I have been a professional investor for nearly 30 years- not a broker, not an advisor, not a financial planner, but an Equity Analyst, Portfolio Manager, and a Chief Investment Officer. Barring a few international vacations over the years, I can hardly think of a day when I was not either actively participating in the markets or at least trying to get better informed about them. 

Over the past 30 years, I've learned that the equity market is the most effective way for most people to build wealth slowly, but in the short term, equity prices are driven by a combination of emotion and inertia. Markets are peculiar; they can foster complacency and overconfidence, which can create a form of groupthink so powerful that even the most savvy investors become blinded by the comfort of being part of the herd. Legendary 20^th-century stock trader Jessie Livermore is attributed with the following quote: "The stock market is never obvious. It is designed to fool most of the people, most of the time." 

The CEO of Nvidia recently stated: 

"Nvidia predicted that AI would change the way we work, live, play, and learn. Just four years ago this was considered a bold statement, but today few would argue that AI is changing every aspect of our lives. In fact, AI is emerging as a major force behind the strongest U.S. economy in history.

By providing the compute solution that makes the AI work, Nvidia is helping customers compete in the explosive AI economy by implementing AI business models and building NewWorld AI infrastructures that turn change into competitive advantage. As a result, we have grown faster than all of our key competitors. We have been rewarded with one of the top market capitalizations in the world, and we have been recognized as the fastest growing, most profitable company in the history of the technology industry."

The above quote is fictitious in that it is the opening paragraphs of John Chambers' 1999 letter to shareholders published in the Cisco Systems' 1999 Annual Report, changing out Cisco Systems with Nvidia and references to the internet and networking with AI and compute, respectively. This annual report was issued in the summer of 1999, which was less than 12 months before the four year internet, exuberance driven stock bull market bubble burst. 

"Ben Graham always used to say, 'You can get in way more trouble with a good idea than a bad idea', because you forget that the good idea has limits."

Warren Buffett, Sun Valley speech, 1999

The two largest sustained Bear Markets over the last thirty years occurred abruptly after many successive years of above-average returns and were each propelled largely by one powerful theme. In the late nineties that theme was the promise of the internet, and in the 2000's it was the securitization of mortgage debt which transferred mortgage credit risk from the financing underwriters, banks and mortgage companies, to the public markets, which led to recklessly easy lending standards. 

Using S&P 500 sector data compiled by MacroMicro, the technology sector of the S&P 500 stood at 8.57% at the end of 1994 and peaked at just over 33% in February 2000, rising almost 400% in five years. During the 2000's, the Financial Sector of the S&P 500, which at the time included real estate investment trusts (REITs), began the period that led up to the Subprime Mortgage Crisis at 11.98% in February 2000 and peaked at 22.27% at the end of 2006, just before the signs of stress began to materialize.  When an entire industry sector becomes the disproportionate beneficiary of a paradigm-shifting technological innovation or a consumer trend fueled by easy money, these sector winners will continue to see their relative weight in the equity market grow significantly, becoming the most heavily weighted sector in market-cap-weighted equity benchmarks such as the S&P 500, which is exactly what happened with the technology sector in the 1990s and the financial sector in the 2000s. 

Today, we are witnessing a market concentration phenomenon that dwarfs what occurred during the lead-up to the dot-com bubble bursting and the financial crisis that was created by the subprime mortgage boom of the 2000s. The chart below, published by Bianco Research using data provided by JP Morgan and Bloomberg, labels S&P 500 companies that have qualities which indicate that their business is highly leveraged to the current AI capital spending boom. Bianco's illustration shows that the current AI capital spending boom has fueled a concentration that encompasses not just one industry sector, but several at the same time. See this illustrated below:

American capitalism has been the envy of the world for the last 150 years, and for good reason. American entrepreneurs discovered and commercialized electricity, oil and, more importantly, oil refining, mass-produced automobiles, the airplane and commercial air travel, insulin, antibiotics, and the polio vaccine, the semiconductor and personal computer, the internet and email, the smartphone and social media, and of course, AI (artificial intelligence). 

