PROVIDENCE, RHODE ISLAND, Nov 16 : Until recently, investors embraced artificial intelligence as one of the biggest bets Wall Street had ever made. Now many portfolio managers are treating the sector with greater caution as they try to assess how much AI risk is embedded in their portfolios.
There is a lot at stake. At least $25.3 trillion is invested in the largest global companies with the most direct exposure to AI, according to the MSCI Global Intelligence Index. Beyond that is an impossible-to-quantify additional exposure to companies that may benefit or suffer as AI reshapes industries.
The latest bruising selloff in AI-related stocks, including the slump on Wednesday, prompted by the resignation of Nvidia's CEO, Jensen Huang, has put a new focus on how much AI exposure investors actually have.
For Andy Goldberg, chief investment strategist at Nomura Asset Management, the scale of the problem hit home last month when he saw that wedding retailer David's Bridal  planned to spin off and list its AI wedding-planning business next year.Â
"Since when was a wedding gown shop a play on AI?" Goldberg said, adding the problem is that exposure to AI exists even in the most unlikely corners of the economy. David's Bridal CEO Kelly Cook said the firm "has been a pioneer in applying AI to retail for several years.â
From trucking companies using AI tools to tackle logistical problems to municipal bonds whose credit ratings may hinge on revenue derived from building new data centers, AI impacts both risk and return, Goldberg said.
"No one fully understands how much exposure they have to AI; itâs really become just one big trade at this point,â said Goldberg. If investors decide to trim even 1 per cent or 2 per cent from their allocations to AI plays, he added, the reaction will be "something that (looks) like a rout."
FORENSIC TASKÂ
For investors trying to map that exposure, the process is increasingly becoming a forensic exercise.
At PGIM, the investment arm of Prudential Insurance, analysts and portfolio managers overseeing nearly $1 trillion in fixed-income investments began this year putting every bond, loan, and structured security through two AI-focused screens. One evaluates AI's impact on a company's business model and credit profile; the other looks for exposure to events ranging from a slowdown in AI-related capital spending to a market crisis in the most unlikely companies.Â
âWe know how to assess the direct exposure, like, I have a data center or I have chipsâ in the portfolio, said Brian Barnhurst, head of global credit research at PGIM. âBut how about the indirect things? And thatâs critically important to understand."
Barnhurst said the approach helps analysts isolate "what does AI mean for this healthcare business, or this industrial business or this telecom business?âÂ
OVEREXPOSED?Â
The amount invested in AI is huge.Â
The Equable Institute, a nonpartisan pension fund think tank, calculated that a basket of 52 publicly traded AI-related companies made up 38.5 per cent of public equity holdings at a group of large pension funds as of June 2026, up from 20.7 per cent at the beginning of 2010.Â
Those funds have between 9 per cent and 11 per cent of their assets invested in AI-focused companies, Equable calculated. Â The figure excludes debt holdings, venture capital and other private equity investments, meaning it almost certainly understates their broader exposure.
"Understanding true portfolio exposure is becoming more important" at this point in the AI cycle, said Jay Jacobs, US head of equity ETFs at BlackRock. The fact that so many AI core holdings now dominate big indexes creates a degree of concentration that is "making diversification harder to achieve than investors may realize."
Adding to the uncertainty is the near-constant news around major AI players such as Nvidia, the looming IPO of Anthropic and concerns over CoreWeave.
Some managers are responding by becoming pickier.Â
The market's willingness to pummel members of the AI trade "can be seen as an early warning indicator that we all need to be more selective in looking at the AI rollout," said Peter Berezin, chief economist at BCA Research, a market, economic and geopolitical analysis firm. "It's very possible that AI will end up making companies a lot more productive but not necessarily more profitable."
Powell Industries, whose power management systems are key components in data centers, illustrates the swings. Its stock, only $8 four years ago, this year traded as high as $328 but has since slumped to $146.Â
Powell Industries did not respond to requests for comment.
Berezin said that bottlenecks in data center construction will eventually ease, diminishing windfall gains for Powell and other AI beneficiaries. When that happens is another unknown.
GASOLINE TO THE FIRE
That is a recipe for continued volatility, investors and analysts agree.
"We've seen a sort of laundry list of concerns pile up" since June, when AI valuations peaked, said Shaon Baqui, a technology analyst at Janus Henderson. He said the US midterm election result "almost feels like putting gasoline on top of the fire," with Democratic gains raising concerns about the future of AI growth. The result, he said, is that investors "increasingly will want to separate the wheat from the chaff."
Liz Ann Sonders, chief investment strategist at the Schwab Center for Financial Research, said that even with careful research, "the math behind all of this AI exposure" makes such a concentrated market particularly difficult to navigate.Â
But a hard-headed assessment of AI portfolio concentration could uncover opportunities as well as risks, said Rob Arnott, founder of Syzygy Asset Management.
"There's a kind of tunnel vision at work here," he said. While investors are "taking on substantial risk in a segment of the market that is frothier, they are overlooking the chance to invest early in other companies for whom AI will create value that has so far gone unrecognized."
Source: CNA
At a glance
- Topic
- Business
- Country
- đ¸đŹ SG
- Publisher language
- EN
- Added to Allnira