1 ETF to Buy and Hold for 2026 and 1 to Avoid

The VanEck Semiconductor ETF (SMH) has been a standout in the past decade with a roughly 31% annualized return and a 0.35% expense ratio, but it is highly concentrated—25 holdings with the top 10 representing 75% of assets and Nvidia alone ~17% (TSMC and Broadcom ~18% combined)—making it an efficient, high-conviction way to play the AI chip cycle but exposing investors to single-stock and sector-concentration risk. By contrast, the Direxion Daily Junior Gold Miners Index Bear 2X Shares ETF (JDST) has collapsed about 90% YTD; as a 2x leveraged inverse on junior gold miners it carries elevated roll/decay and active-management costs (0.89%) and illustrates the large downside and unsuitability of persistent leveraged short products for most portfolios. The practical takeaway: SMH can serve as targeted AI exposure if concentration is managed or offset with broader holdings, while leveraged inverse ETFs should generally be avoided except for short-term tactical use.

Analysis

The VanEck Semiconductor ETF (SMH) has generated a compound annualized return of roughly 31% over the past decade while charging a 0.35% expense ratio. The fund is highly concentrated: it holds 25 names with the top 10 comprising 75% of assets, Nvidia represents about 17% of the portfolio, and Taiwan Semiconductor Manufacturing plus Broadcom together account for ~18%; notably, Nvidia was still the weakest top-10 performer over the past year with a 33% gain.

That concentration makes SMH a high-conviction way to play the AI/chip cycle but exposes investors to single-stock and sector concentration risk despite strong long-term returns. The article positions SMH as efficient for AI exposure for those willing to accept concentration, while recommending that investors seeking diversification add broader ETFs alongside SMH.

By contrast, the Direxion Daily Junior Gold Miners Index Bear 2X Shares ETF (JDST) has declined about 90% year-to-date; it is a 2x leveraged inverse on junior gold miners with a 0.89% expense ratio, and the piece flags leverage, active-management costs and decay as red flags. The coverage therefore frames leveraged inverse ETFs like JDST as unsuitable for most buy-and-hold investors and better reserved for short-term, actively managed tactical trades.

Key Decisions for Investors

• Use SMH selectively as a targeted AI/chip exposure but cap allocation to limit single-stock and sector concentration given top-10 holdings are 75% of the fund and NVDA is ~17% of assets
• Pair SMH with broader, non-leveraged technology or market-cap diversified ETFs to retain AI exposure while reducing idiosyncratic or single-stock risk
• Avoid holding JDST as a long-term position because its ~90% YTD drop, 2x inverse structure and 0.89% fee make it appropriate only for short-term, actively monitored tactical trades with strict position sizing
• Monitor weights of NVDA, TSM and AVGO within SMH and rebalance or trim the ETF holding if any single name materially increases concentration risk or if fund performance diverges from broader semiconductor/AI indicators