Michael Hartnett, Chief Investment Strategist at Bank of America Merrill Lynch, has issued his most explicit bubble warning to date, stating that the current AI-driven market bubble is the largest since the 19th-century railway bubble. However, he cautions investors that it is premature to reduce positions until two key conditions are met.
In his latest "Flow Show" report, Hartnett notes that the market concentration of the AI sector has approached approximately 48%, surpassing all historical bubbles including the "Roaring Twenties," the "Nifty Fifty" of the 1970s, the Japanese stock market of the 1980s, and the tech-media-telecom bubble of the 1990s. The only historical peak not yet exceeded is the 63% concentration seen during the height of the 1880s railway bubble. Concurrently, BofA's Bull & Bear Indicator has risen to 8.0 this week, officially triggering a contrarian sell signal for risk assets—the 17th such signal since 2002.
Despite the clear top signal, Hartnett believes that bullish investors are unlikely to exit easily at this moment. He outlines two conditions to wait for: first, the pricing of historic IPOs such as SpaceX and OpenAI; second, a rise in CPI to 4-5% in the coming months, which could trigger significant policy tightening.
The yield on the 30-year U.S. Treasury note has reached a 19-year high this week, putting pressure on emerging market currencies across the board—the Korean won is nearing a 30-year low, the Japanese yen is approaching a 35-year low, and the Indonesian rupiah and Indian rupee have both fallen to record lows, indicating that risks are rapidly spreading from the periphery.
Hartnett warns that the surge in global capital costs is eroding the peripheral layers of risk assets, encompassing housing markets, consumer sectors, and private equity. Historically, emerging markets have been the starting point for large-scale risk-off sell-offs. If yields and oil prices ultimately reassert dominance in market pricing, the impact on equities could equate to a drop of over 1000 points in the S&P 500.
Bubble Scale: Market Concentration Nears Railway-Era Extremes
Hartnett characterizes current market sentiment as "price strength, retail frenzy, and low volatility," describing the current setup as "so bubbly."
On market concentration, if upcoming major tech IPOs are included in the AI sector, the concentration of AI-related assets could easily surpass approximately 48%, exceeding all notable bubble cases of the past century. The only historical record yet to be surpassed is the extreme level of 63% of total market value seen at the peak of the 1880s railway bubble.
About a year ago, when many analysts labeled AI as a "super bubble" comparable to or surpassing the dot-com bubble, Hartnett was more reserved, viewing AI then as a "mini bubble" and not yet a true peak. In hindsight, investors who reduced positions early in anticipation of a bubble burst missed out on significant subsequent gains. Now, his stance has shifted notably.
Sell Signal Triggered, But Two Prerequisites Remain Unmet
BofA's Bull & Bear Indicator has risen to 8.0, triggering a contrarian sell signal. Factors driving the indicator higher include continued net inflows into tech stocks and emerging market bonds, the largest single-month increase in equity allocation in the Fund Manager Survey (FMS), and a drop in FMS cash levels to 3.9%.
Since 2002, this indicator has triggered 17 sell signals, with a hit rate of approximately 60%. Global equities have typically declined by 2-3% over the following 2-3 months, with maximum drawdowns potentially reaching 15-20%. Meanwhile, the latest Fund Manager Survey shows that market consensus is extremely optimistic on both positioning and earnings expectations, while a breakout in yields suggests some profit-taking pressure is building.
Nevertheless, Hartnett advises waiting.
He notes that no one will voluntarily cut long positions before historic IPOs are priced, and underwriting banks will not allow a market crash before then, as it would cost them billions in underwriting fees.
Additionally, Hartnett provides two bond yield indicators to watch: if XBI rises to $120, it signals yields entering a circuit-breaker-like surge channel; if XRT breaks above $85, it suggests a bond market shock may be delayed.
Yield Surge and Emerging Market Currency Pressure
Hartnett characterizes the rise in the 30-year U.S. Treasury yield to a 19-year high as a typical path for the end of this cycle's boom and bubble, noting that bond vigilantes are actively at work. He warns that if the reality reflected by yields and oil prices ultimately impacts equity valuations, the resulting pressure could equate to a drop of over 1000 points in the S&P 500.
From a geopolitical and imported inflation perspective, the AI wave is reshaping Asia's tech industry landscape. Asia is becoming an exporter of inflation—South Korean semiconductor export prices have risen 148% year-over-year, while DRAM prices have surged 223% year-over-year.
Simultaneously, the Korean won is hovering near a 30-year low, the Japanese yen is approaching a 35-year low, and the Indonesian rupiah and Indian rupee have both hit record lows. Hartnett warns that the surge in global capital costs is eroding the periphery of risk assets, and emerging markets have historically been the starting point for large-scale risk aversion.
Bonds Continue to Attract Funds, Equity Allocation Remains Cautious
The latest weekly fund flow data shows a clear divergence.
Bonds attracted $30.5 billion, the largest inflow among major asset classes; equities saw only $2.4 billion in inflows; cash inflows were $1.2 billion; gold experienced outflows of $1.1 billion; and cryptocurrencies saw outflows of $1.5 billion. This pattern aligns with the elevated sentiment revealed by the Bull & Bear Indicator—while substantial funds chase yield, marginal increases in risk asset allocations have become more cautious.
Hartnett clearly distinguishes between a "wealth-price spiral" and a "wage-price spiral": the wealth effect from equities does exist, but it benefits only a very small portion of the population, highlighting significant wealth inequality.
The strain on the consumer side is particularly evident. The ratio of equal-weight consumer stocks to the S&P 500 has fallen below lows seen during the Lehman crisis, reflecting the difficult situation of highly indebted U.S. consumers.
Meanwhile, Hartnett notes that AI is suppressing labor prices (though not employment numbers), and as the U.S. government begins addressing affordability concerns and energy inflation, the U.S. Strategic Petroleum Reserve (SPR) has declined by 10% since March, releasing approximately 41 million barrels, rapidly approaching near-historic lows.
Post-Bubble Positioning: Contrarian Opportunities Emerge
Looking ahead, Hartnett believes emerging markets and commodities remain in a structural bull market, and consumer stocks will be the best contrarian investment targets after a bubble bursts.
Within the AI sector, he points out that the most cost-effective direction from now on will be small-cap tech AI adopters and transformers capable of disrupting monopolies, duopolies, and oligopolies—a path highly reminiscent of the historical trajectory following the burst of the "Nifty Fifty" bubble in the late 1970s, when waves of technological innovation swept through traditional giants.
Hartnett concludes with an ironic observation on the zeitgeist: "Now everyone believes stocks are the best inflation hedge." He explicitly counters this consensus, stating that stocks are not an inflation hedge—a point that often only becomes "obvious" in hindsight.
Comments