Dotcom Bubble

In the late 1990s, the internet became a mass medium for private individuals. Many investors believed that any company with an internet address would soon make enormous profits. They therefore bought shares in such companies at ever-higher prices. The “.com” ending in web addresses gave this era its name. Between 1995 and March 2000, the share prices of these companies multiplied many times over, even though many of them had never earned a single euro. Starting in March 2000, sentiment turned, prices collapsed, and hundreds of these companies disappeared within two years.

What the 2000 crash cost

The damage was greater than in a normal price decline. The US Nasdaq index, which lists mainly technology companies, lost around 78 percent of its value from its peak to autumn 2002. Several trillion dollars evaporated on paper. Anyone who bought at the peak in 2000 had to wait until 2015 for the index to reach that level again.

It wasn’t just professionals who were affected. In Germany, millions of small investors bought shares for the first time, often on the so-called Neuer Markt, a separate stock exchange segment for young technology companies. This segment lost more than 95 percent and was shut down completely in 2003. Many people avoided stocks for decades afterward.

But the other half of the story matters too. The internet was not an illusion — it actually transformed the economy. Amazon fell by more than 90 percent and is today one of the most valuable companies in the world. The technology was right; only the timing and the prices were wrong.

How a bubble inflates

A speculative bubble arises when prices are no longer tied to a company’s actual earning power but to the expectation of being able to resell it later at a higher price. As long as enough buyers keep joining in, this expectation confirms itself. Rising prices are then taken as proof that the thing works. That attracts even more buyers, and the cycle spins faster.

Several amplifiers came together during the dotcom era. Money was cheap to borrow due to low interest rates. Companies went public just months after being founded. And because profits were lacking, substitute metrics were invented, such as the number of page views per month. This made it possible to show growth even when no money was coming in.

The end rarely comes from a single event. It’s enough for the supply of new buyers to dry up. Then prices fall, expectations reverse, and everyone wants to sell at the same time. Companies that lived off fresh investor money rather than customers then had no financing left and went bankrupt.

The recurring comparison in the AI debate

The term appears in economic news almost daily today, usually as a question: Is the AI euphoria a new dotcom bubble? What’s meant is the sharply risen share prices of companies like Nvidia, which makes chips for AI data centers, and the billions being invested in new data centers. Critics point to start-ups with high valuations and low revenues. That reminds them of 1999.

The other side points out that today’s major AI buyers are highly profitable corporations. Microsoft, Google, and Amazon pay for their data centers out of ongoing profits, not with money borrowed by stock market newcomers. That is a real difference from 1999. Anyone asking the bubble question should therefore look not only at prices but also at where the invested money is coming from.

A common mistake is to confuse a bubble with a fraud. In a bubble, prices are inflated, but the underlying technology can still be real and important. That’s exactly what makes the comparison so uncomfortable for investors: it says nothing about whether AI will succeed, only whether today’s prices match that success.