On Friday, July 17, the Philadelphia Semiconductor Index fell 4.3%, confirming a technical bear market for the industry and sending the benchmark measure down 22% from its mid-June record high. The defeat was the harshest test yet for the AI-driven surge that characterized the first half of 2026, wiping out trillions in global semiconductor market value and extending from Wall Street to Asia and Europe.
A Worldwide Epidemic
On Friday, the selloff echoed around the Pacific. Taiwan’s market plunged 6.5% a day after TSMC revealed intentions to invest an extra $100 billion in U.S. fabrication units, while Japan’s Nikkei 225 slumped 4% due to significant selling in chipmakers. Before markets closed for a vacation on Wednesday, South Korea’s Kospi, which had more than doubled in the first half of 2026, fell as high as 7.6%, with SK Hynix and Samsung leading losses.
Memory stocks suffered the most: Western Digital sank more than 9%, Micron fell more than 5%, SanDisk fell more than 12%, and SK Hynix ADR fell more than 13%. The shares of Japanese chipmaker Kioxia, which had increased by around 800% earlier this year, fell 15% on Thursday alone.
Promote the News
Corporate earnings have been strong, which is the paradox at the core of the panic. TSMC raised its full-year spending projection and posted a 77% increase in quarterly earnings on Wednesday, marking its fifth consecutive quarter of record performance. In a similar vein, ASML exceeded projections and increased its yearly revenue forecast to €43 billion to €45 billion. However, neither report improved mood; TSMC shares dropped in premarket trade, and ASML’s stock closed lower after earlier rising 7%.
In what traders referred to as a typical “sell-the-news” dynamic exacerbated by profit-taking following an exceptional run, investors sold down heavyweight chip stocks even as economic data remained encouraging, according to Reuters.
The Question of Capex
Growing concerns about whether hyperscalers’ massive capital expenditures can yield equivalent returns are the root cause of the reversal. The four biggest U.S. AI companies, Meta, Alphabet, Microsoft, and Amazon, all increased their 2026 capital spending forecasts to about $725 billion after first-quarter earnings. As big asset managers reduce leveraged holdings in crowded AI trades, Goldman Sachs said that its high-beta momentum index had dropped 20% since the beginning of July. “The investor deleveraging phase is still ongoing,” the strategy desk at J.P. Morgan cautioned.
Although Morgan Stanley has described the decline as a mid-cycle adjustment rather than a peak, the rate of loss—the sector’s worst weekly performance since the tariff shock in April 2025—indicates that the market’s appetite for AI optimism may have peaked.