The dot-com bubble explained simply, with the AI bubble as a mirror

What is the dot-com bubble, explained simply in one minute?

Picture a new shop street that everyone expects to fill with customers. Builders rush to construct shops, investors pay high prices for the first tenants, and newspapers print the success stories. The customers arrive more slowly than the prices assumed. Rents fall, tenants leave, and prices drop. The dot-com bubble followed that path, with the internet as the street and the Nasdaq as the price tag.

The Nasdaq Composite peaked at 5,048.62 on 10 March 2000, and the NASDAQ-100 fell 78% to its low on 9 October 2002 (Wikipedia, Dot-com bubble). The technology stayed in use while the prices reset. That split between a lasting technology and a temporary price level is the single idea to keep from this page.

What promise started the dot-com bubble, and what promise drives the AI bubble?

In 1995 Amazon.com and eBay launched, and over 40 million people used the internet that year (Wikipedia, Dot-com bubble). The promise was simple: every business would move online, and early movers would win. Investors bought the promise before the revenue existed, because waiting looked more costly than paying early.

The AI promise has the same form: generative AI will run through every company. The EQOS Asia AI study opens with that picture, then adds a measured result, 95% of deployed projects failing in production (MIT 2025). A promise and a result can both be true. The gap between them is where the price lives, and it is what the rest of this page watches.

How did money flow into the dot-com bubble, and how does it flow into AI?

Money arrived by 3 doors. Telecom equipment firms spent more than $500 billion in the 5 years after the Telecommunications Act of 1996, mostly borrowed. Venture capital was easy to raise. And listings let early backers sell to the public (Wikipedia, Dot-com bubble). A simple way to see it: lenders, venture funds, and public buyers each took a turn at paying.

For AI, the study records about $200B a year spent on research and deployment. The simple question is the same: who pays? When customers pay, the spending rests on sales. When new investors pay, it rests on the hope that more investors follow. Look for the door that the money uses, then check whether customers or investors stand behind it.

Why did stories matter more than profits in the dot-com bubble?

A telling example is Amazon, valued at over $14 billion in 1998 before it earned a profit. Another is the Super Bowl of 30 January 2000, where dot-coms bought 12 of the 61 ads (Wikipedia, Dot-com bubble). In both cases the story did the work that earnings normally do. People paid today’s price for tomorrow’s results, and the price of tomorrow moved first.

AI has its own stories: an assistant for every worker, an agent for every task. Some will prove right. The simple test is to ask what a story would be worth if the next 3 results arrived late. Stories also shape decisions inside companies: the EQOS Asia AI study reports that 47% of executives took a major decision based on a hallucination.

What made the dot-com bubble crack, explained in simple terms?

A bubble cracks when the money that feeds it stops arriving on time. On 20 March 2000 a Barron’s article warned that internet companies were running out of cash, and the Federal Reserve raised rates the next day (Wikipedia, Dot-com bubble). Companies that depended on fresh funding met a higher price of money at the moment they needed it. That is the crack in plain words.

Pets.com shut down on 9 November 2000, 9 months after its IPO, and most internet stocks were down 75%, which wiped out $1.755 trillion (Wikipedia, Dot-com bubble). The fall needed only one thing: the gap between price and results had to become visible. A few months of missing funding made it visible.

Who survived the dot-com bubble, and what does that mean for the AI bubble?

48% of dot-com companies survived through 2004, at lower valuations, and by 2002 stocks had lost $5 trillion in market capitalisation since the peak (Wikipedia, Dot-com bubble). Survival and loss sat side by side. A company could live, serve customers, and still be worth far less than its peak price. The simple word for this is repricing: the business continues while its price resets.

For AI, repricing means a company can succeed as a business and disappoint as an investment. The study projects productivity gains of +40% and 300 million jobs disrupted, which marks real change. Those gains and a lower price can coexist. The question for your own decision is which side of that split you hold, the business or the price.

Which 5 mirror questions explain the AI bubble simply today?

5 questions summarise the whole page. What is the promise? Who pays for it? Which story carries the price? What would end the funding? Who survives a repricing? Answer each for AI in one sentence and date the answer. As of October 2026 the EQOS Asia AI study gives the market a φ(S) score of 0.337, in the fragile zone, where a shock can tip the system.

The EQOS Asia 30-minute pre-diagnostic turns the 5 answers into a score for your own case, and the Λ simulator gives a first estimate in 10 minutes. You leave with a number you can recompute, which is simpler to defend than an opinion. The same questions work for a board, a family office, or a founder.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile