Tech stocks fall: a correction or a turning point for AI?

The stock market sell-off in technology stocks has brought artificial intelligence (AI) back to the center of the investment debate. Following an especially difficult session on Wall Street, the sell-off spread to Asia and reopened a question the market has been grappling with for months: whether AI valuations and investment levels are sustainable. Some see this as a turning point; others view it as a necessary adjustment after a particularly demanding run.

The correction has hit semiconductor companies and infrastructure providers hardest, while other market segments have held up better. Three industry voices, representing Asian equities, the macroeconomic outlook, and multi-asset allocation, agree on one key point: structural demand for AI has not faded, but the market is drawing a sharper distinction between those able to finance their growth and those that are not.

Why are technology and AI stocks falling?

The scale of the move is notable. The Nasdaq fell 5.5% and the Philadelphia Semiconductor Index dropped around 12%, according to Pruksa Iamthongthong, Head of Asia-Pacific Equities at Aberdeen Investments, who describes it as "a pronounced risk-off move led by the artificial intelligence and technology sectors". The decline erased the initial rally following the U.S. midterm elections and hit Taiwan and Korea hard, given their significant exposure to semiconductors and memory chips.

Clémence Rusek, Senior Investment Strategist, Boutique Multi Asset at Vontobel, attributes the correction to four factors: higher-than-expected U.S. inflation, which pushed bond yields to their highest levels in more than two decades; concerns about financing the AI boom; advances by Chinese memory chipmaker CXMT, fueling doubts about pricing power; and concentrated positions in a handful of winners that amplified the sell-off.

Iamthongthong adds the political factor. Democratic gains have heightened concerns about increased scrutiny of data center development, at a time when investors were already questioning the financing and profitability of AI capital expenditure. This concern also explains the decline in Treasury yields, which reflect expectations of slower expansion.

Is this a turning point or a valuation reset?

Asset managers are avoiding talk of a collapse. Rusek (Vontobel) argues that "the AI cycle has not yet peaked": demand remains strong, and the latest corporate results confirm that computing capacity is in short supply. What is changing is the criteria, as investors become more selective and reward companies that can finance their investments with their own earnings.

Along the same lines, Iamthongthong (Aberdeen) notes that Asian memory companies were already turning AI spending into revenue, profits and cash flows, while supplies of high-bandwidth memory remain scarce. Therefore, a correction in expectations for data centers in the U.S. does not mean that end demand has disappeared. Iamthongthong adds: "Over time, we expect the impact to be manageable."

Giordano Lombardo, founder, CEO and CIO of Plenisfer Investments (part of Generali Investments), sees little doubt about the ultimate destination, with more growth and productivity, but warns that the transition may be uneven. His point of reference is the rise and fall of telecommunications in the 1990s: "We have seen this before." That investment in fiber and broadband boosted productivity, but first fueled a stock market bubble.

What risks does spending on AI infrastructure pose?

Lombardo (Plenisfer) emphasizes that the demand for capital for AI infrastructure is much greater than that of the Internet expansion relative to the size of the economy. In the short term, this pushes up the prices of chips, energy and raw materials such as copper, adding to inflationary pressures. If returns on invested capital prove modest, some of that investment will have to be written off, and competition from open-source models could reduce margins across the sector.

Rusek (Vontobel) focuses on financing: as borrowing costs rose, bonds issued by the most heavily indebted AI computing providers plunged, and industry leaders warned that access to financing could constrain growth more than chips or electricity. In addition, "concentration carries risks", since portfolios heavily weighted toward a few large companies may be less diversified than they appear.

Where might opportunities emerge after the correction?

Iamthongthong (Aberdeen) acknowledges that short-term volatility could continue as the market assesses potential delays at data centers and limits on computing capacity. Even so, she sees stocks correcting beyond their expected impact on earnings: "value is beginning to emerge". AI capacity may also shift geographically, and the case for Asia extends beyond AI, with opportunities in financials, consumer, healthcare, industrials, and domestic demand.

Rusek (Vontobel) views the correction as "a healthy adjustment within an ongoing investment cycle". He notes that earnings have been encouraging across all regions, bond yields appear to be stabilizing from their highs, and sentiment and positioning are less stretched. Defensive sectors and quality and low-volatility styles held up better, although he warns that inflation, energy, and geopolitics could keep volatility elevated.

Conclusion

The market has shifted from rewarding the AI narrative to demanding returns and financing capacity. All three views suggest that technology remains a structural trend, but that the transition is less linear than valuations had implied. The lesson for professional investors is twofold: distinguish between those turning investment into earnings and those relying on cheap debt, and review the true concentration of portfolios in a handful of major names.