“Speculative bubbles are often associated with a sort of social contagion: stories of easy fortunes attract public attention, and then fuel further price increases as more and more investors, many with little knowledge of fundamentals, rush to participate.”
(Shiller, Irrational Exuberance)
The word “bubble” has become a kind of financial clickbait, thrown around in both social and traditional media to explain every market rally without any precision. But true bubbles, like 1929 or 2000, were defined by broad participation and euphoria, when households across the spectrum felt the lift. I remember just coming back from Brazil and its depression post devaluation and many of my Morgan Stanely colleagues were bragging about their internet longs and my high school friends were buying houses with AOL stock being collateral. In those moments, everyone was benefiting: public mood was high, job confidence was strong, and speculative enthusiasm spread widely. Equally important was not having corrections creating the original buy the dip mentality. From 1991 to 2000, there was only one drawdown of at least 15% in the SPX and that was short lived for the LTCM fall that led to a sharper rally into the bubble before we finally saw a prolonged bear market. We have seen three 20+% falls in the last five years. Concentration for me is less stable and that means more, not less, scary falls whenever the AI fears spread.
So that doesn’t mean today is healthy, it makes it different. What we are seeing is not a bubble but a concentrated market. That concentration extends to the consumer which is driven by a K-shaped economy where the top 1% own roughly a third of all assets while the bottom 50% owns an astonishing 2.5%. We hear these numbers but think about that ratio and wonder why we are not in a revolution already. Consumer sentiment which is a survey of everyone sits near historic lows, anger at politicians simmers globally, and workers are more worried about losing jobs than finding new ones. If stock market retail participation was broad based it would show up in sentiment like past examples. This is not the broad mania of a bubble; it is the imbalance of narrow concentration.
Fear of bubbles has already proven costly. Many investors have sat on the sidelines this year, spooked not only by tariff fear but also by the constant warnings of an AI bubble, only to miss what is a megatrend. Unlike past manias, this surge is not merely about corporate profits or IPO fueled speculation but about a global race. The United States and China see artificial intelligence as a matter of national security, ensuring sustained investment regardless of sentiment. At the same time, mega-cap winners of the prior decade are racing to avoid obsolescence in a world where innovation cycles compress like new model releases of language models, faster, sharper, and more unforgiving. And governments, weighed down by record debt and confronted with widening inequality, are choosing growth over austerity through fiscal dominance in the hope that AI might ultimately provide solutions. Far from a bubble destined to pop, AI sits at the intersection of security, survival, and systemic necessity.
“What distinguishes a true bubble is the popular participation: when speculation ceases to be confined to professionals and instead grips society at large.”
(Edward Chancellor, Devil Take the Hindmost, 1999)
In past bubbles, the broad-based nature of the euphoria showed up not only in markets but in the real economy. The dot-com period is the clearest modern example. In January 2000, consumer confidence as measured by the University of Michigan reached its highest level on record, reflecting a public mood of optimism and abundance. Auto sales provide another striking marker: in the first quarter of 2000, the U.S. recorded the highest five-month average of seasonally adjusted annualized sales in history. These were not isolated signals; they captured a time when households felt flush, job prospects looked secure, and spending surged. The economy was expanding in ways that touched nearly everyone, and speculation spread well beyond Wall Street. That was the definition of a bubble: broad enthusiasm that spilled into everyday life.
“Men, it has been well said, think in herds; it will be seen that they go mad in herds, while they only recover their senses slowly, and one by one.”
(Charles Mackay, Extraordinary Popular Delusions and the Madness of Crowds, 1841)
Today’s market looks very different. Instead of widespread consumer exuberance, the gains are concentrated in a narrow set of companies tied to artificial intelligence. Michael Cembalest’s recent “Blob” report highlights how just a handful of mega-cap firms, NVIDIA, Microsoft, Apple, Alphabet, Amazon, and a few others account for the bulk of equity market performance. Corporate investment is similarly skewed, with capital expenditures and revenue growth dominated by AI-related spending in semiconductors, cloud infrastructure, and data centers. These same companies, however, were also at the center of what some called the “last bubble” during the 2010s, when software and platform dominance reshaped the economy. Rather than bursting, that dominance entrenched itself and widened the wealth divide further. What we are experiencing now is not a broad speculative mania, but a continuation of that concentration, only this time powered by AI.
