How the AI boom really compares to the dot-com bubble

How the AI boom really compares to the dot-com bubble

The dot-com boom of the mid to late 1990s was a wild mix of hype, easy money and genuine technological progress. It created the basic infrastructure for today’s online world, but when the bubble burst in 2000 it wiped out trillions of dollars in market value and pushed the broader economy into a brief recession. Many of the companies at the centre of that mania were fragile: young, unprofitable start-ups built more on promise than on sound business models, the New York Times reported.

The current artificial intelligence surge clearly echoes that earlier moment. Once again there is talk of a glorious technological future. Fortunes are being made quickly, including by people who were present and successful in the dot-com years. Start-ups with little track record are attracting lofty valuations. Yet beneath the surface, this boom is being shaped by a very different kind of financial and corporate muscle.

The most important difference is who is funding and controlling AI This time, the core technology is backed by multitrillion dollar giants such as Microsoft, Google, Meta, Amazon and Nvidia. These are businesses with established revenue streams and global franchises, not small firms a few bad quarters away from collapse. They are pouring hundreds of billions into data centres, chips and foundational models without having to slash their existing operations. Amazon is not selling fewer everyday products to pay for AI investments, and Google has not had to sacrifice its ad business to chase the next wave.

Regulation and politics are also aligned differently. In the 1990s the Clinton administration’s most notable technology move was an antitrust case against Microsoft. Today the Trump administration is broadly encouraging AI expansion, with relatively few regulatory barriers standing in the way. That supportive environment gives large players even more room to move.

As in the dot-com era, capital is heavily concentrated in one theme. Around two thirds of venture investment this year has gone into AI start-ups, a modern version of putting most of the eggs in a single basket. But the scale is completely different. At the height of the internet bubble, the most valuable enablers of the new economy, such as Cisco, Microsoft and Intel, were each worth roughly 500 billion dollars. Nvidia alone is now valued at more than 4.5 trillion dollars, and a cluster of major AI players is together worth more than the entire United States stock market in 2000.

Despite the familiar frenzy, this boom is less populist than the last one. The dot-com era drew thousands of hopefuls to San Francisco in search of start-up riches, and more than 2,000 internet companies went public in a few years. By contrast, AI is dominated by a small number of large firms and specialist labs. There are many companies using .ai domains, but only a fraction have scale, funding or credible products. The real action is concentrated in a relatively tight circle of corporations with deep expertise and deep pockets.

Supporters of the AI boom argue that the underlying technology is already delivering real value. In the 1990s, building online services was slow, expensive and constrained by immature software, unreliable bandwidth and a small internet population. Many dot-com businesses failed simply because the economics did not work. AI systems today can be deployed over a global, mature internet almost instantly. Investors and executives point to rapid adoption and strong revenue growth as proof that the technology is useful now, not just in some distant future.

Some analysts go further and say that concern about a bubble is actually a sign that markets have not yet reached the point of blind euphoria that usually precedes a crash. In their view, the fact that people keep asking whether this is a bubble shows that scepticism and caution are still present.

Even so, a familiar set of dangers lurks beneath the surface. Any period of fast growth and high valuations creates temptations to embellish numbers, overstate progress or structure deals in ways that confuse where the real risk lies. In the dot-com era, circular arrangements among media groups and internet companies were used to manufacture revenue. Today, complex webs of investment and partnership between leading AI firms are drawing increasing scrutiny from banks, regulators and investors who remember how the last cycle ended.

Ultimately, whether this boom ends in a gentle cooling or a sharp break will depend less on the slogans and more on the cash flows. The large AI players have profits and business models that their dot-com predecessors never did. At the same time, valuations in private markets are clearly stretched in some corners. Even enthusiastic backers acknowledge that some companies will run out of money before their products truly work. What seems certain is that Silicon Valley’s largest firms are better positioned than ever to survive whatever comes next, even if smaller AI hopefuls do not.