David Leibowitz, global market strategist at JPMorgan Asset Management, recently said that investors' attitudes toward AI are shifting from blind enthusiasm to rational evaluation. As AI gradually becomes a trading theme across various assets and industries, the ability to accurately identify risks and returns has become a crucial lesson for Wall Street.

Leibowitz pointed out that investors have begun to more carefully distinguish different segments of the AI supply chain, assessing which areas may face oversupply and which still have strong demand. He stated plainly: "The market is no longer simply thinking, 'We like AI, so we buy anything related to AI.' " This shift from "following the crowd" to "picking and choosing" marks the transition of AI investment from being emotion-driven to value-driven.

Robot AI

Chip hardware faces the greatest risk of oversupply, while data centers have more structural support

In specific directions, Leibowitz gave a clear judgment. He believes that compared to semiconductor and hardware production, the demand for data center construction and operation is more likely to have structural support, while the chip sector indeed faces the risk of over-supply.

His words were even more direct: The supply risk mainly lies in the chip and hardware sectors, where investors have poured a lot of enthusiasm. Historical experience repeatedly shows that when the market is overly enthusiastic, it often moves too quickly. This assessment also explains recent market performance— as the frenzy of AI infrastructure investment begins to cool down, the stock prices of some chip manufacturers have seen significant declines. For example, SK Hynix's stock once fell by more than 20% from its June peak, although it has still risen more than two times year-to-date.

AI is everywhere, the key has shifted from "whether to participate" to "how to participate"

Leibowitz ultimately summarized the issue into a more fundamental judgment. He said that after the rapid expansion of the AI landscape, everything has become an AI trade, and AI is everywhere. Therefore, the key is not whether to participate, but how to participate.