The PHLX semiconductor index (SOX) — a benchmark tracking major chip companies — dropped 4.3% on July 16, 2026, extending a month of steep losses across the sector. The S&P 500 technology sector fell 1.8% the same day, according to Reuters.

The July 16 decline follows a 4.65% SOX drop on July 7, which trimmed the index's 2026 year-to-date gain to roughly 74%. For the week ending July 3, SOX declined 4.2% as hedge funds sold semiconductor stocks for a fourth consecutive week.

These losses build on volatility from June. On June 23, the semiconductor index fell 7.9%, with Micron Technology shares dropping approximately 13% the same day.

The gap in performance within the chip sector has been striking. The divergence in returns between AI-adjacent chip stocks and semiconductor equipment makers reached about 35 percentage points year-to-date in 2026. Index concentration in AI-adjacent sectors reached record levels as of July 2026 — meaning a larger share of the index's value was tied to a small group of AI-related companies than ever before.

The 74% year-to-date gain on SOX, even after the July 7 pullback, remains substantial. The recent selling pressure has not erased the broader upward move in the index over the first half of 2026. The scale of the recent single-day drops, however, means the index is testing levels that raise questions about how durable the AI-driven rally really is.

What stands out in the data is the 35-percentage-point return gap between AI-adjacent chip stocks and semiconductor equipment makers. That gap is a function of capital flowing into a narrow set of names at record concentration levels. The July 16 sell-off suggests the crowded positioning behind that trade is being unwound.

Hedge funds exiting chip stocks for four consecutive weeks through early July aligns with that unwinding. When positioning becomes concentrated and the market hits a trigger, the exit narrows quickly. Think of it like a crowded theater: when everyone rushes for the same door at once, the squeeze is sharp. The single-day declines on June 23, July 7, and July 16 are consistent with forced de-grossing — funds simultaneously cutting both long and short positions to reduce risk — rather than a fundamental repricing of long-term earnings expectations.

The broader context here is about concentration risk. With index concentration at record levels, the broader market's performance is increasingly tied to a handful of AI-adjacent names. When those names sell off, the spillover into the S&P 500 technology sector is immediate. The 1.8% drop in the S&P 500 technology sector on July 16 is a direct reflection of that linkage.

In my view, the pattern across these three sharp sell-offs — June 23, July 7, and July 16 — looks less like investors reassessing what chip companies will earn over the next decade and more like a mechanical unwind of a crowded trade. That distinction matters. A fundamental repricing would suggest new information about future cash flows. Forced de-grossing, by contrast, is about position sizing and risk management — and it can reverse just as quickly as it began. The fragility on display is real, but so is the 74% gain that preceded it.