The dot-com bubble: when a real technology met an unreal price
In the late 1990s, investors bet that the internet would transform the economy. They were right: it did. But they paid prices that assumed the profits would arrive almost at once, and in enormous size. When it became clear they would not, the Nasdaq lost about three-quarters of its value. The episode is the reference point whenever a new technology drives a stock market rally, as AI did until this November.
Up sixfold, down three-quarters
- 1995–98The boom beginsNetscape's stock market debut in August 1995 starts a wave of internet listings. The Nasdaq Composite, heavy in technology firms, rises from about 755 at the start of 1995.
- 1999The maniaHundreds of young, loss-making companies go public, many doubling on their first day. The Nasdaq rises about 86% in the year.
- 10 Mar 2000The peakThe Nasdaq closes above 5,000. The Fed has been raising rates since mid-1999, to 6.5% by May 2000.
- 2000–02The bustProfits disappoint and funding dries up. Many dot-coms go bankrupt. By October 2002 the Nasdaq is down about 78% from its peak.
- 2001Into the real economyBusiness investment in technology collapses and the US enters a recession in March 2001. The Fed cuts rates from 6.5% to 1% by June 2003.
How a good story becomes a bubble
Bubbles usually start with something real. They grow when rising prices themselves become the reason to buy.
The internet did create huge companies and change how people shop, work and communicate. That made it reasonable to pay more for firms that would benefit.
As shares rose, more money poured in: from individuals trading online for the first time, from funds that could not afford to miss out, and from companies raising cheap capital. Valuations came to assume decades of perfect growth.
By 2000, technology was a large share of the US market's value. When it fell, it pulled down index funds, pension funds and household wealth with it, and investment spending followed.
- Asset bubble
- When the price of an asset rises far above what its expected future income can justify, driven mainly by buyers expecting prices to keep rising. Bubbles are easy to name afterwards and hard to identify in real time.
- Contagion
- When trouble in one market or country spreads to others, often because investors pull money from anything that looks similar, not because the others have the same problems.
- Risk-off
- Market shorthand for days when investors cut risk: they sell shares and other risky assets and buy safe ones. The opposite, when they buy risk again, is called risk-on.
- VIX (volatility index)
- A measure of how much investors expect the US stock market to swing over the next month, worked out from options prices. Often called the fear gauge: it is usually between 12 and 20, and jumps in a panic.
- Recession
- A broad fall in economic activity lasting more than a few months, with output, jobs and incomes all falling. A common rule of thumb is two quarters in a row of falling real GDP.
A mild recession, a long recovery for tech shares
The 2001 recession was relatively short, partly because the bubble was mostly financed with shares rather than bank loans, and because the Fed cut rates fast. But the Nasdaq did not regain its 2000 peak until 2015.
Much of the fibre-optic cable and data capacity built in the boom was later used by the companies that did succeed. Amazon's shares fell more than 90% from their peak before the company became one of the world's largest.
What assets did when the bubble burst
Tech shares lost three-quarters; government bonds did well as rates were cut.
| Asset | Period | Move | What happened |
|---|---|---|---|
| Nasdaq Composite | Mar 2000 → Oct 2002 | −74% | −76% after inflation. It did not regain its March 2000 level until 2015. |
| US shares (broad) | Sep 2000 → Oct 2002 | −31% | −34% after inflation. The low came in March 2003. Companies outside tech fell much less. |
| US 10-year yield | Jan 2000 → Jun 2003 | 6.66% → 3.33% | The Fed cut rates and investors sought safety, so bond prices rose. |
| US dollar | Mar 2000 → Feb 2002 | +11% | The dollar kept rising for two years after the peak, then fell 16% by December 2004. |
Monthly averages unless stated; daily highs and lows were more extreme. "After inflation" deflates by US consumer prices. Past moves describe what happened then; they are not a guide to what will happen next.
Then and now: what rhymes
- A real technology, AI, drove a rally led by a small group of chip and technology companies.
- On 5 November 2026 the US semiconductor index fell 12.3% in a day and the Nasdaq entered a correction.
- The selling spread from shares to credit: AI-linked bonds and loans came under pressure.
What's different
The leaders of the AI boom are mostly large, profitable companies, unlike many dot-coms. But today much of the build-out is financed with debt, from bonds to private credit, which can spread losses to lenders and banks. And the build-out depends on a few customers. Whether this is a correction or the start of a longer bust is not yet known; what history suggests is to watch profits and credit, not prices alone.
The same mechanism in this year's news
Recent stories where the forces behind this episode are at work again, newest first.
Sources and notes
- Asset moves — FRED: OECD share price indices (SPASTT01USM661N, SPASTT01KRM661N), NASDAQ Composite (NASDAQCOM), 10-year Treasury yield (GS10), WTI oil (WTISPLC), Fed major-currencies dollar index (TWEXBMTH), CPI (CPIAUCNS).
- Nasdaq Composite: daily closes (NASDAQCOM, via FRED); the chart shows month-end values, so it does not show the intraday peak of 10 March 2000 or the exact low of October 2002.
- Federal funds: effective rate, monthly averages (FEDFUNDS, via FRED). Recession dates: National Bureau of Economic Research.
- Figures for 2026: our stories of 6–8 November 2026.
The Sunday note
One email every Sunday: the week’s most important stories, the mechanics behind them, and the numbers to watch next week. Free.