While the stock market has climbed overall in recent months, bouts of volatility have raised concerns about overconcentration in artificial intelligence. Indeed, the No. 1 question from Fidelity’s advisor clients this year has been about whether or not an AI bubble is forming.
Processing Content
Experts say there are clear indicators that advisors can monitor to assess that risk. Regardless of whether AI enthusiasm proves excessive, they say proactive diversification of client portfolios is the best hedge to any potential volatility.
Warning indicators advisors should watch
All the signals that a bubble is forming are present to one degree or another, said Rick Wedell, chief investment officer at RFG Advisory in Birmingham, Alabama.
Valuations are high across the industry, growth is incredibly rapid and there is some degree of circular investment among the industry players, he said.
“While the growth levels are explosive, the actual steady state cash flows of these businesses are unclear since so much investment continues as the businesses refine themselves,” he said.
The first sign of a bubble isn’t irrational exuberance, but disconnection from underlying fundamentals, said Shawn DuBravac, CEO and president of the research and strategy consultancy Avrio Institute in Washington, D.C.
Investors should be worried when asset prices decouple from revenue generation, he said, or when valuations are justified by potential measured across decades rather than quarters.
“The real canary in the coal mine is the gap between what companies are spending on AI infrastructure and what they’re actually generating in return,” he said. “When that gap is impossibly large, we have a problem.”
Capital allocation behavior is another tell, said DuBravac. To him, late-stage warning signals include: venture dollars flowing to companies whose primary business model is proximity to AI; special purpose acquisition companies (SPACs) reemerging with AI branding; and retail investors treating AI exposure as a category rather than a thesis.
The market is still trying to determine what AI will mean for the long-term profitability of many of the companies making up the S&P 500, said John O’Connell, founder and CEO of wealth management consultancy The Oasis Group. Also uncertain is what broader AI consumption will look like over time.
“Companies across every sector are implementing AI to promote new levels of growth and to reduce overall operating expenses,” he said. “The effects of those initiatives in the market are unrealized yet.”
Advisors should also closely watch enterprise adoption trends, said DuBravac. If Fortune 500 companies continue piloting AI tools without converting them into scaled, revenue-generating deployments, that may signal the technology isn’t yet delivering the productivity gains that the market is likely pricing in.
“We’re not necessarily in a bubble today, but we are in a moment where the burden of proof is increasing,” he said. “AI has to start showing up in productivity data, earnings and measurable economic output and not just in press releases.”
READ MORE: How financial advisors can buy a wealth book of business
Dot-com or GFC? Assessing systemic risk
To date, Wedell said the companies involved appear to have relatively little leverage applied to them, making the rally largely equity-fueled, more similar to 2000 than a credit-fueled rally like 2008.
“That’s good, because it means that when things start to pop, it’s easier to wash it through the system,” he said. “You don’t necessarily have to have a banking crisis at the same time.”
DuBravac agreed that any AI correction would likely look more like 2000 than 2008. The dot-com bust was painful and deep, but it was not a systemic financial crisis, he said.
“The underlying economy absorbed the blow, recovered and ultimately benefited from the infrastructure the bubble left behind,” he said. “The internet didn’t disappear. And the survivors got stronger.”
If there is a correction, the fallout would likely be concentrated, not systemic, said Jake Miller, co-founder and chief solutions officer at Opto Investments in Los Angeles. The speculative fringe could be wiped out, much as Pets.com was while Amazon survived, he said.
“The infrastructure layer and vertical AI applications driving measurable ROI in healthcare, construction, manufacturing and defense will keep compounding because the demand is structural,” he said. “The dot-com bust destroyed the pretenders but did not destroy the internet.”
The irony of a bubble is that the technology itself survives, said DuBravac. Though the railroad bubble of 1873 wiped out thousands of investors, we still have railroads, he said.
“AI will follow a similar arc,” he said. The applications that generate real, measurable value will persist. The rest will be a footnote.”
READ MORE: Using AI to write that client email? Think twice.
How should advisors position portfolios?
These companies might be overvalued, but that doesn’t mean they won’t continue to appreciate, said Wedell. That’s why he said he recommends holding a steady allocation to stocks and bonds that participate in market upswings while cushioning against downturns.
AI capital expenditure, fiscal expansion, tariffs and deregulation are colliding to create a high-pressure economy that rewards selectivity and punishes passive exposure, said Miller.
The best-positioned clients are those with exposure to domestically oriented businesses where AI is not a valuation story but a value-creation tool, he said, citing examples of firms whose AI products compress construction timelines, automate health care billing or improve manufacturing quality inspection.
“These are companies large enough to have meaningful workflows but small enough to implement AI without multi-year transformation programs,” he said. “They also don’t come with the aggressive valuations of late-stage venture AI bets or public hyper-scalers.”
The fallout of any correction would be most intense for clients who have failed to diversify across all market segments, said Wedell. It could also be particularly painful for investors who remained on the short end of the duration curve, since a bursting bubble could lead to a Treasury rally, he said.
“Stop trying to time the correction,” he said. “That’s not actually possible.”
