AI & Compute

BofA survey: Tech overweight falls to 12% on AI capex fears

A net 12% of fund managers are overweight technology, down from 37% in October, while 52% expect an AI hyperscaler to trim capital spending within six months.

By
Rebecca Stone
Filed
Channel
AI & Compute
Read
3 min read

By Rebecca Stone

A net 12% of fund managers were overweight technology in Bank of America’s November Global Fund Manager Survey, down from 37% in October. BofA analysts called the 25-percentage-point drop the largest one-month reduction since June 2006.

BofA ran the survey from Nov. 6 to Nov. 12 and published the results on Tuesday, Nov. 17. It included 193 panelists managing $526 billion; 169 of them, managing $461 billion, answered the global questions.

The report landed hours before Nvidia was due to report third-quarter results after the close, with U.S. stock futures lower that morning. The timing matters: the positioning data predates Nvidia’s release and any market response to it.

What is driving the retreat from technology?

The spending concern was unusually broad. A record net 38% of respondents said companies were overinvesting, while only 16% wanted companies to raise capital spending, the lowest share since May 2020.

Meanwhile, 52% expected an AI hyperscaler to announce a capital-spending cut within six months, up from 30% in October. BofA summarized the survey as a call to pause AI capex.

BofA analysts also said the survey’s “big message” was a pause or slowdown in AI capital expenditure, describing investor expectations rather than confirmed reductions in spending.

The downside scenarios reflected that concern. An AI bubble ranked as the top tail risk at 34%, followed by an AI hyperscaler credit event at 31%. Among investors expecting a credit event, 38% identified hyperscaler bonds as the likeliest source and 30% pointed to private credit.

Worries about inflation and a disorderly rise in bond yields eased in the November reading.

Do investors still favor semiconductor stocks?

Long global semiconductors was named by 39% of respondents as the most crowded trade, down sharply from 82% in July. Other crowded positions were narrower:

  • Short Chinese stocks: 14%.
  • Long Magnificent Seven stocks: 12%.

A crowded-trade label identifies popularity, not the size of each fund’s allocation. Still, the split is more nuanced than a wholesale exit: the broad technology overweight fell sharply while global semiconductors remained the survey’s most crowded position, although the July reading was far more extreme.

For chip suppliers, the survey therefore points to selective risk reduction rather than an abandonment of semiconductor exposure. The positioning data captures investor sentiment, not current orders.

What do bonds, cash and gold show?

Investors reduced their bond underweight by 13 percentage points, the largest monthly move toward bonds since April 2025, but still held a net 23% underweight. They said a 10-year Treasury yield of 5.8% would prompt a switch from stocks to bonds.

Average cash holdings declined to 3.6% from 3.9%, their lowest level since August. A record net 23% called gold undervalued.

What is the macroeconomic backdrop?

The survey’s macro readings remained broadly benign:

  • Soft landing: 50%.
  • No landing: 42%.
  • Hard landing: 3%.
  • Best 2027 performers: international stocks at 36%, U.S. stocks at 24% and gold at 17%.

The share expecting a boom fell to 41% from 47% in October.

Michael Hartnett’s team said they “stay tactically risk-off.” Against a 50% soft-landing majority, that recommendation reflects the survey’s tension: the macro backdrop was not universally pessimistic, but expectations for AI spending had turned defensive.

For AI semiconductor suppliers, the next six months hinge on whether the expected hyperscaler spending cuts become actual reductions rather than merely a shift in investor positioning. The survey contains no capacity or pricing data with which to judge the effect. For now, the market is divided over AI returns even as global semiconductors remain the most crowded trade.

Source: mlq.ai

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

1,049 articles

Related articles

« Previous article