Are We on the Brink of an AI-Fueled Stock Market Crash?
The whispers are growing louder: is the stock market inflating an artificial intelligence (AI) bubble destined to burst? You’re not alone in your concern. Google Trends reveals a surge in searches for “AI bubble” in August 2025, dwarfing even the frenzy around “crypto bubble” searches. Having witnessed firsthand the Japanese asset bubble, the commodity bubble, the U.S. housing crisis, and the meme stock mania, I can attest that the 1990s tech bubble remains the most captivating—and the one most often invoked when discussing today’s AI-driven market. But here’s where it gets controversial: while parallels exist, the differences are just as striking, and the stakes for investors couldn’t be higher.
From 1992 to 1999, the S&P 500 soared nearly 250%. The 2020s are on a similar trajectory, but history reminds us that the 1990s ended in a devastating crash, with the index halved and the Nasdaq plummeting 80% from its peak. This is the part most people miss: the stock market today is undeniably expensive by historical standards, but is it a bubble, or is one merely forming? Should investors flee now, or ride the wave until the first signs of deflation? A deep dive into history might offer clues—and a strategy.
My Journey Through Bubbles Past and Present
I bought my first stock in 1992 and landed my first job in the investment industry in 1994 as an intern at Smith Barney. By 1998, I was at a boutique firm specializing in macroeconomic allocation. Like many, I dabbled in day trading—a statistically losing game, yet in the late 1990s, it felt impossible to lose. Fast forward to today, and while the 2020s echo the early 1990s, predicting the next five years feels like reading tea leaves. But I have some educated guesses.
In the mid-1990s, the future felt electric. I remember Netscape’s green browser crawling on a dial-up modem, thinking, “This will change everything.” Today, the buzz is about GPUs and data centers, not 56k modems. Investors now ask: if the past five years mirror the early 1990s, will the next five mirror the late 1990s? The short answer: the parallels are real, but the differences matter—and portfolios hang in the balance.
The Bubble Question: How Close Are We?
Nick Colas of Datatrek offers a rule of thumb: whenever the S&P 500 doubles in a short period, it’s a bubble. If the S&P 500 rises another 7%, it will have doubled since October 2022. Is three years “short”? And has AI made this time truly different? The 1990s index took six years to double, then doubled again from 1996 to 1999. What’s next for the S&P 500? Another double, or an imminent crash?
1990–1994 vs. 2020–2024: Echoes of the Past
Both eras began with shocks followed by recoveries. The early 1990s saw a recession tied to the savings and loan crisis, while the 2020s started with a pandemic and the fastest bear-to-bull transition ever. By late 2023, GDP growth stabilized around 3%, mirroring the mid-1990s productivity boom. Both periods also witnessed industrial revolutions: the internet then, AI now. Generative AI, for instance, has gone from novelty to necessity at lightning speed, with 65% of organizations using it regularly in 2024, up from 33% in 2023. ChatGPT hit 100 million users in just two months—unprecedented.
Markets thrive on adoption curves. The S&P 500’s back-to-back outsized years in 2023 and 2024 echo the 1990s, though such streaks are rare. Tech dominates again, accounting for nearly a third of the index, slightly below the 1990s peak but still unnerving. Yet, this concentration is typical when platform technologies scale—desktop operating systems then, cloud computing and AI now.
IPOs and Market Frenzy: Then and Now
Market fervor is measured by IPO activity. Netscape’s 1995 IPO opened the floodgates, leading to 544 U.S. offerings by 1999, raising $69 billion ($130 billion adjusted). Today, the IPO market, frozen from 2022 to 2024 under Lina Khan’s FTC, reopened in Q3 2025 with 64 IPOs raising $14.6 billion—a rebound, but far from hysteria. Unlike the 1990s, today’s IPOs are generally profitable, not just hopeful.
Fed Speak: “Irrational Exuberance” vs. “Fairly Highly Valued”
Alan Greenspan’s 1996 warning of “irrational exuberance” is legendary. The market doubled before crashing. Jerome Powell’s 2025 remark that equities are “fairly highly valued” sparked a brief flinch but no panic. Both Feds acknowledged risks without calling a top—their job is risk management, not market timing.
Risks and Opportunities: What Rhymes and What Doesn’t
The 1990s had crises—the Mexican peso crisis, Asian contagion, Russian default—yet each kept complacency at bay, ironically extending the rally. The 2020s crises are heavier: pandemic, wars, inflation, bank wobbles. Yet, the bull case rests on two key factors: productivity and stimulus. AI could drive productivity gains, and stable inflation might allow the Fed to keep rates accommodative, fueling growth.
What If the Next Five Years Mirror 1995–1999?
The optimistic scenario: disinflation, productivity boom, and investment super-cycle could drive the S&P 500 to compound at 20% annually, as it did from 1995 to 1999. If AI mirrors the internet’s impact, earnings will likely drive stock prices. Year-to-date, the S&P 500’s 14% rise is mostly earnings-driven, not speculative. But here’s the counterpoint: while bubbles often pop due to central bank rate hikes, today’s rates are lower than the 1990s peak and expected to decline, not rise. Is this time truly different?
The Bottom Line: Stay In, But Stay Alert
While a correction could come anytime, the current environment favors bulls. My base case: earnings-led gains, periodic volatility, and productivity growth keeping inflation in check. But valuation risks loom. The right strategy? Stay invested, but be ready to act if necessary.
Thought-Provoking Question: Is AI the next internet, or are we overestimating its transformative power? Share your thoughts in the comments—let’s debate!
