BofA Survey Sounds the "Crowding Alarm": Tech Overweight Hits Six-Year High, AI Bubble Debate Nears a 50-50 Split

BofA's October Global Fund Manager Survey shows that global investors continued to add to tech stocks amid the AI-driven rally, with overweight positioning rising to 37% from 30% last monththe highest since August 2020. About 45% of respondents expect an AI bubble in 2027, while 42% disagree. The 37% net overweight is not an isolated data point. Looking back at the survey trajectory this year, net equity overweight hit a record high in May; in July, long semiconductors was seen by 80% of respondents as the "most crowded trade," setting an all-time survey record; in August, cash allocation fell to an extremely low 3.5%, and the Bull & Bear Indicator rose to 9.3, triggering a BofA "sell" signal. Macro optimism persists, but cracks are emerging In this survey, respondents identified global semiconductor stocks as the current market's "most crowded trade." In the prior June survey, crowding in long semiconductors had already declined notably, yet long semiconductors still topped the most-crowded-trade list for multiple consecutive months. AI has been the hottest trade in the stock market, driving the S&P 500 up nearly 120% over a four-year bull market. But enthusiasm has cooled as investors question whether returns can justify the massive spending in the tech sector. On Tuesday, South Korean memory chip giants Samsung Electronics and SK Hynix fell sharply, dragging Asia's benchmark stock index to its lowest since August. Futures tracking the tech-heavy Nasdaq 100 fell 0.6%. The survey was conducted from October 2 to 8, with 161 respondents managing a combined $453 billion in assets. Tech overweight tops the list, AI trade narrows to leaders Notably, beneath the surface, positioning is diverginghedge funds are "trimming on the surface while concentrating in reality." The latest Goldman Sachs data shows hedge funds last week pushed tech stocks to the sector with the largest net selling, with the long-short ratio falling to the 17th percentile of the past five years. At the same time, however, Mag 7 positioning as a share of hedge funds' total U.S. equity exposure has risen from 14% in June to 22%, a record high in prime book history. This contradiction of "cutting sectors, concentrating in individual names" reveals a dangerous structure in the current market: no one is truly leaving the AI trade, they are just rotating from broad tech into the most core leaders. Once leader earnings disappoint, highly concentrated positioning will amplify volatility. Goldman derivatives data also shows that VIX upside convexity is extremely sought after while S&P 500 downside protection is "extremely cheap"investors are hedging against a volatility spike, not a stock market decline, and the pricing logic itself is asymmetric. Stock-bond switching threshold emerges, AI capex becomes a concern Bond pessimism fell to its lowest since May 2025. About 37% of respondents said a 10-year Treasury yield of 6% would prompt them to shift from stocks to bonds; about 19% said they are already adjusting allocations. The current 10-year Treasury yield is around 5.19%. The survey shows most fund managers believe a rise in Treasury yields to around 6% would be needed to trigger a large-scale shift from stocks to bonds, with the largest Treasury short covering in a year currently underway. Equities remain net overweight, but the overweight fell from 49% to 44%; cash positioning held at a low of 3.9%, with both BofA's Cash Rule and Bull & Bear Indicator triggering sell warnings. In terms of allocation shifts, funds added to bonds, Japanese and UK equities, and utilities, while significantly cutting banks, U.S. equities, and consumer discretionary. By sector, tech overweight reached its highest since August 2020, while consumer discretionary underweight hit its worst level since July 2008, with positioning highly concentrated. Institutions ranked a disorderly rise in Treasury yields as the top tail risk, with the AI bubble second. 39% of investors believe a contraction in AI capex by major tech companies is most likely to trigger systemic credit risk; 30% of institutions expect AI companies to cut capex in the next six months. Goldman derivatives data also shows that VIX upside convexity is extremely sought after while S&P 500 downside protection is "extremely cheap"investors are hedging against a volatility spike, not a stock market decline, and the pricing logic itself is asymmetric. Overall, global institutions remain betting on tech and high-quality earnings, favoring large-cap and high-dividend styles, but low cash positioning, crowded trades, AI capex uncertainty, and Treasury yield disruptions constitute the main market constraints. The contrarian approach suggested by the BofA report is: go long bonds, REITs, and consumer sectors, and reduce exposure to overheated tech and banks. Surveys only reflect the current consensus of institutions, and consensus reversals often bring violent swings. When tech overweight hits a six-year high and the AI bubble debate nears a 50-50 split, what the market really needs to guard against may not be whether a bubble exists, but the chain reaction once crowded trades begin to unwind.