1. Professional Investors Turn Wary of the AI Buildout
The November BofA fund manager survey, conducted from November 6 to November 12, 2026, found that 52% of respondents expect at least one artificial intelligence hyperscaler to cut capital spending within the next six months. A month earlier the share was 30%. Net overweight positioning in technology stocks fell from 37% to 12% over the same period.
Taken together, the two readings show a rapid change of mood among the professionals who direct large pools of money. The belief that the companies financing the AI buildout will soon pull back has gone from a minority opinion to a slim majority in a single month, and managers have reduced their technology holdings at a pace not seen in two decades.
The results do not show investors walking away from artificial intelligence as a theme. They show investors asking harder questions about what the buildout costs, who pays for it and when the spending will produce a return. The enthusiasm that carried technology stocks has given way to scrutiny of the bill.
2. What Did the BofA Fund Manager Survey Find?
The survey found that a majority of fund managers, 52%, now anticipate a spending cut by at least one AI hyperscaler in the coming six months, up from 30% in October. It also recorded a 25-percentage-point drop in net overweight technology positioning, a record net 38% of respondents saying companies are overinvesting in AI, and an AI bubble named as the largest tail risk facing markets.
Bank of America's Global Fund Manager Survey is a monthly poll of institutional investors about their outlook, their positioning and the risks they consider most pressing. Because it asks the same kinds of questions each month, it is widely used to track how professional sentiment shifts over time, and sudden moves in its readings tend to draw attention.
The November edition contained several such moves at once. Expectations for hyperscaler spending, the weight of technology in portfolios, views on corporate investment and the ranking of market risks all moved in the same direction. Each pointed to growing doubt about whether the scale of AI investment can be sustained.
3. What Is a Hyperscaler Capex Cut?
A hyperscaler capex cut means that one of the giant cloud operators decides to build less than it had previously planned. The term combines two pieces of market shorthand that are central to the AI trade.
Hyperscalers are the very large cloud computing companies that construct and operate the data centers on which AI systems are trained and run. Their facilities house the chips, servers, cooling and networking equipment that make large-scale artificial intelligence possible, and their budgets determine how quickly that capacity expands.
Capital expenditure, usually shortened to capex, is money spent on physical assets such as buildings and machinery. For a hyperscaler, capex is the line that captures data center construction and the hardware installed inside. When investors talk about a capex cut, they mean a reduction in those planned outlays.
The reason such a cut would matter is that hyperscaler spending sits at the top of a long chain. The money flows to chipmakers, equipment suppliers, construction firms, power providers and many other businesses whose sales depend on data centers continuing to be built. A decision by one large operator to slow down would be read as evidence that demand for AI capacity is not keeping up with supply, and it would raise questions about the revenue of every company further down the chain.
That is why the jump from 30% to 52% carries weight. A month ago, fewer than one in three managers expected such an announcement. Now more than half do, which means the possibility has shifted from a risk scenario to something close to a base case among those surveyed.
4. Technology Positioning Falls at the Fastest Pace Since June 2006
Expectations about spending were matched by action in portfolios. Net overweight positioning in technology stocks dropped from 37% to 12%, a decline of 25 percentage points in one month and the steepest fall since June 2006.
An overweight position means a manager holds more of a sector than its share of the benchmark the fund is measured against. The net figure subtracts the proportion of managers who are underweight from the proportion who are overweight, so it summarizes how the group as a whole is leaning. A reading of 37% indicated a strong collective tilt toward technology. A reading of 12% indicates that the tilt has largely, though not entirely, been unwound.
Two details stand out. The first is the speed. Shifts in sector allocation among large institutions normally happen gradually, because moving significant sums takes time and managers are reluctant to abandon positions that have been working. A change of this size within a single month is unusual enough that the survey had to reach back twenty years to find a larger one.
The second is where the reading ended up. At 12%, the figure is still positive, which means managers as a group remain overweight the sector. They have cut back sharply, but they have not moved to an outright bet against technology. The data describes a retreat from an aggressive stance, not a wholesale exit.
5. A Record Share Says Companies Are Overinvesting in AI
Underlying both moves is a change in how investors judge corporate spending itself. A net 38% of respondents said companies are overinvesting in AI, the highest reading the survey has recorded.
The significance of that number lies in what investors usually want from the companies they own. For much of the AI boom, heavy spending was rewarded, because it was taken as a sign that a company was positioning itself to capture a transformative technology. A record share now saying the investment has gone too far indicates that the market's tolerance for ever-larger budgets is wearing thin.
The view links directly to the expectation of a capex cut. If managers think spending has overshot, it follows that they would expect at least one of the biggest spenders to rein it in. The two findings describe the same judgment from different angles: one about what companies are doing, the other about what they are likely to do next.
6. An AI Bubble Tops the List of Tail Risks
Asked to identify the greatest tail risk confronting markets, fund managers named an AI bubble. Close behind came concern about credit events among AI hyperscalers.
A tail risk is an outcome considered unlikely but severe if it occurs. The survey's ranking shows which low-probability dangers investors spend the most time worrying about, and in November the top two were both tied to artificial intelligence.
The second of those concerns adds a dimension beyond stock prices. A credit event refers to trouble in the debt of a borrower, such as a failure to meet obligations. Its presence near the top of the list signals that investors are thinking about how the AI buildout has been financed as well as how it has been valued, and about what would happen if the returns from data centers fall short of what is needed to service the money raised to build them.
Having both risks at the head of the ranking shows how concentrated investor anxiety has become. The same theme that drove technology shares higher is now the main source of the scenarios that managers fear most.
7. Inflation and Bond Worries Fade as AI Returns Come Into Question
The November results also record a change in the subject of investor concern. Worries about inflation and bond yields have eased, according to the findings, and attention has moved to whether large-scale AI data center projects will pay off.
That rotation is notable because inflation and interest rates have dominated market anxiety for years. Their retreat from the top of the list does not mean investors feel more comfortable overall. It means the source of discomfort has changed, from the broad economic environment to a specific question about one sector's investment cycle.
As a consequence, managers are reassessing their positions in technology and in semiconductors. Chipmakers are among the most direct beneficiaries of data center construction, so doubts about the payoff from those projects extend naturally to the companies that supply them.
The underlying question is one of timing as much as of scale. Data centers require large sums to be committed up front, while the income they generate arrives over many years. Investors who were content to wait for that income while optimism was high are now asking how long the wait will be and how much of the spending it will ultimately justify.
8. How Bank of America's Strategist Reads the Results
Michael Hartnett, the Bank of America strategist associated with the survey, interpreted the findings as a sign that a period of excessive optimism around technology and AI investment is coming to an end.
That reading fits the pattern across the individual questions. Managers expect spending to be trimmed, they have sharply reduced their overweight in technology, a record share believes companies are investing too much, and the risks they rank highest are an AI bubble and credit stress among the largest builders.
It is worth keeping in mind what a sentiment survey can and cannot show. The poll captures what professional investors think and how they say they are positioned at a point in time. It does not establish that a hyperscaler will in fact reduce its budget, or that technology shares will fall. What it does establish is that the consensus among large money managers has moved quickly, and that the AI trade is now being judged on its costs as closely as on its promise.