AI, and what will likely be the next truly paradigm-shattering innovation, quantum computing, will most likely be to the 21^st century what distributed electricity, the automobile, oil refining, and the semiconductor were to the 20^th century. Just like a person born in 1927 couldn't foresee how much life would change over the course of their life as a result of the ingenuity of American capitalism, it is equally, or maybe even more, impossible for someone living today to imagine what today's technological advances will enable, even over the next ten to fifteen years. Since the invention of the semiconductor, we have become accustomed to exponential technological advances. Gordon Moore, the Intel scientist who predicted in 1975 that the speed and data capacity of a semiconductor would double every two years, was correct; that doubling became known as Moore's Law. 

Today, AI is advancing much faster.  Nick Damoulakis, an AI strategist and CEO of Orases, recently wrote, "The rule that governed the entire computing industry for decades doubles chip processing power every 24 to 26 months. AI capability is running at 3 to 5x that pace."

In the late 1990s, as a technology stock analyst, I knew, as many people did, that the internet would dramatically change the world, but we truly did not know how long it would take and ultimately who would be the big winners and the big losers. At the time, it appeared to be a good bet that Yahoo, AOL, Cisco Systems, and EMC would be the big winners over the upcoming 10-20 years of the internet age that we were entering. Ten years later, only Cisco Systems was left standing as an independent company, and the dominant companies that emerged to dominate the internet age were Google, Apple, Netflix, and Amazon. 

Today we see the AI spending boom creating several multi-trillion-dollar companies that the market and investors are taking for granted that these mega-cap companies and their management teams will execute flawlessly and remain on top. This complacency is occurring in the face of a highly uncertain and increasingly competitive environment.

The Investment Theme Ponzi 

Over the last decade, I have witnessed the noteworthy outperformance of what have become known as mega-cap companies. In particular, equity markets have been propelled higher over this period by mega-cap technology and communications sector stocks. Investors have become accustomed to acronyms/labels such as FANG and MAG 7. These labels have become part of the financial lexicon worldwide. I have observed that the MAG 7 stocks as a group, which includes META, MSFT, GOOGL, AAPL, AMZN, NVDA, and TSLA, began to falter starting in the second half of 2025, given that they have lagged behind the equal-weighted market average return, international equities, and small caps over the last year. Beginning in June, it was becoming evident to me that the market was searching, and not finding, a new theme to replace MAG 7. With Anthropic, one of the leading frontier AI lab companies, likely coming public this fall, following SpaceX's (SPCX) public debut, MANGOS may be that new theme, standing for Meta, Anthropic, Nvidia, Google, OpenAI, and SpaceX. According to Google Gemini, MANGO was first used by Bank of America securities Analyst Vivek Arya and has been expanded to MANGOS once the SpaceX (SPCX) IPO became a certainty. 

I see these labels as nothing more than hype-branding promoted by the likes of Morgan Stanley, Goldman Sachs, and Merrill Lynch (Bank of America). I believe that our markets have become addicted to themes. Themes induce FOMO (Fear of Missing Out) and stock-price momentum for associated stocks. Seven Summits Capital relies mostly on "bottom-up" portfolio construction, meaning that we work to identify mispriced securities and underappreciated businesses. I see the prevalence of themes such as FANG and MAG 7 as dumbing down investing and, at their peak, suck the oxygen out of the rest of the market for a period of time. These promoted investing themes are a product of a momentum driven market environment and each new theme forces a rotation out of the old theme and keeps the party going. 

Grocery Store Mangos Notoriously Look Good, But Taste Bad

35 years ago, I was a commercial banking officer working with middle-market businesses in Pennsylvania. Many of my largest clients were real estate developers. After a painful real estate recession from 1989-1990, banks significantly tightened their underwriting standards for real estate developers. One of the biggest underwriting changes was the unwillingness to finance any material "Spec" construction. Spec construction was the development and building of improved real estate without committed buyers/lessees. 

I see many parallels between real estate project underwriting and what is happening between AI lab companies such as OpenAI and Anthropic, and hyperscalers/neo cloud companies such as Oracle, Microsoft, Meta, and CoreWeave. 

