The AI trade is facing its most serious test in months. In the two weeks leading into the July 4th holiday, the Nasdaq 100 dropped more than 4% from mid-June highs. Micron Technology (MU) shed roughly 14% in five trading days. Seagate Technology (STX) dropped nearly 18%. SanDisk (SNDK) tumbled close to 20%. Financial media ran with it — “The AI Trade Is Cracking” and “Yet Another Way 2026 Is Looking Like 1999.” The bears were growing louder, and understandably so. But context matters — and the Monday rebound, with both the S&P 500 and Nasdaq bouncing back sharply, is telling.
According to veteran portfolio managers closely tracking institutional flows, last week’s selloff had far more to do with how Wall Street manages books at the start of a new half than any deterioration in AI fundamentals. Portfolio managers trim their biggest H1 winners to harvest gains, rebalance into laggards, and clear inventories before holiday weekends. The hardest-hit stocks — Micron, Seagate, SanDisk — were all AI darlings that had made enormous first-half runs. The simultaneous rotation into beaten-down software names confirms the thesis: this was repositioning, not panic selling. The Atlanta Fed’s recent GDP downgrade added macro noise, but it didn’t change the underlying AI infrastructure demand picture. Earnings from major tech companies — due to report starting around July 23 — will be the next real test.
The actionable point for investors is this: AI earnings are still rising, even as AI stock prices pulled back. When earnings go up and prices go down, forward valuations improve — that’s the opposite of a bubble. Market tops are historically marked by euphoria and FOMO, not by headlines warning of a 1999-style crash. The very fact that bears are this vocal, this early, is itself a contrarian data point. Investors who historically used sharp pullbacks in fundamentally sound AI names as buying opportunities — rather than exit signals — have fared well once earnings season confirmed the bull case. With Micron’s 16 long-term AI supply contracts, Nvidia’s continued data center dominance, and hyperscalers still racing to build AI capacity, the infrastructure buildout shows no signs of reversing. For long-term investors, the question isn’t whether AI will remain a major theme — it’s whether you’re positioned ahead of the next leg up.
