The iShares Semiconductor ETF (NASDAQ: SOXX) has delivered a staggering 118% return over the past 12 months, dividends included, stunning even seasoned chip investors.
To put that figure in perspective, the fund has averaged 14.2% annually since launching on July 10, 2001, meaning the past year compressed roughly six years of typical gains into one.
That acceleration was driven largely by explosive demand for graphics processing units and the broader infrastructure buildout supporting artificial intelligence applications across industries.
The fund now holds approximately $43 billion in assets, making it one of the largest semiconductor-focused investment vehicles available to everyday investors.
Its three biggest positions are Nvidia (NASDAQ: NVDA) at roughly 9% of assets, Advanced Micro Devices (NASDAQ: AMD) at 8.2%, and Broadcom (NASDAQ: AVGO) at 8.2%, with all three stocks riding the same AI spending cycle.
Together, those three names account for about a quarter of the entire fund, meaning this is a highly concentrated bet on continued AI infrastructure growth rather than a broadly diversified semiconductor play.
Digging into the fund’s 25-year history reveals only two prior 12-month periods that came close to matching the current run, though neither quite reached it.
The 12 months ending January 2004 produced about 97% as chip stocks rebounded sharply from the dot-com collapse, and the 12 months ending March 2021 generated roughly 109% during the pandemic electronics boom and first wave of chip shortages.
At its peak in June, the fund’s trailing-year return had climbed to approximately 170%, well beyond anything in its history, before pulling back to the current 118% figure.
Both of those earlier banner years were followed by disappointing second years, with the fund falling roughly 23% in the year after the 2004 peak and returning only about 12% in the year following the 2021 surge before dropping 35% in 2022.
The five-year outcomes after those two episodes diverged sharply, however, with the post-2021 period ultimately delivering about 142% in total, or roughly 19% annually, because AI demand arrived and created a larger demand wave than the pandemic ever produced.
What separated those two outcomes was not the scale of the initial run but whether a new, sustained wave of end-market demand materialized to justify the elevated starting prices.
Today’s fund trades at approximately 67 times earnings, a valuation that already prices in continued compounding of AI infrastructure spending at a significant scale for years ahead.
The fund carries 30 holdings and a 0.33% expense ratio, making it a structurally efficient vehicle, but the uncomfortable variable remains the starting valuation after such an extraordinary 12-month run.
As the historical record shows, the year following both prior surges came in well below the fund’s 14.2% long-run average, and there is no guarantee the AI cycle will prove more durable than the booms that preceded it.