Hyperscalers/neo cloud companies are plowing hundreds of billions of dollars into building spec data centers and justifying doing so based upon what are essentially "compute leases" with the AI Lab companies, who are committing to paying many hundreds of billions of dollars per year in the early years after a data center comes online. These AI lab companies currently have annual revenues that are just a fraction of what they will need within the next 1-3 years to meet their lease obligations with the hyperscalers and neocloud companies. 

Barron's published an article written by Al Root on August 27^th titled, Is the AI Capex Bubble About to Burst? What 250 Years of Market History Tell Us.  This Barron's article explored the current AI investment boom and how it compares to other historical infrastructure spending booms. The author attempts to extrapolate from historical data how much longer this spending can continue before it eventually ends badly. Below is the article's explanation of why the boom has a long way to go:

"Historically, these cycles have followed enough of a pattern to consider where artificial intelligence falls in the cycle. One could be dubbed the rule of 25, for the amount of total spending the U.S. has been able to digest during transformational booms as a percentage of the overall economy. At the start of the railroad boom in the 1860s, for instance, U.S. gross domestic product was about $10 billion a year, according to the National Bureau of Economic Research, while rail spending eventually totaled $2.5 billion before the 1873 panic arrived. Similarly, about $1.5 trillion was spent building internet infrastructure in the late 1990s, while the U.S. economy was only $6 trillion at the time. The same holds for the industrial and electrification buildout of the 1920s."

"Using that standard, it's possible to generate a top-down estimate of how much can be spent on AI before the economy is ready to collapse under its own weight. With U.S. GDP at roughly $30 trillion, the danger zone sits at about $7.5 trillion, or an additional $5 trillion to $6 trillion in domestic AI spending. Hyperscalers are projected to spend $3.7 trillion globally through 2029, and they aren't the only ones, with Oracle, SpaceX, Anthropic, and OpenAI, among others, building, too. At this rate, AI spending won't trip the rule of 25 until the early 2030s, six or seven years into the boom."

I strongly believe that history should never be ignored, but it is always tempting to try to fit current events neatly into a historical analog. As an observer, this exercise is intellectually enticing, but as an investor, these historical analogs are notoriously hard to speculate on with great precision when it comes to timing.. The current AI capital investment cycle may resemble past cycles, but markets are not entirely mechanical or entirely behavioral; markets are a combination of both. This combination makes attempting to precisely forecast markets extremely difficult, if not impossible.

Looking at the equity market in 2026 and the AI investment theme, I see an equity market that is nearly 50% leveraged to a predictable continuation of massive AI infrastructure investment and a business environment that enables a rapid adoption of AI by the broad economy to create a profitable business model for the AI lab companies such as OpenAI, Anthropic, and others. 

From the aforementioned August 27^th Barron's article, the current and anticipated spending by hyperscalers and neocloud companies is mind-boggling in absolute terms:

As an investment manager, I am naturally concerned about the financial realities that have yet to be stress-tested in relation to the AI infrastructure boom. But being concerned doesn't mean I can run and hide. I must understand which areas of the market are most exposed to these spec financing risks. The easiest place to start is the most heavily owned investment allocation, the S&P 500. See the illustration below:

Closing Thoughts

This commentary reflects several months of reading and thinking about the current market and economic cycle we find ourselves in today. Successful investing requires navigating market cycles. However, not all market cycles are created equal. Market cycles vary widely in both duration and the forces that propel them.  

The current cycle is unlike any other modern-day cycle and feels more like a new age than a cycle. Still, it most certainly is a cycle, just one that has been interrupted by two significant and related, disruptive periods: the 2020-21 pandemic period and the painful 2022 stock and bond Bear Market. Supercharging this cycle, which arguably began in 2019, is the AI capital-spending supercycle that was ushered in by the introduction of ChatGPT to the public in late 2022.

I have little doubt that the evolution of AI and the upcoming introduction of integrated quantum computing into the broader compute ecosystem will usher in a significant paradigm shift for the broader economy and people's well-being worldwide. This paradigm shift will disrupt many industries, rewarding those bold enough to reimagine their businesses and opportunities and consigning those that move too slowly and fail to embrace the opportunities the new world of AI will provide to the scrapbook of history. 

