The AI Stock Bubble: Are We Reliving the Dot-Com Era?

A strange pattern starts to emerge on Wall Street late in the afternoon, as trading screens flicker between green and red. Every day, the same few companies—Nvidia, Microsoft, Meta, and Alphabet—seem to dominate the discourse. Over the past two years, their stock charts have risen sharply, dipped occasionally, and then risen again as if gravity were pulling them upward in reverse, giving them an almost theatrical appearance.

It’s difficult to avoid having a slight déjà vu feeling when watching the market these days.

It’s a loaded term that evokes the dot-com era of the late 1990s, when internet startups with dubious business models reached enormous valuations before collapsing spectacularly in 2000. Although the comparison isn’t perfect, there are enough parallels to cause people to think twice.

There’s a startling detail tucked away in the numbers. Approximately 30% of the S&P 500’s total value is currently held by a small number of tech giants, which is a remarkable concentration by historical standards. Approximately half of that share was held by the most powerful tech companies during the peak of the dot-com boom. History indicates that markets may become fragile when they get this concentrated.

However, the narrative is more complicated than just repeating 1999. Not obscure startups squandering venture capital in rented offices are driving the AI boom. Selling chips that run the world’s AI data centers, Nvidia is extremely profitable. While investing billions in artificial intelligence infrastructure, Microsoft continues to make money from enterprise software. Google and Meta continue to be enormously profitable advertising machines.

That distinction is important. Many of the priciest businesses made little to no money during the dot-com boom. Expectations drove nearly all of their stock prices. In contrast, today’s AI behemoths are enormous businesses already making real money. Because the companies themselves are unquestionably strong, investors appear to be willing to put up with high valuations.

However, even well-established businesses can become overpriced. Earlier this year, the market sent out a subtle but significant signal one afternoon. Microsoft’s earnings exceeded projections. That would normally cause the stock to rise. Rather, it fell precipitously the following day. Investors questioned whether the billions being spent on AI would truly yield significant returns, not Microsoft’s earnings.

Around the same time, Meta had the opposite reaction. The stock skyrocketed when executives persuasively claimed that AI was enhancing ad targeting and increasing revenue. It served as a reminder that markets now require evidence rather than merely assurances.

Huge infrastructure spending is another phenomenon that is subtly influencing the AI boom behind the scenes.

Large data centers are emerging from the earth like industrial cathedrals throughout the United States and parts of Europe. The scale is evident when you drive past some of them in Texas or Virginia. They are long concrete structures filled with processors meant to train machine-learning models, humming with cooling systems. Over the next ten years, tech companies are anticipated to invest hundreds of billions of dollars in this infrastructure.

Those who recall the telecom boom of the late 1990s will find that level of spending uncannily familiar. In anticipation of the surge in internet demand, businesses hurried to install fiber-optic cables across continents back then. Eventually, demand came in, but not fast enough to save many of the companies that constructed the networks.

AI may follow the same pattern. Some preliminary data have already raised questions. According to MIT research, roughly 95% of corporate AI initiatives had not yet produced noticeable financial returns. Although companies are actively experimenting, many still struggle to translate AI capabilities into quantifiable profits.

Bubbles frequently start to form in that space between excitement and actual revenue.

The financial framework for AI development is another subtle red flag. Circular financing arrangements, in which businesses invest in partners who subsequently use that money to purchase their products, are similar to some agreements between large technology companies. On paper, it appears to be growing quickly. However, the structure may begin to appear more brittle if actual customer demand does not materialize.

None of this guarantees a crash. In fact, many economists believe the result won’t be as dramatic as the dot-com bust. What analysts refer to as a selective correction may be a more likely scenario. While speculative startups stealthily vanish, the strongest businesses—those that are truly making money from AI—may continue to expand.

That pattern would resemble a gold rush more than a bubble bursting.

No matter which AI applications end up being successful, the companies selling the “shovels,” especially chipmakers like Nvidia, seem to be in a good position. In the meantime, thousands of smaller businesses are racing to develop AI tools for robotics, marketing, customer support, and coding, many of which might never turn a profit.

The size of the wager becomes clear when you stand outside a contemporary data center and hear the constant hum of cooling fans. In essence, trillions of dollars are betting that artificial intelligence will transform industries to the same extent that the internet did.

However, markets seldom move in a straight line, and enthusiasm frequently temporarily surpasses reality. As this develops, it seems like there is both real innovation and a good deal of speculative optimism in the AI boom. It’s still unclear whether that mixture will turn into a historic bubble or just the messy beginning of a technological revolution.

Wall Street screens continue to glow green for the time being. And investors keep buying.