacity guarantees used to secure these borrowings.

This structure makes the AI database buildout vulnerable to the weaker links in what is becoming a very complex financing chain. By funding short-lived silicon with multi-year fixed liabilities, these structures could introduce severe collateral vulnerability if hardware pricing continues to fluctuate. Consequently, any downward revision in enterprise AI software revenue has the potential to trigger cascading collateral calls across the private debt market. This seizing of liquidity stops time and forces immediate price discovery—no matter what that price might be. Here’s how Munger phrased it: “It’s not the fall that kills you; it’s the sudden stop at the bottom.” 

*        *        *

Some rays of heat are beginning to be felt on Wall Street. Yields are rising across the globe and, over the last month, the ten-year U.S. Treasury yield has risen dramatically to levels not seen since 2007. Some of this increase in yields is in response to the strong economic numbers that the current AI growth cycle is driving. Some of it, however, is because governments, investors, and companies are all demanding leverage at the same time. There is simply more demand for debt than available supply. The squeeze of higher rates not only creates a headwind for economic activity going forward, but it also shortens the clock for many levered enterprises that were already under pressure. Steve Miller might add that, “Time keeps on slipping, slipping, slipping, into the future,” but for some that means their time is running out.

Looking Ahead 

As we move into the fourth quarter of 2026, earnings growth continues to be strong across a reasonably wide range of companies. This is true even in the face of rising input costs, and the monumental buildout of AI data centers across the globe surges on tirelessly. For some, this can lead to complacency. 

To us, this is a good time to be keeping balance in portfolios, reviewing historic market tolerances, and paying attention to our hot investment atmosphere. We believe that the current race to build out massive data center capacity introduces significant capital allocation and financing risks. We also see tremendous potential for significant productivity and efficiency advances in a number of industries as the next wave of technology expresses itself across the global economy in the coming years. And, we continue our focused work on portfolio companies we believe have the industry leadership and balance sheet strength to take advantage of AI and other new technologies as pricing, platforms, and products come into clearer focus. As always, we have our eyes set on long-term goals—while managing risk prudently along the way.

Print and Download the Newsletter Here