The Nasdaq 100 has entered a correction phase. How can you protect your portfolio if the AI bubble bursts?
Investors are concerned about both the high costs of AI and the possibility that those costs might be reduced
The technology-focused Nasdaq-100 index has entered a correction phase / Photo: DCStockPhotography / Shutterstock.com
The Nasdaq 100 technology stock index fell 1.4% on October 2 and closed 10.4% below its all-time high on October 6, entering correction territory. October was its worst month since March 2025. The last week of the month proved particularly difficult for tech stocks: following Alphabet’s quarterly earnings report, concerns intensified over the cost and financing of the AI boom, according to Bloomberg. Strategists offer three responses to these fears: combine safe-haven assets with riskier ones, add defensive stocks and exchange-traded commodities, or buy more stocks on pullbacks, according to Business Insider.
The push toward AI needs a counterbalance
Investors have long feared that hyperscalers—the largest owners of cloud platforms and data centers—are spending too much on AI with an unclear return on investment. In recent months, however, a countervailing threat has emerged: some tech giants have begun talking about cost discipline, and capital expenditures on AI could drop sharply. Such a shift could trigger a chain reaction among suppliers, the publication notes.
David Rosenberg, founder of the research firm Rosenberg Research, noted that shares of both cloud giants and AI hardware providers fell sharply last week. “When stocks on both sides of the same investment thesis fall on opposing news, the problem lies with the thesis itself. Simply put, the market is pricing in a lower return on AI spending,” he wrote to clients.
The “barbell” strategy. BCA Research advises combining risky and safe assets in a portfolio: if the former decline, the latter protect against losses. Overall, the firm is optimistic about the outlook for stocks and recommends increasing the share of technology and healthcare stocks, as well as high-yield debt instruments, in portfolios.
Safe-Haven and Real Assets. Paul Dietrich, chief investment strategist at Wedbush Securities, suggests that the AI bubble could burst as early as the beginning of 2027, by which time the peak of AI investment will be behind us. He advises investing in gold, which he calls a “life insurance policy” for a portfolio, as well as in healthcare stocks, copper, and dividend-paying stocks.
Buying on the Dip. Asset management firm Franklin Templeton has a positive outlook on U.S. stocks but expects their prices to remain highly volatile for the time being. Therefore, according to its chief market strategist, Chris Galipo, it makes sense to buy stocks at a discount during market pullbacks. “Earnings expectations are high—very high. Any minor misstep is punished, and this leads to capital rotation,” he explained.
Will the AI bubble burst?
Ron O'Hanley, CEO of State Street, does not yet see any serious risks in the AI market, as the massive investments in AI are backed by demand. “These data centers aren’t being built on a whim. They’re backed by real orders and, in most cases, real companies. So it’s hard to see a bubble here right now,” he said on Bloomberg TV.
O'Hanley also commented on the market decline: “A minor sell-off was probably due. And right now, it certainly doesn’t look like a chaotic one.” He did not rule out the risk that there might end up being more data centers than needed: “The question will be whether AI becomes so much more efficient that we won’t need all the capacity currently under construction.”
In the race for AI, companies have raised more than $650 billion in the markets in just under two years, according to Bloomberg. David Solomon, CEO of Goldman Sachs, assured attendees at an investment summit in Hong Kong on November 3 that capital markets will continue to finance the development of AI infrastructure. “There will be financing to meet this demand, and it will be substantial,” Solomon promised.
This article was AI-translated and verified by a human editor
