As the artificial intelligence concept permeates every corner of the market, investor perception of this theme is shifting from the fervor of “everything rises with AI” to a more nuanced and granular approach to portfolio allocation. David Lebovitz, Global Market Strategist at J.P. Morgan Asset Management, stated that the market is becoming increasingly adept at distinguishing the potential risks and rewards among different segments of the AI space. He pointed out that people are beginning to make more distinctions between various sectors, judging where there might be an oversupply and where there is robust demand, and that the mindset of “if you like AI, you buy all related assets” is now outdated.
Lebovitz explicitly noted that there are significant structural differences within the AI industry chain. In his view, the demand for data center construction and operations will be more “structural” than that for semiconductors and hardware production, whereas the chip sector may face the risk of oversupply. He pointed out that chips and hardware are precisely where investor enthusiasm is at its peak, and historical experience suggests that when enthusiasm runs high, the market tends to overextend itself.
These remarks coincide with a period of intense volatility in the AI chip sector. Shares of South Korean chip giant SK Hynix (SKHY) have fallen more than 20% from their June all-time high, despite the stock still being up over 200% cumulatively since the beginning of 2026. The company was originally scheduled to list on the Nasdaq this week, but the chip sell-off has cast a shadow over its debut. Even more shocking to the market was the case of Samsung Electronics (SSNLF), which on July 7 issued a record-high second-quarter earnings guidance. However, this milestone failed to boost its stock price and instead triggered a global chip sell-off, during which South Korea’s KOSPI index plunged more than 8% intraday, triggering a circuit breaker. The market widely interpreted this as a classic case of “buy the rumor, sell the news.” In addition, news that Meta plans to sell its surplus AI computing power externally also fueled concerns about demand saturation.
In another research report, J.P. Morgan pointed out that over the past year, AI chip and memory manufacturer stocks have significantly outperformed cloud service providers such as Amazon (AMZN), Microsoft (MSFT), and Google (GOOGL), and that this “earnings gap” has widened to an unsustainable level. The report argued that whether the AI rally can continue in the future now depends less on capital expenditure and more on the ability to commercialize and realize returns. Although the chip sector faces short-term pressure, the firm remains optimistic about the long-term prospects of AI infrastructure, forecasting that the combined capital expenditure of the four major U.S. cloud service providers will grow by 80% year-over-year in 2026, and raising its total global AI infrastructure construction investment forecast for 2030 to $5.5 trillion.
The most popular trading strategy in the U.S. stock market in 2026—”buy chip stocks, sell software stocks”—is now showing signs of unraveling. In previous months, the market feared that AI would erode the growth potential of software companies, leading to sustained selling pressure on that sector. Now, software stocks are finally seeing a rebound. Meanwhile, chip manufacturers, which had benefited from surging AI demand earlier this year, have been underperforming as investors begin to question how much of the trillions of dollars in planned capital expenditures will actually materialize.
Institutional views suggest that chip stocks are facing valuation concerns at elevated levels, with their volatility index having nearly doubled since the start of the year. Some analysts have pointed out that sharp fluctuations, combined with the sector’s proximity to historical highs, could signal an extended consolidation phase or a peak and subsequent pullback. Nevertheless, the fundamental outlook for chip stocks remains strong, with earnings growth projected at 47% for 2027, while the software and services sector is expected to post earnings growth of 16.5% this year. Some argue that AI has reshaped market perceptions, and that the high-margin, recurring-revenue business models of the software industry no longer enjoy long-term certainty, prompting the market to reassess their ultimate value.
