What Happened to the Biggest Stocks of the Dot‑Com Bubble?

Most of them lost 77% to 94% of their value, and most needed more than a decade to get it back. Exent’s data on 11 large, widely held tech stocks of 1999–2000 that still trade today shows a median fall of 82% from each stock’s bubble peak to its bust low. With dividends reinvested, the median stock took 14 years to close above that peak again.

The table is sorted by how long the wait was. The peak is each stock’s highest close in 1999–2000, so it is not always the same day. Microsoft, Amazon and IBM peaked in 1999, before the Nasdaq’s top in March 2000.

Daily closes, dividends reinvested. “Back to peak” is the first close at or above the bubble peak after the bust low. Microsoft and Applied Materials fell further in the 2008 crisis, to −69.5% and −85.2% below their peaks, before they recovered.

Yes, but it took more than two decades. Cisco was briefly the most valuable company in the world in March 2000 (Cisco’s own history site). Its close of $80.06 on Mar 27, 2000 was followed by $8.60 on Oct 8, 2002, a fall of 89%.

With dividends reinvested, Cisco first closed above its 2000 peak in August 2021, 21.4 years later. On the share price alone, it did not close above $80.06 until Dec 10, 2025, 25.7 years later. Intel’s story is similar: its price‑only peak of $74.88 on Aug 31, 2000 was not passed until Apr 24, 2026.

Neither company collapsed; both stayed large, established businesses. The problem for buyers at the top was the price they paid. Cisco traded at about 222 times its earnings for the fiscal year that included its peak, as the valuation table below shows. Illustrative: $500,000 put into Cisco at the March 2000 close was worth about $54,000 at the October 2002 low.

About 15 years with dividends, and 16.5 years on price alone. QQQ, the fund that tracks the Nasdaq‑100, closed at $117.88 on Mar 24, 2000 and at $20.06 on Oct 9, 2002, a fall of 83%. With dividends reinvested it got back to its peak in February 2015. On price alone, it first closed above $117.88 in September 2016.

The broad market fell far less. The S&P 500 fund SPY fell 47.5% with dividends and was back at its peak in October 2006, before the 2008 crisis pushed it down again. How QQQ behaved in later falls, including 2022, is in how much QQQ can fall in a crash, and recovery times across many crashes are in how long markets take to recover.

A decade later, most were still underwater. At the end of 2010, 7 of the 11 stocks were below their bubble peak, even with dividends. Over the full 26 years to Oct 23, 2026, the picture splits in two. The table compares each stock with the S&P 500 bought on the same day as that stock’s peak.

Dividends reinvested; eBay adjusted for the 2015 PayPal spin‑off. Five of the 11 (Cisco, Intel, Oracle, Qualcomm and IBM) trailed the S&P 500 from their own peak, some by a wide margin. Six beat it. Apple, Amazon and Nvidia did so by a huge amount, but in 2000 Nvidia was a small chipmaker and Apple was far from the largest company. The giants of the day, Cisco and Intel, roughly doubled in 26 years while the index rose about eightfold.

One caution about this list: these are the survivors. Several other stars of 2000 never recovered at all. Nortel Networks, one of the largest telecom equipment makers of the era, filed for bankruptcy protection on Jan 14, 2009 (IndustryWeek). A list built only from companies still trading today flatters the outcome.

Today’s largest companies are priced at far lower multiples of their profits. At their bubble peaks, the profitable 2000 leaders traded at a median of about 70 times earnings, and Amazon had no earnings at all. The 10 largest US stocks traded at a median of about 28 times trailing earnings at the Oct 23, 2026 close, in Exent’s data.

P/E at the peak = peak share price divided by diluted earnings per share for the fiscal year that included the peak. Median of the 10 profitable companies: about 70.

Median of the 10: about 28. Today’s P/E uses the last four quarters of earnings, so the two tables are not measured the same way; read the gap as a rough size, not a precise one. Trailing earnings can include one‑off items. Alphabet’s 15.8 reflects large non‑operating gains in 2026, and Tesla’s 376 reflects unusually thin recent profits, so neither is like‑for‑like.

Other measures look closer to 2000. The Shiller CAPE for the whole S&P 500 was 41.87 on Oct 9, 2026, against its December 1999 record of 44.19 (multpl.com). By mid‑2025 the 10 largest companies made up almost 40% of the index, the highest share since the mid‑1960s (S&P Dow Jones Indices). Lower multiples on the leaders did not stop sharp falls in the past either: a hypothetical equal‑weight Magnificent 7 basket lost 45.1% in 2022. The full set of measures is in are we in a stock market bubble.

History cannot answer that, and this page does not try. What it shows is the range of outcomes when a market’s favourite stocks were priced for perfection: deep falls even for great businesses, waits measured in decades, and a handful of huge winners that were hard to tell apart from the rest in 2000.

Bubbles are only confirmed afterwards. The 1990s also had several sharp falls that did not end the boom. What a 2000‑style or 2022‑style fall in AI stocks did to whole portfolios, including Nvidia and chip funds, is in what happens to your portfolio if the AI bubble bursts. Nvidia’s own history of large drops is on the NVDA data page.

A few checks follow from the record above (education, not advice):

Every major fall and its depth is listed in stock market crashes in history.

Exent market data, daily closes from Jan 2, 1998 to Oct 23, 2026, dividends reinvested and adjusted for splits and spin‑offs. The peak is each stock’s highest close in 1999–2000. The bust low is its lowest close from that peak to June 2003. “Back to peak” is the first close at or above the peak after the bust low; Nvidia briefly passed its June 2000 high in 2001, before falling to its 2002 low. Price‑only figures, without dividends, are stated as such.

The 11 stocks are large or widely held tech names of 2000 that still trade today; companies that went bankrupt or were bought have no complete price history and are left out of the tables, which makes the results look better than the era really was. P/E figures use reported diluted earnings. All results are past and describe one period; they are not a forecast.