Inside Bank of America’s AI Bubble Warning: From Extreme Concentration to Jumbo IPOs

BofA’s Internal Stress Indicators Are Flashing

Michael Hartnett, Bank of America’s chief investment strategist, has been the most persistent voice flagging parallels to the 1998–2000 period. His warnings center on a handful of quantitative measures:

The bubble risk indicator is heating up. BofA's internal Bubble Risk Indicator shows the market moving toward a state where a burst looks "inevitable" over time, though the bank acknowledges core AI assets have not yet fully detached from underlying fundamentals .

Price-to-book has surpassed the dot-com peak. The S&P 500’s price-to-book ratio hit a record high of 5.3 in August 2025, exceeding the 5.1 level reached in March 2000 at the very top of the dot-com bubble. Other classic valuation measures, including the 12-month forward price-to-earnings ratio and the Shiller cyclically adjusted P/E, are at or near their highest levels since the dot-com era .

Credit spreads are dangerously tight. As of August 2025, credit spreads on U.S. technology companies sat at just 56 basis points — a level indicating extreme market optimism by historical standards. Hartnett has warned that a sharp widening from these compressed levels could signal serious trouble ahead, noting that spreads soared past 400 basis points during the 1999–2000 meltdown .

The Valuation Gap Between Winners and Losers Is Hitting Extremes

One of the most striking BofA findings concerns market concentration. The bank warns that U.S. market concentration is approaching levels seen only during the 1880s railroad bubble — the single most extreme concentration episode in modern market history . This is not merely a matter of a few stocks doing well. BofA's "regime indicator" suggests that the era of mega-cap leadership may be ending, a pattern that preceded the dot-com crash in 2000 .

Earlier in 2025, BofA released a report warning that U.S. growth stocks had entered a valuation bubble exceeding both the "Nifty Fifty" mania of the 1960s and the dot-com era of the late 1990s. The bank cautioned that a correction could take the S&P 500 down 40% from peak levels .

A more contemporary and perhaps more telling metric is what BofA calls the "ROI Gap." In 2025, an estimated $400 billion was spent on AI infrastructure, but AI software generated only about $100 billion in incremental revenue — a 4:1 spending-to-revenue ratio that echoes the telecom and fiber overbuild that preceded the dot-com bust . The bank has also noted that aggressive capital expenditures from hyperscalers are increasingly reliant on debt, raising the stakes for investors still waiting for returns to materialize .

A Pipeline of Mega-IPOs Raises the Concentration Risk

The accelerating pace of large-scale initial public offerings is another red flag that BofA has explicitly tied to the 1999–2000 playbook. In a May 2026 Flow Show report, the bank calculated that if SpaceX, OpenAI, and Anthropic were to go public and join the existing "AI Big 10," the resulting group would concentrate close to half of the S&P 500's total market capitalization. That level of concentration would exceed every bubble in modern history except the 1880s railroads .

Hartnett has explicitly linked the surge in large-scale AI IPO speculation to the mounting "historic extremes" in concentration risk, warning that the combination of soaring public offerings and already stretched valuations is driving U.S. markets into dangerous territory . The SCMP similarly reported that the gap between richly valued stocks and cheaper ones had reached extremes last seen only before the dot-com bust, while the pace of new listings was matching the run-ups that preceded both the 2000 and 2008 market peaks .

Not Everyone Agrees This Is 1999 All Over Again

It would be misleading to present BofA's view as monolithic. The bank's own analysts disagree sharply on what the current signals mean.

Savita Subramanian, BofA's head of U.S. equity and quantitative strategy, has argued consistently that today's environment is not the year 2000. She describes the situation as an "air pocket" rather than a full bubble, pointing to strong cash flows at major tech companies, high utilization of AI computing power (unlike the "dark fiber" glut of the dot-com era), and less extreme speculation in unprofitable stocks .

A separate team led by analyst Vivek Arya published a note in October 2025 acknowledging the "AI doom headlines" but listing four structural differences from the dot-com era: high compute utilization rather than idle capacity, capex funded by operating cash flows rather than debt, a Federal Reserve more likely to cut than raise rates, and real earnings growth supporting valuations .

European equity strategists at BofA, led by Sebastian Raedler, have drawn a different historical parallel entirely — not to the dot-com bubble but to past capital-expenditure-driven boom-bust cycles . Meanwhile, a December 2025 report from the bank's Global Equity Volatility Insights team argued that the core AI-linked segments of U.S. equities remain "well short of conditions typically associated with an imminent bubble peak," even as speculative pressure intensifies in the broader market .

What Investors Should Watch Next

The picture that emerges from BofA's research is not a clean consensus call. It is a live debate inside one of Wall Street's largest institutions. On one side, Hartnett and the global macro team see compressed credit spreads, record price-to-book ratios, a yawning infrastructure-to-revenue gap, and a surge of mega-IPOs that collectively resemble the conditions that preceded the dot-com crash. On the other, several equity and technology analysts argue that the underlying cash generation and computing utilization of today's AI leaders make this cycle fundamentally more durable than the late 1990s.

The question for investors is which set of signals proves predictive. Historically, the combination of extreme valuation dispersion, extremely tight credit spreads, and a rush of large IPOs has been a reliable precursor to market turbulence — but past cycles have also shown that such conditions can persist longer than skeptics expect before finally breaking.