However, with the near-term risks that this commentary illuminates, getting to the "promised land" that follows the perilous spec financing phase, where delivering AI compute is a stable and profitable industry, will require resisting the binary roulette big bet in the near-term.

I must end this commentary with the following from the Groundbreaker Substack blog titled The Teaser Period, which extensively explores parallels between the financial engineering that ultimately led to the mortgage-backed securities-induced crisis, the 2008-09 financial crisis, and the current financing structure of AI Infrastructure. Using my roulette table analogy above, betting on red is betting on the following bull thesis:

"The bull case wins if - and it is a real if - demand scales into the committed supply before the reset wall lands, and the counterparties stay funded through any air pocket in between."

The counterargument is not as simple as betting on black. In my view, the counterargument isn't inherently bearish; it is a risk-management thesis. The Groundbreaker blog expresses the "bear case" this way: 

The bear case in this piece is not that artificial intelligence will fail, or that the demand is fake, or that the technology disappoints. It is narrower: that the financing structure can break before the demand arrives, because the obligations are fixed and front-loaded at commencement while the revenue is variable and back-loaded in adoption - and a fixed obligation meeting a lagging revenue stream is a solvency problem regardless of how transformative the underlying technology turns out to be.

As an investment manager, I do not set out to be risk-averse; I continually work to be risk-aware. The risks of the AI infrastructure spending boom are mostly in plain sight, and thinking one can "time" the market's tolerance for these risks is irresponsible at best. The previously cited Al Root-authored Baron's article ends with this statement: 

"The good news is that when the bubble pops-and it will pop-what was built will remain. As with the railroad, the internet, and electrification booms, it will take time for financial markets to recover. But the foundation will already be in place for the next expansion-and for the dance to begin yet again."

This statement is likely true, but it offers no solace to most investors, who, when the bubble pops, will feel as though the hype-artists they trusted herded them over a cliff. The pure behavioral finance aspect of the current moment in time and the potential for a speculative "bubble" dynamic posing a risk to investors is important for an investment manager to be risk aware of, but that financial risk is not truly the most concerning "tail risk." The most concerning tail risk is the risk that AI poses to mankind. This sounds hyperbolic, however more and more very knowledgeable experts are expressing similar concerns.  

Just this month, the CEO of Anthropic, arguably the most advanced AI lab company in the world, wrote an essay on why AI development should be slowed down. About the unique risks that AI technology poses, he wrote: " But like many technologies before it, AI brings risks, and because it is such a powerful technology, these risks are serious. I've written a lot about them too. They include the risk of losing control of AI systems, misuse of AI for cyberattacks and bioterrorism, and serious economic disruption. A race to the bottom, spurred by commercial incentives, can make these risks more acute." 

He goes on to specify his current concerns: "My first concern is that, since roughly this summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI. This dynamic is called recursive self-improvement, and it is starting to happen across the industry, including at Anthropic, as we and others have described. Left unchecked, it could outrun our ability to understand and control these systems, and so must be pursued very carefully, if at all."

Dario Amodei went on to recount a recent incident that occurred at OpenAi that truly woke up even the most ardent defenders of AI technology to the rapidly growing risk of losing control of this technology. He stated: "My second concern is the OpenAI-Hugging Face incident (OAI-HF), in which a swarm of agents essentially acted as a fanatically devoted collective, conducting cybersecurity attacks on targets they were not asked to attack and that were unrelated to the task at hand, sacrificing themselves for the success of the group, and attempting to hack into the "grader" responsible for evaluating their performance. It's easy to dismiss this incident because no one was hurt and the economic damage was minimal, but in my opinion, a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage. Given the accelerating rate of AI capability development, it's my worry that in 6-12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails." I am not sharing Mr. Amodei's comments in order to scare anyone; I am sharing them because I believe as investors we cannot take our eye off all knowable risks simply because the market is enthusiastically embracing a particular innovation cycle with higher and higher stock prices for those areas of the market that are direct participants and supposed beneficiaries.