So far, 2026 has belonged to semiconductor stocks, and the numbers are staggering. In the second quarter alone, the sector surged more than 87 percent, marking its best quarterly performance on record. The rally has been powered by relentless demand for artificial intelligence infrastructure, lifting chipmakers, suppliers, and memory companies across the board. The two largest U.S.-listed semiconductor ETFs, SMH and SOXX, hold roughly $70.6 billion and $45.9 billion in assets respectively, and both have delivered exceptional returns across multiple time frames.
The performance gap between the two funds largely comes down to how they are weighted. SOXX holds the edge over the past year, gaining about 127 percent compared with 102 percent for SMH. But SMH has been the stronger performer over the prior three- and five-year periods, rising 279 percent versus 227 percent and 362 percent versus 283 percent. That longer-term outperformance reflects SMH’s heavier concentration in industry giants like Nvidia, Taiwan Semiconductor, and Broadcom. Its Nvidia allocation is more than double that of SOXX, which has helped propel long-term returns even as SOXX’s broader allocations to Intel, Micron, and AMD have given it an edge more recently.
Analysts remain broadly bullish on both funds, though expectations suggest plenty of room to run may already be priced in. The consensus price target on SMH sits around $749, implying roughly 30 percent upside from current levels, while analysts have a target of approximately $692 on SOXX, suggesting about 27 percent upside. Valuations within the Semiconductor Index have historically peaked near 45 times forward earnings, with support often emerging around 15 times over the past decade and closer to 18 times over the past two years. Earnings expectations for this year and next have also climbed sharply, reflecting just how much confidence investors are placing in continued AI-driven growth.
Still, the risks facing semiconductor investors are substantial. The sector’s fortunes depend heavily on hyperscaler spending from companies like Microsoft, Amazon, Alphabet, and Meta, meaning any slowdown in AI infrastructure budgets could trigger a rapid reset of both demand projections and earnings estimates. Layered on top of that are export controls, tariffs, supply-chain disruptions, shifting pricing power, and the constant threat that rapid technological change could reshuffle market leadership without warning.
Semiconductors remain one of the most compelling corners of the AI trade, but they are not suited for every investor. Valuations sit below prior peaks yet well above historical lows, and sentiment can turn quickly. The long-term opportunity appears intact, but anyone buying in at these levels needs to weigh recent gains against the very real potential for sharp pullbacks, softer-than-expected demand, and elevated expectations that leave little room for disappointment.