Dot.com bubble burst 25 years ago this month. Could AI be next?

San Francisco – A revolutionary new technology comes along and infatuates investors with its seemingly limitless possibilities. Euphoria sparks a stock market rally. Eventually, things get overheated and share prices become ridiculous. Then it all collapses.

Does this sound familiar?

It happened exactly 25 years ago when the roughly five-year dot.com bubble popped, leaving trillions of dollars of investment losses in its wake. On March 24, 2000, the S&P 500 Index posted a record level it would not see again until 2007. Three days later, the tech-heavy Nasdaq 100 Index also closed at an all-time high, the last time it would do that for more than 15 years. 

Echoes of that era are reverberating now. The technology this time is artificial intelligence (AI).

After a wild stock market rally that sent the S&P 500 soaring 72 per cent from its trough in October 2022 to its peak in February, adding more than US$22 trillion (S$29 trillion) of market value in the process, signs of trouble are emerging. Stocks are starting to sink, with the Nasdaq 100 losing more than 10 per cent to fall into a correction and the S&P 500 briefly dropping to that level. And the symmetry is raising frightening memories from a quarter century ago.

“Investors have two emotions: fear and greed,” said Mr Vinod Khosla, billionaire venture capitalist and co-founder of Khosla Ventures, who was a key rainmaker during the internet boom and remains one today. “I think we’ve moved from fear to greed. When you get greed, you get I would say indiscriminate valuations.”

Difference of degrees

The main difference between the dot.com and AI eras, however, is a matter of degrees. The most recent boom has been eye-popping, but it pales in comparison to the extremes of the internet bubble.

“The internet was such a big idea, had such a transformative impact on society, on business, on the world, that those who played it safe generally got left behind,” said Mr Steve Case, the former chairman and chief executive of America Online (AOL). “That leads to this kind of focus on massive investments to make sure you’re not left behind, some of which will work, many of which won’t work.”

Mr Case became a face of the dot.com boom when he bought Time Warner in January 2000, right at the peak of the bubble as AOL’s stock price was flying high. But the deal quickly went sour as the combination that made sense on paper did not in reality, something Mr Case himself has acknowledged. The combined AOL Time Warner’s shares tanked as earnings worsened, and the combination was finally unwound in 2009. 

This is the kind of thing Wall Street is worried about now. 

“There was a lot of hype around the internet, which materialised well before anybody had a business model for making money from the internet,” Nobel Prize-winning MIT economist Daron Acemoglu said. “That’s why you had the internet boom and the internet bust.”

The companies involved in the AI boom are also very different from the firms that dominated the dot.com era. The internet bubble was largely built on unprofitable start-up businesses, some of which capitalised on the mania by appending a “.com” to their names so they could sell stock to the public. The euphoria surrounding AI, on the other hand, is centered around a small group of tech companies that are among the most profitable and financially stable corporations in the world, like Alphabet, Amazon, Apple, Meta Platforms, Microsoft and Nvidia. 

For example, look at the amount of capital generated by today’s tech giants. This year alone, Alphabet, Amazon, Meta and Microsoft are expected to plow a combined US$300 billion into capital expenditures to develop their AI capabilities, according to analyst estimates compiled by Bloomberg. And even with all that spending, they are still projected to generate US$234 billion in combined free cash flow.

Burn rate

The dot.com boom had nothing like that, since it was more about speculative investments in emerging companies that had no profits. 

“You had a huge number of companies in the top 200 market cap that had a negative burn rate,” said billionaire investor Ken Fisher, chairman of Fisher Investments. “The thing that makes a bubble a bubble is the negative burn rate. Companies in 2000 were just accepted as good, there was a ‘it’s different this time’ mentality because of the internet.”

The differences between the companies driving the euphoria make it difficult to compare equity valuations between the two periods using traditional measures like price-to-earnings ratios. In 1999, the Nasdaq Composite Index, which housed most of the companies in the internet boom, saw its price-to-earnings ratio reach roughly 90 times, based on estimates at the time. Today, it is roughly 35.

But really, the entire concept of valuing share prices on price-to-earnings ratios seemed passe during the internet bubble because so many of the hottest companies were unprofitable. So Wall Street even invented new metrics, like “mouse clicks” and “eyeballs”, to try to measure their growth without involving money.

While that may seem crazy now, many investors at the time did not mind because they were betting on the limitless future baked into the internet’s growth assumptions. Plus, the stocks kept rising. 

Right call, wrong time

In time, a confluence of factors ended the dot.com bubble. The US Federal Reserve began raising interest rates aggressively, in part to slow the stock market’s exuberance. Meanwhile, Japan slumped into a recession, which raised fears of a global slowdown. With markets booming to all-time highs, investors suddenly turned sceptical about stocks that had no profits.

Said Mr Rob Arnott, founder and chairman of Research Affiliates: “Human beings are creatures of habit, and the embrace of the internet for most of us was gradual. Today, we use the internet for everything. Back in 2000, that wasn’t as true.”

But that does not make the roughly US$5 trillion dot.com wipeout any less painful. In so many ways, the internet turned out to be the right investment, but at the wrong time.

AI dreams

The risk for investors today is that scenario could be playing out again. AI inspires dreams of computerised personal assistants intertwined in every facet of our lives. The new technology will handle our transportation, help teach our children, provide routine medical care, create entertainment, and manage daily household errands and chores.

And all that may turn out to be true. But when it comes to tech-driven bubbles, the real winners are rarely revealed right away. The idea that the biggest beneficiaries from AI may not even exist yet is a dot.com lesson that Ms Julie Wainwright, former chief executive officer of dot.com era darling and now defunct Pets.com, thinks about a lot these days.

“All the innovation came from very small companies,” she said. “That may still be happening.”

The emergence of an advanced chatbot built in China called DeepSeek exposed this risk just a few months ago. The revelation sparked a US$589 billion rout in Nvidia’s shares on fears that cheaper AI models would sap demand for computing gear. For investors, it was a stark reminder that dominance in AI is far from assured.

This also explains why tech giants have been willing to pour so much money into developing the technology, even at the expense of hurting profit margins. The challenge is making sure they are constantly on top of a rapidly evolving revolutionary industry. And that is the price they are willing to pay.

“I’ve seen almost every phase of technology change since 1980,” said Mr Khosla, who made one of the shrewdest investments of the dot.com era in Juniper Networks and more recently was an early investor in OpenAI. “And I can’t see a change that’s larger in the past than AI.” BLOOMBERG