Best 3 AI ETF Picks for the Second Half of 2026

The article argues the AI trade remains favored into 2H 2026, citing strong tech/S&P 500 leadership and continued semiconductor strength despite some first-half exhaustion signals. It highlights three AI/semis-focused ETFs—SMH (concentrated ~25 stocks; ~30% tied to Nvidia + TSMC; ~24x next-12-month earnings), AIQ (more diversified; ~36% semis), and DTCR (about 50% REITs/50% tech)—as ways to participate as leadership shifts toward second-order beneficiaries like Micron and Sandisk and Intel shows a resurgence.

Analysis

This reads more like a flow/positioning note than a fresh fundamental catalyst, which matters because crowded thematic exposure tends to amplify short-term moves while compressing medium-term alpha. The cleanest expression of the AI trade is no longer just the index-heavy chip complex; the opportunity is shifting toward breadth, with memory and infrastructure names likely to capture incremental spend before the mega-cap leaders re-rate again.

SMH is still the highest-beta vehicle, but it is also the most fragile because a few names effectively dominate its outcome. That makes the better second-order trade a relative-value basket in beneficiaries that sit one step down the stack: MU and SNDK on memory content, and potentially INTC only if its turnaround can translate into actual share gains rather than headline momentum. If AI capex broadens, the market should reward operating leverage in these laggards more than multiple expansion in the already-owned leaders.

DTCR is the most interesting contrarian angle because the real bottleneck may be power, land, and interconnect capacity, not chips alone. The REIT sleeve gives it a different earnings driver, but also a higher sensitivity to rates and cap-rate expansion; that makes it a better trade only if the 10-year yield is stable to lower. Consensus is probably underestimating how quickly a broad tech de-rating would rotate capital from concentrated semis into infrastructure proxies, but overestimating how durable the current momentum is if NVDA/TSM execution merely normalizes.

Key Decisions for Investors

• Long MU / short SMH for 1-3 months: express the view that AI spend is broadening into memory faster than it is re-accelerating the mega-cap index weights; abandon if MU guides down or NVDA/TSM reassert leadership.
• Buy AIQ on pullbacks as a lower-beta breadth trade for 6-12 months; expect it to underperform SMH in a straight-line momentum tape but outperform if semis de-rate while software/hardware participation improves.
• Hold off on DTCR unless long rates stabilize or decline; the upside from data-center scarcity is real, but the REIT half makes the risk/reward poor if Treasury yields back up 50-75 bps.
• Treat INTC as a tactical catalyst trade only, not a core AI holding; add only on evidence of real product or foundry traction in the next earnings cycle, otherwise fade strength.
• If SMH rallies another 5-7% without breadth confirmation, consider taking profit and rotating into MU/SNDK or AIQ; that would signal the trade is becoming too concentrated and vulnerable to a sharp unwind.