AI stocks and dot-com stocks share a powerful market pattern: enthusiasm for a transformative technology has lifted a relatively small group of companies and driven a large share of market gains. But the comparison is not one-to-one. Federal Reserve and Nasdaq analyses published in 2025 found that many leading AI-related companies had established earnings and that Nasdaq-100 profitability and valuation measures were less extreme than around the 2000 peak. Those differences matter, but they do not make current prices safe or predict what comes next.
Are AI stocks like dot-com stocks?
In market enthusiasm and concentration, yes. In the earnings and valuation evidence cited here, the resemblance is incomplete. Both periods featured rapid price rises around a major technology shift, with a narrow set of technology-linked companies contributing disproportionately to market performance. But the Federal Reserve’s November 2025 comparison described many dot-com firms as having little realized earnings and speculative revenue prospects, while firms most closely associated with AI generally had established and growing earnings at that time.
That distinction is about the broad market backdrop, not a guarantee for any specific company. A profitable business can still be overpriced, and a promising technology does not ensure that every company associated with it will earn an adequate return.
How do the two market periods compare?
| Comparison | Dot-com boom | AI-linked market | How to interpret it |
|---|---|---|---|
| Price appreciation | The Federal Reserve reported that dot-com firms’ stock prices rose more than 200% from 1996 to 1999. | As of November 2025, the rise in AI-related firms’ stock prices since 2022 was somewhat slower, according to the Federal Reserve. | The Fed’s categories are not exhaustive or perfectly matched populations. The comparison shows a similar scale of enthusiasm, not identical market baskets. Federal Reserve speech, November 21, 2025. |
| Earnings and profitability | Many dot-com firms had little or no realized earnings and speculative revenue prospects. In the Nasdaq-100, 21 companies had negative net margins in 1999. | The Fed said leading AI-related firms generally had established and growing earnings. Nasdaq reported that 99.9% of Nasdaq-100 exposure was profitable in its 2025 comparison, using consensus estimates. | Nasdaq’s 1999 margins use realized results; its 2025 figure is exposure-weighted and based on forecast net income divided by sales. It is not the percentage of companies with realized profits. Nasdaq Global Index Research analysis, November 2025. |
| Valuation | Nasdaq imputed a Nasdaq-100 price-to-earnings ratio (P/E) of 104 at year-end 1999 and estimated it likely reached 150–200 at the first-quarter 2000 peak. | Nasdaq reported trailing P/E ratios largely in the low 30s during the year through November 14, 2025. | These are Nasdaq-100 figures, not measures of every AI stock. The 150–200 peak figure is an estimate. P/E comparisons are also less useful for companies with losses. |
| Technology’s share of market gains | Technology stocks contributed 74% of U.S. equity market gains in the two years leading up to the March 2000 peak, according to MSCI. | Technology stocks contributed 56% in MSCI’s comparison for March 2023–February 2025. Amundi’s 2026 analysis also describes a narrow group of AI-related stocks as driving a disproportionate share of index returns. | Concentration remains substantial even when profitability and valuation measures look less extreme. MSCI analysis; Amundi analysis, 2026. |
| Technology share of the S&P 500 | Technology firms represented 47% of S&P 500 market capitalization in 2000, after doubling from 23% in less than two years. | The share reached 49% by the end of August 2024, after taking nearly a decade to double. | These are historical capitalization observations, not a current 2026 figure. Bank for International Settlements, December 2024. |
Why earnings change the comparison—but do not settle it
Investors value future profits, so earnings are not a complete measure of whether a stock is attractively priced. Still, realized profits and cash generation provide a different foundation from a business whose value rests mostly on hoped-for revenue. The Federal Reserve’s November 2025 assessment was that many leading AI-related firms had established, growing earnings streams, unlike many dot-com companies. Nasdaq’s index data likewise showed far greater profitability in its 2025 comparison than in 1999, with the important caveat that the modern margin figure was based on forecasts.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#1 Best Overall
Do not treat these broad comparisons as proof that every AI-related company is profitable, or that forecasts will be achieved. “AI stock” is not a single, consistently defined market category: it can include established companies selling AI-related products as well as firms whose future depends heavily on AI demand. The Federal Reserve noted that one measure counted about 50 publicly traded AI-focused firms, compared with more than 1,000 publicly listed dot-com firms near the late-1990s peak; those counts use different category definitions and periods.
What the valuation numbers can—and cannot—tell you
Nasdaq’s figures make the valuation gap between its index at the dot-com peak and its 2025 comparison striking: an imputed P/E of 104 at the end of 1999, a likely estimated peak of 150–200 in early 2000, and trailing ratios largely in the low 30s over the year before its November 2025 analysis. But the figures apply to the Nasdaq-100, not to a universal basket of AI stocks, and the early-2000 peak range is an estimate.
A P/E ratio divides a share price by earnings. It becomes difficult to interpret when earnings are negative or unusually small. Trailing P/E uses realized earnings; forward P/E uses estimates, which can be wrong in either direction. The BIS cautions that forecasts may reflect optimism or pessimism. A lower index multiple than in 2000 is useful context, not evidence that every holding is fairly valued or that prices cannot fall.
What investors can learn from the dot-com era
Separate a technology’s promise from a stock’s prospects
The internet’s long-run importance did not guarantee that every dot-com company would survive or that every purchase price would prove rewarding. Amundi notes that many projections during the earlier boom failed to materialize. Apply the same distinction to AI: demand for a technology and returns earned by a particular company or investor are different questions.
Test earnings, cash generation, and investment returns
For an individual business, examine realized earnings and cash generation, then ask whether its investment in AI-related infrastructure and products can plausibly earn an adequate return. Aggregate market statistics cannot validate any specific security. Expectations can be disappointed even at companies with real profits if spending rises faster than returns or growth falls short of what prices imply.
Check concentration in your whole portfolio
A broad index fund can still depend heavily on a small number of companies. MSCI’s 56% technology contribution to U.S. market gains in March 2023–February 2025 illustrates how much a sector can influence broad-market results. Amundi also warns that ordinary equity allocations can embed substantial exposure to AI and long-duration growth. Review direct holdings and the underlying concentration of funds rather than assuming that a diversified-looking portfolio has little exposure to the theme.
Monitor fundamentals alongside market breadth and financing
Amundi identifies the sustainability of earnings and capital expenditure, market breadth, issuance activity, and renewed acceleration in valuations as indicators to watch. These can help frame risk, but no one indicator establishes that a bubble exists or says when a downturn will begin.
Financing is another variable. Federal Reserve Vice Chair Philip N. Jefferson said both eras generally showed limited reliance on debt, while noting that borrowing for AI infrastructure could rise. That is a risk to monitor, not evidence of an established debt crisis.
Best Value
Why the analogy cannot predict a crash
History can provide a useful reference without supplying a timetable. Jefferson put the caution plainly in his November 21, 2025 speech: “history can only be a useful reference and not a predictor of future outcomes.” The earnings mix, valuation measures, concentration, investment needs, and financing conditions can all change, so similarity to the 1990s does not establish that a crash is imminent—or that one will not occur.
This is a comparison of market structure, not personalized investment advice or a forecast. The cited measures cover different universes and dates: Nasdaq’s figures are for the Nasdaq-100, MSCI’s are U.S.-market comparisons, and the Federal Reserve’s AI and dot-com categories are not identical populations. The available comparisons do not establish a single directly comparable measure of the total economic value of AI versus the internet.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