“The mass escape into make-believe, so much a part of the true speculative episode, reached its apogee in 1929, when the great and the humble alike took leave of reality in the expectation of fabulous gains.”
(John Kenneth Galbraith, The Great Crash of 1929, 1955)
This concentration has also produced a K-shaped economy, where asset owners thrive while much of the population feels left behind. In past bubbles, jobs and wages rose alongside markets, giving workers a sense that they were part of the boom. Today, the opposite dynamic is unfolding. For the first time, we are witnessing GDP growth accelerating while job creation falters, a break in the historical link between output and employment. Surveys capture this divergence clearly the New York Fed recently reported the lowest ever probability that workers believe they could find a job within three months if they lost their current one. In effect, the economy is expanding on the back of AI-driven productivity and capital spending, but the labor market is not keeping pace. That disconnect fuels the anger toward politicians, the malaise in consumer confidence, and the sense that growth is happening somewhere else, in markets and corporate balance sheets, not in people’s lives. This is all happening before the true job disruption from AI begins with the rise of the agentic world, robotaxis and humanoids.
“In euphoric booms, speculation becomes the dominant source of profits and households as well as businesses are drawn into the expectation of ever-rising asset values.”
(Hyman Minsky, Stabilizing an Unstable Economy, 1986)
The contrast in consumer confidence underscores just how different today’s cycle is from past bubbles. At the peak of the dot-com frenzy, households dominated by the baby boomers felt secure in their jobs, flush with paper wealth, and willing to spend. Today, by contrast, consumer confidence, especially for the younger generation, sits near historic lows despite equity markets trading at record highs. Instead of optimism, surveys capture anxiety over job security, the cost of living, and the perception that the economic system is rigged to benefit only a narrow slice of society. The current mayoral race in New York is a reminder of how deep this has become. In a true bubble, the masses are swept up in collective enthusiasm; in this cycle, the public mood is one of skepticism and unease. This mismatch between soaring markets and sour sentiment is another reason why “bubble” is the wrong frame for what we are experiencing.
“As firms or households see others making profits from speculative purchases and resales, they tend to follow. When the number of firms and households indulging in these practices grows larger, and the belief spreads that the object of speculation can only go up, speculation for profit leads away from normal, rational behavior to what have been described as ‘manias’ or ‘bubbles.’”
(Charles Kindleberger, Manias, Panics, and Crashes)
When viewed through Kindleberger’s framework, the distinction becomes clear. A bubble requires breadth, households, firms, and entire sectors piling in, driven by the belief that prices can only go higher. That is what defined 1929 and 2000, when speculation spread from trading desks to living rooms and consumer exuberance reinforced the cycle. Today’s markets show the opposite. Gains are concentrated in a handful of firms tied to artificial intelligence, while the broader economy reveals weakness in jobs, wages, and sentiment. Far from the speculative fever Kindleberger described, this is a narrow dependence on corporate capex and productivity improvements that most households cannot access. In other words, the problem is not excessive mania but excessive concentration and that distinction matters.
“Speculative bubbles are fueled by the contagious spread of market stories, which inspire enthusiasm and a desire to participate among a growing number of investors. But just as the stories can drive prices up, the eventual disillusionment brings them back down.”
(Shiller, Irrational Exuberance)
To separate AI from the lazy shorthand of “bubble,” we have to recognize that what’s unfolding is not speculative mania but a structural arms race. At one level, it is geopolitical: the United States and China are pouring resources into artificial intelligence as a matter of national security, ensuring that capital will keep flowing regardless of sentiment surveys. At another level, it is existential for the companies themselves, many of which see AGI as both an opportunity and a risk, a god they are racing to build, even if it means cannibalizing their own business models. That does not mean there won’t be massive losers. Just as once-dominant firms like GE, Ford, AOL, and BlackBerry eventually fell to faster-moving competitors, many of today’s leaders will be displaced. The difference now is speed. In the age of AI, obsolescence arrives like a new language model release, faster, sharper, and more unforgiving. This is not a bubble but an extension of the concentration and disruption that has defined modern capitalism: innovation compresses the distribution of wealth across households, across companies, and increasingly across nations. AI accelerates that process, amplifying both the imbalance and the opportunity. Bubbles lift everyone into euphoria; concentration leaves most behind.
