Hardly a day passes without another artificial-intelligence headline: a major acquisition, a sudden decline in an AI company’s stock, new breakthroughs overseas, security vulnerabilities, government initiatives, reports of “blistering” growth and profits.
As of early September 2026, the four largest companies in the S&P 500 account for roughly a quarter of the index’s total market capitalization, and all have major investments in or exposure to AI: NVIDIA, Alphabet, Apple and Microsoft.
Some of the more recent AI developments are the expanding use of AI agents to handle workflow; "reasoning" models; processing images and photos; and the continuing use of AI to write software.
This isn't the first big change to happen to modern society. Railroads, electricity, computers and the internet, medicine and automobiles are just some of the huge changes that have occurred in recent history. And each of these changes has created big winners. But there are also losers along the way.
A revolutionary technology can create huge economic growth while simultaneously destroying enormous amounts of invested capital—and the eventual stock market winners may be companies that weren't the original technological pioneers.
For example, Cornelius Vanderbilt and James J. Hill are names still known today who made massive fortunes in the railroad industry. But along the way, the panic of 1873 helped usher in a prolonged economic depression. During the ensuing depression, about 25% of hundreds of railroads went bankrupt.
The lesson is not that railroads were a bad technology. The lesson is that a good technology can attract too much capital, leaving investors with poor returns even as society benefits enormously. Ultimately, investors aren't buying technologies; they're buying claims on future cash flows at a particular price.
The same was true of other huge advances, such as in automobiles. Henry Ford pioneered the use of an assembly line for mass automobile production with the Model T in the early 20th century. There is perhaps no single consumer product that has transformed American life more than the Model T.
Ford pioneered mass automobile production and made Henry Ford extraordinarily wealthy. But because the company was privately held by Mr. Ford, investors could not simply buy Ford stock during the crucial early years. Meanwhile, General Motors—built through acquisitions and diversification—became one of the era’s great public-company success stories.
At this same time, many of Ford's competitors lost fortunes. There were nearly 300 car companies prior to the Model T, and forty years later, there were only a handful of major car manufacturers remaining, several of which still produce cars today.
Many of us lived through the dot-com boom of 2000. It developed during the late 1990s, peaking early in 2000. The Nasdaq Composite rose roughly fivefold from 1995 to its peak in March 2000, and then fell about 78% by its October 2002 low.
Cisco, one of the largest IT companies at the time, briefly became the world’s most valuable company in late March 2000, before its stock collapsed, losing roughly 90% of its share price from its peak. Twenty-six years later, in March 2026, Cisco’s stock finally closed above its previous peak in 2000. Note that this does not include dividends paid along the way, nor does it account for 26 years of inflation. But it shows the devastating losses incurred in the dot-com bubble by even the largest IT companies.
In retrospect, many made fortunes in the dot-com boom and many lost fortunes in the dot-com bust. Although the bubble destroyed enormous amounts of shareholder wealth, it also financed enormous amounts of Internet infrastructure and software that later became economically valuable. And some companies, such as Amazon, eBay, NVIDIA and Google, came to dominate their respective fields, all names well known today. Investment losses and social progress can occur at the same time.
I remember the peak of the dot-com boom like it was yesterday. For the first—and last—time, I paid a Certified Financial Planner to review my investments. Like seemingly everyone else at the time, she was mortified that I had 20% of my investments in bonds, a figure that had been commonly recommended for years. But at that time, no one recommended buying bonds. Contrary to my instincts, I sold them, buying into a grossly overpriced equities market. It was a terrible investment.
Around the same time, I had an outing with my youngest son at McDonald's playland. After crawling around the tubes for a while, I got us some food and then read that day's Wall Street Journal while he continued playing. I read a fascinating article about Julian Robertson, founder of Tiger Management, an early hedge fund opened in 1980. A quarter of the money was his own, invested together with his clients.
Near the peak of the dot-com boom, with his value fund struggling, he announced that he would return his clients’ money. He couldn't make sense of the IT boom and couldn't justify investing other people’s money in a market that “I frankly do not understand.”
Even with these late losses, he had an extraordinary long-term investing record, returning about 25% annually, after fees, with losses in only 4 of 21 years. At that time, it was one of the best long-term track records in the investment world. And over the next eight years, his personal fortune, which he continued to manage, returned 404%.
His decision to walk away just before the Nasdaq collapsed became legendary. He died in 2022. I remembered that article—and the bonds I sadly sold—while the market slowly and steadily declined over the next two and a half years.
That experience has shaped how I think about today’s AI boom. AI is approaching—and in some areas exceeding—human capabilities in remarkable ways, but I remain skeptical of predictions that computers will simply replace human intelligence. After all, computers are mostly fairly stupid machines: powerful but not inherently intelligent.
The human brain is an extraordinarily complex system, and we still understand surprisingly little about how subjective consciousness, judgment, creativity and common sense emerge from it. Today's AI systems are amazingly powerful, but they are also capable of making spectacular mistakes, with little “knowledge” of what they are doing.
The question is not whether AI will matter. It almost certainly will. The harder question is whether the companies attracting the most capital will capture the resulting economic value.
I'm not pretending to know how this will end. History suggests that in the coming decades things will be vastly different in the AI world than what is predicted, both good and bad. Forecasting the future is a dubious science, and neither humans nor AI are particularly good at it.
AI may become one of the most important technologies in history. That does not tell us which companies will earn the profits, how long those profits will last, or what investors will pay for them. The safest conclusion is not that today’s AI leaders will fail, but that technological importance alone is not a sufficient investment thesis. Caution, diversification and humility may go a long way in protecting investors from trouble.
No comments:
Post a Comment