SMH vs SOXQ vs SOXX: Which Is the Better Pick in 2026?

Overview

SMH, SOXX, and SOXQ are all low-cost, cap-weighted index ETFs that own the U.S.-listed semiconductor sector — the same universe of chip designers, foundries, and equipment makers. They are not strategy bets against each other; they are three wrappers around slightly different indexes at slightly different price points. SMH (VanEck) tracks the MVIS US Listed Semiconductor 25 Index and is by far the largest and most liquid. SOXX (iShares) is the long-tenured veteran, tracking the ICE Semiconductor Index. SOXQ (Invesco) is the cost leader, tracking the PHLX SOX Semiconductor Sector Index at the lowest fee and lowest share price.

How they differ

The clearest difference is cost. SOXQ charges 0.19%, versus 0.35% for both SMH and SOXX — roughly a 16-basis-point edge that compounds over long holding periods. For a buy-and-hold sector allocation, SOXQ is the cheapest way in.

The second difference is index construction, and it matters more than the fee. SMH's MVIS index holds about 25 names and is the most top-heavy, concentrating weight in the very largest chip stocks (think NVIDIA and TSMC). SOXX (ICE index) and SOXQ (PHLX SOX index) each spread across roughly 30 holdings with somewhat more balanced weighting. SMH's mega-cap tilt is what has driven its recent performance edge during the AI rally — and it is also what concentrates more single-name risk.

The third difference is scale and access. SMH holds about $65B, SOXX about $37B, and SOXQ about $2.6B. SMH's size gives it the tightest spreads and the deepest options market. SOXQ's much lower share price (near $105 versus roughly $600 for both SMH and SOXX) makes it easier to size in small-dollar accounts and cheaper to trade in options. None of the three is an income vehicle — trailing yields sit between 0.18% and 0.29%.

Who each is best for

SMH: Investors who want the most liquid semiconductor ETF with a mega-cap tilt and the deepest options market, and who are comfortable with a top-heavy, concentrated portfolio.

SOXX: Investors who want a slightly broader, ~30-holding index from iShares with the longest live track record (inception 2001) and don't mind paying the same 0.35% as SMH.

SOXQ: Cost-conscious, buy-and-hold investors who want the cheapest semiconductor index exposure (0.19%) and a lower per-share entry point, and who accept the smallest asset base of the three.

Key risks to know

• Sector concentration and cyclicality: All three are pure semiconductor plays with no diversification outside the group. The industry is cyclical and capex-driven, so a slowdown in AI spending or a chip down-cycle would hit all three together.

• High beta and drawdown risk: These funds carry high market betas (roughly 2x the broad market), so sharp, correlated drawdowns are the norm in a sector selloff — SOXX has run the highest beta of the three.

• Single-name concentration at SMH: SMH's 25-name, top-heavy index means a handful of mega-cap holdings dominate returns. That has helped during the AI rally but magnifies the impact if a top holding stumbles.

• Smaller scale at SOXQ: At roughly $2.6B, SOXQ is far smaller than SMH and SOXX. It is liquid enough for most investors, but spreads can be wider and depth thinner than SMH's during stress.

• Index tracking divergence: Because the three follow different indexes (MVIS 25, ICE, and PHLX SOX), their returns will drift apart over time even though the holdings overlap heavily — the choice of benchmark, not the fee, drives most of the long-run difference.

Bottom line

These are three low-cost index wrappers around the same sector, and they will move largely in lockstep. SMH is the liquidity-and-mega-cap choice: biggest, deepest options, and a top-heavy tilt that has led during the AI rally with more single-name risk. SOXX is the veteran ~30-holding iShares option at the same 0.35% fee. SOXQ is the value pick — the cheapest at 0.19% with the lowest share price — best for cost-sensitive buyers who accept its smaller asset base. The real decision is index construction and cost, not strategy; check the baked total-returns section for how those differences have actually played out. Past performance does not guarantee future results.