Three Warning Signs the AI Trade Is Flashing a Dot-Com Red Alert

Valuation multiples for AI-related firms have also reached eye-watering levels. Pre-revenue companies in sectors such as quantum computing and eVTOL (electric vertical takeoff and landing) are commanding valuations 5 to 10 times higher than their dot-com-era peers, as flagged by analysts at Goldman Sachs and others . Palantir, a data-analytics company frequently associated with AI, traded at 123 times sales at one point .

Bank of America equity and quant strategist Victoria Roloff has acknowledged that "narrow market breadth and lofty multiples rhyme with 2000" but cautioned that she sees "more of an AI air pocket than a fully inflated AI bubble (at least for now)" . The distinction matters: an air pocket suggests overvaluation concentrated in specific names, whereas a true bubble implies systemic excess.

Further metrics reinforce the valuation concern. The S&P 500's price-to-book ratio reached 5.3 in August 2025, exceeding the 5.1 level recorded during the dot-com bubble's peak in March 2000 . The cyclically adjusted price-to-earnings (CAPE) ratio has risen above two standard deviations from its historical average, according to economist David Rosenberg, a level he argues is consistent with a market that has been in a price bubble for more than a year .

2. A surge of jumbo IPOs matching pre-2000 and pre-2008 pace

The pipeline of initial public offerings from AI-focused companies has swelled to proportions that echo the pre-peak IPO booms of the late 1990s and the mid-2000s. Business Insider reported in June 2026 that the race to go public among mega-cap tech companies is "sending a signal that last flashed during the boom of public offerings from the late 1990s to the early 2000s" . Analysts at TS Lombard argue that the wave of mega-IPOs indicates a desire among company insiders and early investors to sell stock to the public at elevated valuations—behavior historically seen near market tops .

The scale of the planned listings is striking. The combined hoped-for valuations of just three companies—SpaceX, OpenAI, and Anthropic—would exceed the entire dot-com IPO wave of 1995–2000 when adjusted for inflation, according to market analysis . In March 2026, a "$200 billion AI IPO wave" was described as reshaping Wall Street, led by CoreWeave, Databricks, and Cerebras Systems . The South China Morning Post noted that the "slew of jumbo initial public offerings" is explicitly echoing the run-ups to past market peaks, matching the pace seen before both the 2000 and 2008 declines .

The historical parallel is uncomfortable: IPO activity surged in the years before the dot-com bubble burst, and a similar pattern played out ahead of the 2008 financial crisis. While IPO volume alone does not guarantee a crash, it has been a reliable signal of late-cycle exuberance in prior episodes.

3. Seven of Bank of America's 10 bear-market indicators are flashing

Perhaps the most systematic warning comes from Bank of America's internal bear-market framework. As of June 2026, strategist Savita Subramanian reported that 7 of the bank's 10 proprietary bear-market signposts have been triggered—up from 5 in April and 4 in March . Historically, the triggering of 7 signals has been the typical level at which a market peak occurs, a pattern that has held across bear markets since 1990 .

The indicators span a broad range of metrics, including consumer confidence, stock performance expectations, credit stress levels, credit tightness, and the relative performance of high price-to-earnings stocks versus low price-to-earnings stocks . Two newly triggered signals in May were particularly noteworthy: the outperformance of high-P/E stocks over low-P/E stocks—a classic sign of excessive speculation—and overly optimistic long-term growth expectations that exceed five-year historical trends .

In response to the accumulated warnings, Bank of America explicitly advised clients to take profits, stating there were "too many red flags" . The bank's European equity strategy team separately warned in February 2026 that "doubts around the AI revolution are emerging" and that the market narrative was shifting from an "upside-only" perspective to concerns that AI could actively destroy corporate profits rather than boost them universally .

The steady progression of triggered signals—from four in March to seven in May—suggests the deterioration has been gradual rather than sudden, but the direction has been consistent. By late May and early June 2026, 70% of the bank's bear-market indicators were flashing red .

The debate: bubble, air pocket, or base camp?

Despite the accumulating warnings, the analyst community is far from consensus. A significant faction argues that the parallels to the dot-com bubble are overstated or at least premature.

Goldman Sachs has warned that current AI investment imbalances closely mirror the 1997 inflection point rather than the 2000 peak, implying that the cycle could still have years to run before a bust . The investment bank pointed to rising capital expenditures alongside falling profits and widening credit spreads as imbalances that will become "more visible as the AI investment boom extends" but stopped short of calling a near-term peak .

Within Bank of America itself, there are contradictory views. A December 2025 report from the bank's Global Research division argued that "pockets of the market are already displaying instability consistent with late-cycle excess" but that "the main AI-linked segments of U.S. equities remain well short of conditions typically associated with an imminent bubble peak" . The same team projected in their year-ahead outlook that fears of an AI bubble are "overstated" and that AI-related capital expenditures would continue to build rather than burst .

Reuters' Return on Investment team published an analysis in May 2026 arguing that while the "hyperscaler-led AI capex boom may eclipse dotcom mania, a repeat of the crash of 2000 is unlikely" even if AI demand fails to match supply . The International Monetary Fund similarly stated that a downturn was plausible but unlikely to evolve into a systemic crisis .

The Bank of England has taken a more cautious stance, warning in October 2025 that "stretched" AI valuations risk a "sharp market correction" if expectations around the technology's impact become less optimistic . Its Financial Policy Committee drew direct comparisons to the late-1990s dot-com craze .

What makes this moment especially difficult to parse is that many of the same institutions issuing warnings are also hedging their calls with upside scenarios. Bank of America's Bubble Risk Indicator shows speculative pressure has intensified, but core AI assets have not fully detached from fundamentals . The debate turns on which analogy is most appropriate: the 1997 build-up phase, the 2000 peak, or something new.

Where the risks are concentrated

For investors watching these signals, several pressure points deserve attention. Credit spreads, which have remained at historically tight levels, have begun to widen in recent weeks—the ICE Bank of America US High Yield Index has shown early signs of stress . The Buffett Indicator—the ratio of total market capitalization to GDP—has reached 230%, more than two standard deviations above its long-term trend, signaling systemic overvaluation beyond just the technology sector .

The IPO pipeline itself creates mechanical risks. The sheer size of upcoming listings means large-scale forced buying by index funds, potentially creating artificial demand that could reverse when the new issuance is absorbed or when early investors begin selling into the market .

None of these signals guarantees a crash, and the AI trade differs from the dot-com era in important ways. Today's leading AI companies are profitable enterprises with real revenues, not speculative startups with no business models. The technology itself is already generating measurable economic value across multiple industries. But the warning signs are sufficiently numerous and historically resonant that even the most bullish institutions are advising caution.