The VanEck Semiconductor ETF (SMH) opened 2026 at $360 and closed at $638 on June 3, 2026 — a 77.13% gain in just over five months (247wallst.com). That kind of move normally plays out over years, not weeks. The Philadelphia Semiconductor Index (SOX) then dropped nearly 20% from its June high amid what's been called AI-related anxiety (Seeking Alpha). The round-trip from euphoric buying to a near-bear-market decline took weeks, not quarters.

Through mid-2026, the iShares Semiconductor ETF (SOXX) returned 169.68% compared to SMH's 135.91% over the same period (ETF.com). A 33.77-percentage-point gap between two funds that most investors would assume track the same sector is unusually wide. SOXX tracks a broader, more equally weighted basket of semiconductor stocks, while SMH's index concentrates more heavily in its top holdings. When the largest chip stocks lead the market, SMH tends to outperform; when a wider range of companies matters more, SOXX pulls ahead. The 2026 spread suggests mid- and smaller-cap semiconductor firms contributed significantly, while the mega-cap names that dominate SMH saw more volatile two-way trading.

The June 3 close at $638, followed by the near-20% SOX pullback, frames the central tension in this story. The same AI-driven demand narrative that powered the surge is now being cited as the reason for the selloff. The Seeking Alpha report attributes the drawdown to "AI-related anxiety" — concern that semiconductor valuations had outrun even the optimistic earnings growth that AI infrastructure spending would imply.

The broader context here is what this means for anyone holding these funds. What matters for positioning is whether the pullback is a healthy reset of sentiment or the first leg of a sustained de-rating — a term for when the market permanently lowers the price it's willing to pay for a given level of earnings. The distinction matters because a reset implies the uptrend can resume, while a de-rating implies a structural shift in how investors value these companies.

For those tracking passive funds, the SOXX-versus-SMH divergence through mid-2026 is the more durable data point. A 169.68% versus 135.91% gap forces a re-examination of what "semiconductor exposure" actually means in a portfolio. The two ETFs deliver materially different risk profiles. SMH's concentration amplifies both upside and drawdowns when the largest names drive price action. SOXX's broader basket smooths that concentration risk but can lag when mega-cap momentum carries the sector. Choosing between them is effectively an active bet on concentration versus breadth, even though both carry the "semiconductor" label.

The 77.13% five-month surge into June 3, followed by the 20% SOX drawdown, is a textbook reminder that parabolic moves in sector ETFs rarely resolve gradually. The speed of the round-trip is itself information: positioning was crowded (meaning too many investors were holding the same trade), and the exit was disorderly. For anyone holding these vehicles in a diversified portfolio, the episode shows that sector ETFs tied to high-beta, thematic narratives can impose drawdowns indistinguishable from those of single-stock positions during sentiment reversals. High-beta, in this context, means the fund tends to move more sharply than the overall market — both up and down.

What remains known from the data: SMH hit $638 on June 3, SOX fell nearly 20% from its June high, and SOXX outperformed SMH by nearly 34 percentage points through mid-2026. What remains uncertain is whether the AI demand cycle underpinning these valuations sustains the earnings growth priced in, or whether the June pullback is the first crack in that thesis.

In my view, the divergence between the two ETFs suggests the market is already differentiating among semiconductor names rather than treating the sector as a single block. That differentiation is worth watching more closely than the headline index level. When two funds labeled "semiconductor exposure" diverge by 34 points, the label itself has become a blunt instrument.