The era of indiscriminate gains from artificial intelligence (AI) investments is giving way to a more discerning and strategic approach as the concept permeates every corner of the market. According to David Lebovitz, a global market strategist at JPMorgan Chase Asset Management, investors are becoming increasingly adept at differentiating between the potential risks and rewards across various segments of the AI landscape, marking a crucial lesson for Wall Street. "People are starting to make more distinctions between different sectors, judging where there might be supply overhangs and where there is strong demand," Lebovitz remarked in an interview. "It's no longer the mindset of 'we like AI, so let's buy everything associated with it.'"
Lebovitz emphasized that there are significant structural differences within the AI supply chain. In his view, the demand for data center construction and operation will be more "structural" than the production of semiconductors and hardware, with the chip sector facing a risk of oversupply. "My supply concerns are primarily in the chips and hardware space," Lebovitz stated. "That's clearly where investor enthusiasm has been highest, and history shows that when enthusiasm runs high, people tend to overdo it."
These comments come as the AI chip sector experiences significant volatility. Chip stocks, which had been on a relentless rally, have recently faced heavy selling pressure, signaling a cooling of the fervor around AI infrastructure investment. Notably, shares of South Korean chip giant SK Hynix have fallen more than 20% from their June peak, despite still being up over 200% for the year 2026. The company's highly anticipated Nasdaq listing this week was overshadowed by the broader sell-off in AI chips.
The case of Samsung Electronics delivered an even more striking market shock. On July 7th, the company issued record-breaking second-quarter guidance, with consolidated revenue of approximately 171 trillion won and operating profit around 89.4 trillion won. However, instead of boosting the stock price, this stellar performance triggered a global sell-off in chips. Both Samsung Electronics and SK Hynix shares plunged sharply that day, with South Korea's KOSPI index briefly crashing over 8%, triggering a circuit breaker.
The market widely interprets this as a classic case of "buy the rumor, sell the news"—where months of gains had already priced in all optimistic expectations, so the actual results became an opportunity for profit-taking. Furthermore, news that Meta is considering selling excess AI computing capacity has sparked concerns about potential oversupply in AI infrastructure.
In a separate report, JPMorgan Chase earlier pointed out that over the past year, AI chip and memory manufacturers' stock prices have significantly outperformed those of hyperscale cloud providers like Amazon, Microsoft, and Google, with the "performance gap" widening to an unsustainable level. The report argues that the core variable for the continuation of the AI rally has shifted from capital expenditure to the ability to commercialize and deliver returns.
Despite the short-term pressure on the chip sector, JPMorgan Chase remains optimistic about the long-term prospects for AI infrastructure. The bank forecasts that capital expenditures from the top four U.S. cloud service providers will grow 80% year-over-year in 2026, surpassing $575 billion. JPMorgan Chase has even raised its forecast for total global AI infrastructure investment by 2030 to $5.5 trillion.
Lebovitz concluded that as the AI landscape rapidly expands, the critical question for investors is no longer "whether to invest in AI," but rather "specifically how to invest." "What I'm realizing is that everything is becoming an AI trade," he said. "AI is everywhere. So the key is how you participate, not whether you participate. I think that's the biggest question for investors going forward."
Comments