Will a Stock Market Crash Follow the AI Boom? Billionaire Ken Griffin Warns About Echoes of the Dot-Com Bubble.

The artificial intelligence (AI) revolution began in earnest with the introduction of ChatGPT in November 2022. The S&P 500 (^GSPC 0.26%) has advanced 72% since then despite headwinds arising from high interest rates, stubborn inflation, and sweeping tariffs.

Michael Cembalest, chairman of market and investment strategy at J.P.Morgan, recently said, “AI related stocks have accounted for 75% of S&P 500 returns, 80% of earnings growth, and 90% of capital spending growth since ChatGPT launched.”

Naturally, some experts have drawn comparisons to the dot-com bubble, a period where rich valuations across the technology sector in the late 1990s led to a stock market crash in the early 2000s. Hedge fund billionaire Ken Griffin recently told CNBC, “There are obviously echoes of the dot-com bubble in this moment.”

How the artificial intelligence (AI) boom differs from the dot-com bubble

Billionaire Ken Griffin runs the most successful hedge fund in history as measured by cumulative net gains, so his opinion should carry weight with investors. However, his recent comment about the artificial intelligence (AI) boom echoing the dot-com bubble should be considered in context.

Griffin in the second quarter purchased a substantial amount of stock in three companies at the very heart of the AI boom: Nvidia, Amazon, and Microsoft. Those stocks now rank among his six largest holdings, which suggests Griffin is comfortable with their valuations and has confidence in their future growth prospects.

Importantly, while valuations are elevated today, they are still a far cry from the valuations seen during the dot-com bubble. The five largest technology stocks in 2000 had a market-cap weighted forward PE ratio of 59, according to JPMorgan. Comparatively, the five largest technology stocks today have a market-cap weighted forward PE ratio of 34.

“During the dotcom bubble, rampant speculation surrounded young companies flaunting internet excitement before profitability,” according to JPMorgan strategist Stephanie Aliaga. “In contrast, today’s AI beneficiaries are already very profitable companies that make their money selling key infrastructure and resources.”

In short, the dot-com bubble was fueled by speculative bets on internet companies, many of which were losing money, and valuations during that period were much higher than what we see today. Furthermore, many of the most heavily weighted AI stocks in the S&P 500 today are paragons of financial strength.

The “Magnificent Seven” in aggregate reported earnings growth of 27% in Q2 2025, while the other 493 stocks in the S&P 500 reported earnings growth of 8%, according to FactSet Research. The “Magnificent Seven” are on pace to report faster earnings growth than the rest of the index for the third straight year, and analysts expect that trend to remain intact through 2026.

The S&P 500 still trades at a pricey valuation that has historically correlated with negative forward returns

While valuations may have been frothier during the dot-com bubble, that does not mean the stock market is cheap today. The S&P 500 recorded a cyclically adjusted price-to-earning (CAPE) ratio of 38 in August, a material premium to the monthly average of 31 during the last decade.

There have only been 40 incidents where the S&P 500’s monthly CAPE multiple exceeded 37 since the index was created in 1957. Even more alarming, economist Robert Shiller has back-tested the data using a precursor to the S&P 500 (called the S&P 90) and there have only been 40 incidents where the monthly CAPE ratio has topped 37 since 1928, a period that covers 1,172 months.

Put differently, the S&P 500 — widely considered the best barometer for the U.S. stock market — has only been this expensive 3.4% of the time in history. And the index has usually performed poorly in the aftermath. The chart below shows the median return during the one, two, and three years following monthly CAPE readings above 37.

Data source: Robert Shiller.

As shown above, the S&P 500 has typically produced negative returns during the one, two, and three years following monthly CAPE readings above 37. And if its performance matches the historical median, the index will decline 19% over the next three years.

Here’s the bottom line: The artificial intelligence boom has carried the U.S. stock market much higher in the last three years. Valuations are nowhere close to the those seen in the years preceding the dot-com crash in the early 2000s, but the S&P 500 still trades at a very expensive CAPE ratio that has historically coincided with negative forward returns.

So, investors should avoid risky behavior. That means scrutinize valuations closely before you invest money, and never buy a stock if you aren’t prepared to hold through a downturn. Also, good opportunities may be difficult to find in the current environment, so it’s OK to focus on building a cash position to capitalize on the next drawdown.