A position in the Invesco PHLX Semiconductor ETF (NASDAQ:SOXQ) bought at the close on May 29, 2025 for $38.08 was worth $101.03 on May 29, 2026, a 165% twelve-month gain on a low-cost, plain-vanilla index fund that anyone with a brokerage account could have bought. Run that back to inception in June 2021 and the same fund, starting at $24.17 on June 11, 2021, has delivered a 318% total move, unusual compounding for a 30-stock market-cap-weighted index with a 0.19% expense ratio.
The recent slope is even steeper than the long arc. SOXQ is up 81% year to date through May 29, up 25% in the past month alone, and tacked on another 5% in the final week of May. The mid-April through late-May run is the kind of move that draws screenshots into group chats and an honest question behind them, which is whether this is real, whether you missed it, and whether the same trade works from here.
The arithmetic, in dollars
A clean way to see what happened is to anchor on round principal. Ten thousand dollars put into SOXQ at the December 31, 2025 close of $55.71 is worth about $18,100 today on price alone, before counting the $0.0706 quarterly distribution the fund has been paying out across March and April. Ten thousand dollars put in at the May 29, 2025 close of $38.08 is worth about $26,500 a year later. Ten thousand at the June 11, 2021 inception price of $24.17 is worth about $41,800.
The dividend matters less for SOXQ than it does for, say, a covered-call income fund, because the distribution is small relative to the price move. The honest framing is that essentially all of the return here is capital appreciation in a market-cap-weighted basket of semiconductor stocks, and the headline numbers do not require a cherry-picked entry date to be true. Buying inside the March 2026 selloff would have done even better, but you do not need to have timed that to be sitting on a triple-digit one-year number.
What actually did the work
SOXQ tracks the PHLX SOX Semiconductor Sector Index, a market-cap-weighted basket of the 30 largest U.S.-listed semiconductor companies. The structure matters because cap weighting concentrates the fund in whatever the market has decided is winning. The top 10 holdings account for 59% of assets, anchored by NVIDIA, Broadcom, AMD, and Micron. When AI accelerator demand is the dominant story in equity markets, this fund is positioned to harvest it almost mechanically, because the names doing the harvesting are the ones the index keeps buying more of.
The macro setup behind the move is unusually clean. Worldwide semiconductor revenue in the first quarter of 2026 came in at $298.5 billion, a 79.2% year-over-year jump, with U.S. sales alone up 83.1% year over year to $101.4 billion. Average selling prices climbed 57.1% year over year as unit volumes barely moved, which tells you the mix shifted toward expensive AI logic and high-bandwidth memory rather than commodity chips. The global market itself reached US$796 billion in 2025, with the largest gains in logic and memory, exactly the segments SOXQ is heaviest in.
On top of that, hyperscaler capex stayed locked in. Goldman Sachs framed the 2026 setup as an economy where "the AI capex boom driving business and investment activity" is propping up growth even as parts of the labor market and consumer spend soften. PineBridge sees datacenter equipment growth "essentially locked in for the next four to five years" at around 25% annual growth, constrained more by electrical transmission and distribution than by chip supply. That is the durable part of the story.
The texture of the recent rally is worth naming directly. The PHLX Semiconductor Index ran a 17-consecutive-day winning streak with gains exceeding 40% in April, which was unprecedented in the index’s 32-year history. Intel’s first-quarter results landed inside that window, helped along by Elon Musk’s confirmation that Tesla would use Intel’s 14A manufacturing process for AI chips, and AMD’s server CPU market share exceeded 40% with firmer pricing, both of which fed straight into the cap-weighted basket. Volatility cooperated as well. The VIX peaked at 31.05 on March 27, 2026 and has since drifted back down to 15.74 on May 28, putting current readings in the 19th percentile of the trailing 12-month range. Semiconductors do their best work in exactly this kind of low-vol, risk-on tape.
What has to keep being true
This is where the article earns its keep, because the next 12 months will not look like the last 12 months unless several specific conditions hold. The first is hyperscaler capex. The bull case for SOXQ from here rests on Microsoft, Amazon, Google, and Meta continuing to spend on AI infrastructure at the pace investors have priced in. Watch the quarterly capex guidance from each of the four in their earnings reports. Any one of them lowering 2027 capex would matter. All four doing it would end the trade.
The second is the AI-related capex line in the U.S. economy, which PineBridge has tracked rising as a share of GDP through 2025 data through September 30. This is the macro version of the same question. If AI capex stops accelerating as a share of GDP, the cap-weighted index loses its biggest tailwind. The data is available from Wolfe Research and the BEA’s investment categories, and it updates quarterly.
The third is concentration risk, which is the part of this story most likely to bite. With 59% of assets in the top 10 holdings, SOXQ is functionally an NVIDIA-Broadcom-AMD-TSMC vehicle with 20-plus names along for the ride. A bad NVIDIA quarter, a renewed U.S.-China export restriction in the spirit of the $8 billion China drag NVIDIA faced in 2025, or a real software-side disappointment that calls AI monetization into question would all hit the fund harder than its "30-stock index" label implies. Seeking Alpha’s January read on the fund was direct about this, calling it "highly volatile, richly valued, concentrated portfolio suitable only for risk-tolerant growth investors", and that was before the spring rally added another leg.
The fourth is volatility itself. The VIX at 15.74 is closer to complacency than to fear, and semis tend to give back gains quickly when the fear gauge climbs back above 20. The March spike to 31.05 is the recent reminder. A move back into that range, on any of the policy or geopolitical catalysts already sitting on the calendar (the EU’s Chips Act 2.0 published May 27, 2026, ongoing tariff posture, China export controls), would be the leading indicator that the regime is changing under the fund’s feet.
The honest read from here
SOXQ did what it did because a structural tailwind (AI infrastructure buildout), a favorable macro tape (low VIX, risk-on flows), and a cap-weighted index hardwired to overweight the winners all lined up at the same time. The first of those is the most durable. The second is regime-dependent and currently helpful, with the VIX in the lower 19th percentile of its trailing year. The third is a feature when the trade is working and a liability when it stops.
What it means in practice is that the same fund at $101.03, after a 165% one-year move, is a different instrument than it was at $38. The mechanism that drove the gain is still intact, but the price you pay to rent it is much higher, and the asymmetry has narrowed. The things to watch are hyperscaler capex guidance, the quarterly WSTS data, and the VIX tape, rather than SOXQ’s chart. As long as capex guidance keeps drifting up, the WSTS number keeps printing year-over-year growth in the high double digits, and the fear gauge stays in the teens, the mechanism is alive. When one of those breaks, the cap-weighted basket will tell you about it faster than anything else, because it was built to.
Top Gaining Stocks
Top Losing Stocks