Nasdaq-100 futures slid nearly 2% on July 17, 2026, as a widening global chip sell-off erased more than a trillion dollars in semiconductor market value amid deepening AI capex concerns.
- Nasdaq-100 E-minis fell as much as 2.05% in premarket trading; the iShares Semiconductor ETF slid 3.7% before the open, on pace for a 6.9% weekly loss.
- Taiwan's benchmark index plunged 6.5% — its worst session since "Liberation Day" tariffs — while Japan's Nikkei confirmed correction territory, down more than 10% from its June 25 all-time high.
- Investor confidence in the AI capital expenditure supercycle cracked after SK Hynix signaled slower high-bandwidth memory expansion and Samsung's record profits still failed to lift shares.
Lead
U.S. equity futures deteriorated sharply in early Friday trading on July 17, 2026, as a tech stock crash originating in Asian chip names spread across global markets. Nasdaq-100 futures declined 1.9% to 2.05% at their session low, S&P 500 futures retreated 0.9%, and Dow Jones Industrial Average futures dropped roughly 335 points, or 0.6%. The catalyst: a cascading global chip sell-off that wiped out gains built on months of AI-driven enthusiasm, exposing the fragility of a semiconductor trade that had surged approximately 65% in the first half of 2026.
What Happened
The immediate trigger arrived from Seoul. South Korean brokerage KIS published a second-quarter 2026 profit estimate for SK Hynix that landed 8% below market consensus, citing slow high-bandwidth memory 4 (HBM4) shipments and outsized reliance on premium memory products. SK Hynix shares plunged 15% on the Korea Stock Exchange, dragging the KOSPI down 9% and triggering a brief trading halt. Samsung Electronics fell 7% in Seoul despite reporting preliminary Q2 operating profit that surged nearly nineteen-fold — a sell-the-news reaction that underscored how demanding investor expectations had become.
The contagion moved west. Taiwan Semiconductor Manufacturing (TSMC) reported a record second-quarter net profit that jumped 77% year-on-year and beat forecasts, yet shares fell 7.3% in Taipei after management raised both capital expenditure and annual revenue projections. Investors interpreted higher capex not as a sign of confidence but as confirmation that the AI spending wave was growing riskier and more capital-intensive without a clear return-on-investment horizon. Taiwan's benchmark index fell 6.5%, its worst single-day loss since the April 2025 tariff shock.
Market Reaction
In the U.S. premarket, the iShares Semiconductor ETF (SOXX) slipped 3.7%, while the VanEck Semiconductor ETF (SMH) dropped 5%. Losses spread broadly across the sector: Applied Materials and Lam Research each fell roughly 5%, Intel and KLA Corporation lost more than 4%, and Arm Holdings, Micron Technology, and Nvidia declined in the 3%–4% range. Advanced Micro Devices shed approximately 7%, while Broadcom retreated 2%. Micron's cumulative slide since the tech stock crash began erased an estimated $138 billion in market capitalization. Netflix added to the risk-off pressure after missing subscriber expectations, broadening the sell-off beyond semiconductors.
Japan's Nikkei 225 closed down nearly 3%–5% — with chip-linked names including Kioxia, Screen Holdings, and Sumco among the steepest decliners — confirming a correction of more than 10% from the index's all-time high reached on June 25. China's blue-chip CSI 300 fell 4%.
AI Capex Concerns Take Center Stage
The deeper driver of the global chip sell-off is a re-evaluation of the AI capital expenditure thesis that powered semiconductor stocks to record levels. Investors had priced the sector for a near-perfect demand trajectory: unlimited hyperscaler spending, seamless memory supply absorption, and durable pricing power. That narrative is now under strain on multiple fronts.
SK Hynix disclosed it was redirecting production toward higher-margin DDR5 and away from HBM4 expansion — a supply-chain pivot that markets read as a signal of moderating AI-driven memory demand growth. Separately, reports indicated that Intel's 18A advanced foundry process, critical to the company's strategy for winning external customers, would not reach commercially viable yields until late 2026 at the earliest. Meta Platforms has stated it holds surplus AI infrastructure capacity it is now seeking to monetize — language that suggests the frantic buildout phase may be approaching an inflection point where supply begins to catch up with surging hyperscaler investment.The Federal Reserve's increasingly hawkish posture under Chairman Kevin Warsh compounded the pressure, raising the discount rate applied to long-duration growth stocks and shortening market patience for distant AI payoffs.
Geopolitical Dimension
Escalating tensions in the Middle East added a secondary risk-off layer. Brent crude rose toward the session on geopolitical premium, pulling investor risk appetite lower alongside the tech stock crash and reinforcing the broad move to reduce exposure to crowded positions. The combination of AI capex concerns and geopolitical instability — both amplified by their own feedback loops in derivatives and volatility markets — produced an unusually synchronized sell-off across U.S., European, and Asian equity sessions.
Strategic Context
The July 17 rout is the culmination of a correction that had been building since early July, when chip stocks first cracked following Samsung's initial earnings signal and rising doubts about whether hyperscaler AI capital expenditure would translate into the memory and logic demand investors had already priced in. The sector had risen roughly 65% in the first half of 2026, creating densely crowded positioning and thin tolerance for disappointment. TSMC's strong results — a 77% profit jump — failing to sustain its share price illustrates the degree to which expectations had outrun even exceptional fundamentals.
Outlook
The global chip sell-off reflects a necessary repricing of AI-linked semiconductor valuations after an exceptional first-half run, rather than evidence of a fundamental collapse in demand. The AI infrastructure buildout remains ongoing, and hyperscaler capital expenditure budgets for 2026 remain among the largest ever committed. However, the timeline for demand absorption, yield improvements in advanced processes, and margin sustainability across the memory supply chain is proving less linear than the prior consensus assumed. Markets will look to upcoming earnings from major U.S. chipmakers and hyperscalers for evidence of whether AI capex concerns are cyclical — a pause in an ongoing supercycle — or the leading edge of a more durable deceleration.
Mentioned tickers: QQQ, SOXX, SMH, NVDA, AMD, INTC, AVGO, MU, AMAT, LRCX, KLAC, ARM, TSM, SSNLF, HXSCL, NFLX




