The question got sharper on November 20, when the Nasdaq fell 5.28% and wiped out its 2026 gains, ending the day 18% below its October 6 record. Nearly every side-by-side comparison of AI and the dot-com era was written before that week, so their "how far has it fallen so far" figures are already out of date. One analysis from early November had the Nasdaq only 6.8% below its peak.
Here's how the two booms compare on the measures people cite most.
| Measure | Dot-com era (2000) | AI boom (2025–2026) |
|---|---|---|
| Nasdaq fall from peak | 77.9%, March 2000 to October 2002 (FRED via AI Bubble Monitor) | About 18% from the October 6, 2026 record (CNN via KSL) |
| Shiller CAPE (long-run valuation) | 44.2, December 1999 | 40.9, end of September 2026 (AI Bubble Monitor) |
| S&P 500 forward P/E | 24.4, March 24, 2000 | 18.7, October 30, 2026 (AI Bubble Monitor) |
| Tech share of S&P 500 value | Nearly 35%, March 2000 | 39.4%, June 2026 (AI Bubble Monitor) |
| Most valuable company | Cisco, $579.1 billion | Nvidia, about $4.99 trillion on November 3, 2026 (AI Bubble Monitor) |
| Profits | About 14% of dot-com companies profitable at the peak | Nvidia: $215.9 billion FY2026 revenue, roughly 53% net margin (IntuitionLabs) |
| Tech investment's share of US GDP growth | 28% (2000) | 39% (first nine months of 2025) (St. Louis Fed via AI Bubble Monitor) |
| Build-out spending as share of GDP | Telecom and fiber peaked at 1.2% | Hyperscaler capex expected near 3% a year in 2027–2029 (Apollo) |
The short version: on measures of size and exposure, the AI boom matches or beats 2000. On profits and prices relative to expected earnings, it doesn't come close.
Where does the AI boom look like 2000?
It looks most like 2000 in how expensive the market is over the long run and how few companies drive it. The Shiller CAPE, which compares prices with ten years of inflation-adjusted earnings, was within about 8% of its dot-com record at the end of September. Tech now makes up a bigger share of the S&P 500 than it did in March 2000, and J.P. Morgan put the top 10 stocks at about 40.6% of the index.
The star stocks follow a similar script. Nvidia's roughly 4,300% five-year gain before 2024 is close to Cisco's ~4,500% run before its 2000 peak. Private money is piling in too: AI startups took 61% of global venture investment in 2025, up from 30% in 2022.
Interest rates have also moved the way they did in 2000. The Fed raised rates on September 16, 2026 for the first time since July 2023, which IntuitionLabs calls a condition from 2000 that has returned.
Where is it different?
The biggest difference is earnings. In 2000 most internet companies had little or no profit to justify their prices. Today's leaders make enormous profits, as Nvidia's results show. Nvidia's revenue also grew 7x in three years, against Cisco's 4x before its peak.
Watch which numbers a comparison uses for this, though. The headline "60x" valuation for 2000 is the Nasdaq-100's forward P/E. The S&P 500's forward P/E in 2000 was 24.4. Putting 60x next to today's S&P 500 figure of around 19 makes the gap look bigger than it is. The gap is real but narrower: 24.4 against 18.7.
Who pays for the boom differs as well. Edelweiss argues that AI spending is funded by cash-rich incumbents rather than loss-making startups, as it was in 1999.
What did the November sell-off change?
It pushed AI further away from 2000 on valuation. According to FactSet data cited by CNN, the S&P 500 tech sector's forward P/E went into November 20 at its lowest level since March 2020, and Nvidia's was the lowest since 2002.
That means prices are falling faster than analysts' profit forecasts. In 2000 it went the other way: there were few earnings to fall back on. Nvidia is now down 47% from its October peak, while its expected earnings have held up much better than its share price.
The catch is that those forward P/Es rest on forecasts. If analysts cut their profit estimates for chipmakers and cloud companies, the gap with 2000 would narrow from the earnings side. Nobody knows yet whether the estimates will hold.
Is the economy more exposed than in 2000?
By the available measures, yes. St. Louis Fed researchers estimated that AI-related investment added 0.97 percentage points to US GDP growth in the first nine months of 2025. That was 39% of all growth, compared with 0.81 points, or 28%, from similar investment in 2000.
Apollo says the late-1990s telecom and fiber build-out peaked at 1.2% of GDP before collapsing. Consensus forecasts put hyperscaler spending at around 3% of GDP a year in 2027–2029, up from 0.3% in 2019.
Most comparison pages focus on stock prices and skip this. If AI spending slowed sharply, more of the economy would feel it than in 2000. Whether that would mean a worse downturn than the 2001 recession is an open question.
Where is the debt?
Mostly not on Big Tech's own balance sheets. Edelweiss points out that leverage is building further down the chain, in data centers, chips, energy and "neoclouds", much as debt-funded fiber networks did in 2000. How much debt sits there, and whether it could spread losses the way telecom debt did, isn't publicly clear.
Some of the newer AI companies are already under strain. On November 20, CoreWeave closed at $39.46, below its $40 IPO price. Anthropic ended the day at $88.30, just above its $83 IPO price. Its prospectus showed a 2025 net loss of about $42 billion, including a roughly $34 billion accounting charge, alongside run-rate revenue above $47 billion.
Does an 18% drop tell us which kind of decline this is?
Not yet. The dot-com bust took about two and a half years to reach bottom, and the Nasdaq needed about 15 years to regain its 2000 peak. A few weeks of losses can't show whether this is a correction or the start of something similar.
The selling has been severe. The S&P 500 posted seven straight losses, its longest streak since February 2020, and the VIX fear gauge jumped to 38.50 from 19.3 a week earlier. Some strategists still read it as a correction. JPMorgan Asset Management's David Kelly called it a "mild cold," and CFRA's Arun Sundaram said the fundamentals "don't really justify all this selling."
For context on what followed last time: after the 2000 peak, defensive sectors held up, value stocks beat growth for years, and commodities and emerging markets rallied. Whether that pattern repeats depends on which side of the table above wins out.