AI fever vs dot-com bubble: WSJ finds 5 similarities and 2 differences

The return of network equipment maker Cisco Systems' shares to March 2000 price levels last week has prompted new comparisons of the current boom around artificial intelligence to the dot-com bubble of the late 1990s, The Wall Street Journal (WSJ) notes.

A number of key multiples - from forward-looking P/E (price to earnings per share ratio) and price to cash flow ratio, which compares stock and bond yields - indicate that the U.S. market looks the most expensive in the last 25 years, the newspaper writes. According to WSJ, a similar situation was already observed during the dot-com bubble: then, as now, investors put into quotations expectations of accelerated profit growth due to new technology. The newspaper highlights several similarities and differences between the current market and the late 1990s.

- Growth hopes. Just as in the dot-com era, when companies were valued "based on the hope that the Internet would usher in a new era of profits for business models not yet proven," today's market is doing the same with artificial intelligence, the WSJ writes. Generative AI has already empowered consumers with chatbots and imaging tools. However, such products are now being sold at prices far below the costs required to create them, resulting in large losses for AI businesses, the publication notes.

- Investment. 25 years ago, the Internet was built on a global network of fiber optic cables laid by telecommunications companies. This led to massive corporate spending financed by debt. The large language models underpinning advanced AI are being built on giant data centers, also leading to significant corporate investments that are increasingly being funded by borrowing as well as Big Tech cash flows from their traditional businesses, the WSJ points out.

In 2000, the scale of investment was enormous: in the late 1990s, more than $100 billion was invested in building new telecommunications networks. Fiber was so plentiful that much of it sat idle for nearly a decade, until Internet traffic grew enough to justify its use, the WSJ recalls.

The race to build data centers looks even more extreme, with leading AI developers operating with investment figures in the trillions of dollars.

The companies selling the "picks and shovels" for these booms are also performing outstandingly well, both now and in the late 1990s. In 2000, we were talking about Cisco, maker of the routers needed to connect the Internet, as well as telecom companies. Today, that role is played by Nvidia and other chip makers supplying computing power to data centers and making significant profits.

- Market imbalance. In 1999, in the S&P 500 Index, the number of falling stocks outnumbered the number of rising stocks. At that time, if a stock was related to the Internet sector, it was booming, and if it was not, there was almost no interest in it. Today, roughly the same situation is seen around artificial intelligence, the publication notes. Since the beginning of this year, 183 companies, or 37% of the U.S. broad market index, are in the negative. Everything related to AI - chip manufacturers, power generators, suppliers of equipment for building data centers - is growing, while much of the rest of the market is declining, the newspaper writes.

- Retail. Private investors are again playing a key role in the market and are actively buying shares of small companies, which are still mostly unprofitable. Such a situation was already observed during the dot-com bubble in 2000 and in 2021 - against the backdrop of the SPAC boom, when companies went public en masse through special structures and investors actively bought up risky assets. As a result, unprofitable small companies outperformed profitable but lower-quality securities, WSJ notes.

This trend is still reflected in the performance of small company indices today. In the 12 months to mid-October 2025, the Russell 2000 index, which includes small companies regardless of their financial performance, outperformed the S&P 600, which includes only profitable companies, by 10 percentage points. Previously, the Russell 2000's outperformance of the S&P 600 had been as marked only three times: in 1999-2000, in 2020-2021, and during the market's recovery from the low reached after the dot-com crash of 2002, the paper notes.

- Bubble talk. "Some people think that if everyone is talking about the bubble, then it doesn't really exist," writes the WSJ. In 1999, however, the bubble debate was in full swing, even as the bubble continued to inflate. The same thing is happening now, the publication notes.

- Revenue performance. Many "pure" dotcom companies 25 years ago had no revenue at all, whereas AI companies are at least already making sales.

- The magnitude of the stock price gains. Although Nvidia's stock has added 54% since its October peak and 30% since the beginning of 2025, with subsequent adjustments, these figures still look modest compared to the late 1990s surges, the WSJ writes. Cisco has more than doubled in a month - since mid-October 1999, the publication recalls. Apple added 150% that year, and Intel grew 75% in less than three months in early 2000.

Ultimately, the question of whether the current market situation is a bubble depends on whether it bursts, the WSJ points out. If it turns out that AI can provide neither the promised productivity growth nor high profits for its creators, this situation will be compared not with the AI boom, but with the painful consequences of the dot-com crash, the publication concludes.

This article was AI-translated and verified by a human editor