What was the dot-com bubble? Lessons for the AI bubble

What was the dot-com bubble in plain terms?

The dot-com bubble was a period when investors paid rising prices for internet companies on the strength of future growth, well before profits arrived. It peaked on 10 March 2000, when the Nasdaq Composite reached 5,048.62, and it ended in a 78% fall by October 2002 (Wikipedia, Dot-com bubble). A bubble has 2 parts: a real change, and a price that assumes the change pays off fast.

That definition works as a test. Ask whether the price assumes more than the business can deliver, and over what horizon. For the AI bubble, the EQOS Asia AI study asks the same question at market level: valuations rose ×4 in 18 months while the market scores 0.337 on φ(S). The test gives you a question to answer and an answer to date.

What fed the dot-com bubble before the peak of 2000?

Three fuels appear in the record. The Telecommunications Act of 1996 opened telecom networks to heavy building, and equipment companies invested over $500 billion in the following 5 years, mostly with debt (Wikipedia, Dot-com bubble). Public listings let young companies raise money before they earned profits. And a shared story said that web traffic would turn into earnings. Each fuel is a funding source you can check for AI.

Stories can outrun numbers. Pets.com shut down on 9 November 2000, 9 months after its IPO, and AOL announced its merger with Time Warner on 10 January 2000, then the largest merger to date (Wikipedia, Dot-com bubble). These 2 dates show a story at full strength and the same story at its end. For AI, the matching habit is to read each announcement against the revenue behind it.

What are the 3 phases of the dot-com bubble from 1995 to 2002?

The timeline has 3 phases: a rise that began in 1995, a peak on 10 March 2000, and a fall to October 2002. The detailed guide, the dot-com bubble timeline, walks through each phase and marks the events that mattered. Reading the phases in order helps because each one asks investors a different question.

For AI the same reading splits the story into where you stand and what comes next. You do not need a peak date to use the timeline. You need to know which phase today’s evidence matches, and what a change of phase would look like. The EQOS Asia method page shows how a score turns that judgement into a number you can recompute.

What did the dot-com bubble in the United States set up for the AI bubble?

The Nasdaq Composite, a United States index, is the standard measure of the dot-com bubble, and the telecom build-out that followed the 1996 Act was American too. The guide on the dot-com bubble in the United States covers the companies, the exchange, and the rules of that market, and asks which parts of the setting still apply.

Today the AI market is global, and decision makers in Singapore, ASEAN, and Hong Kong hold exposure through listed shares, chip supply chains, and data centres. The American lesson travels through those channels. The useful question is where your own exposure sits, and the Λ simulator estimates φ(S) for your case in 10 minutes, free and anonymous.

What does a dot-com bubble chart show that numbers alone do not?

A chart shows shape: a long climb, a sharp top, and a drop that took 31 months from March 2000 to October 2002. Shape tells you about speed and symmetry, which a single percentage hides. The guide on the dot-com bubble chart compares that curve with the path of AI valuations and explains what to check before you trust a comparison of 2 lines.

Charts also mislead. A scale that starts in 1995 or in 1999 changes the story, and 2 curves drawn on one axis can look alike for unrelated reasons. Check the start date, the scale, and the units first. The EQOS Asia AI study reports a ×4 valuation rise over 18 months as a stated interval with a source, which gives a chart something firm to draw.

How can the dot-com bubble be explained simply, with the AI bubble as a mirror?

A short version fits in 3 sentences. A new technology promised a lot. Investors paid for the promise before the results. When results arrived slower than the price assumed, the price fell. The guide on the dot-com bubble explained simply tells it without jargon and holds the AI bubble up as a mirror at each step.

The mirror works step by step. The promise is generative AI used everywhere. The price is a ×4 valuation rise in 18 months. The results are the 95% of projects that fail in production (MIT 2025, cited in the EQOS Asia AI study). Whether the gap closes by growth or by repricing is the open question as of October 2026.

What lessons does the dot-com bubble give the AI bubble?

4 lessons carry over. Check who funds the spending. Separate the technology from its price. Expect the run to last longer than a sceptic wants and end faster than a believer hopes. Write down the triggers in advance. Each lesson is a decision rule, so none of them requires a forecast of the date. They suit boards, investment committees, and founders alike.

Pairing the lessons with a measurement makes them usable. The EQOS Asia 30-minute pre-diagnostic scores your AI exposure, and the pricing page lists the formats for a deeper study. You bring the decision in Singapore, ASEAN, or Hong Kong, and you leave with a score and ranked options. The result is reproducible, so you can rerun it each quarter.

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