Here is a fact that will annoy anyone who has been told NVIDIA is the only way to play artificial intelligence. Over the past year, the Global X Artificial Intelligence & Technology ETF (NASDAQ:AIQ) returned 52%, while the Invesco QQQ Trust (NASDAQ:QQQ) returned 35%. NVIDIA (NASDAQ:NVDA | NVDA Price Prediction) itself, the supposed only AI trade worth making, returned 42% over the same stretch.
The basket beat the chip. And it did so because it owns AI stocks most US investors cannot easily buy on their own. I have been studying AI ETFs for the better part of two years now, and AIQ is the one I keep coming back to when someone asks how to own the AI build-out without betting the farm on a single ticker like NVDA, Taiwan Semiconductor (NYSE:TSM), or Broadcom (NASDAQ:AVGO).
Why AIQ Stopped Tracking the Nasdaq
For most of 2024, AIQ and QQQ traded like twins. Then something changed. According to the 247 Wall St. Best AI ETFs analysis, AIQ “stopped tracking the QQQ so closely and began soundly beating it,” driven by foreign holdings that no Nasdaq tracker can match. SK Hynix returned 211% in 2025, Samsung returned 111%, and Taiwan Semiconductor returned 50%.
Those are the kind of returns that bend a chart. And they are the kind of names that are genuinely hard for a US brokerage account to touch. SK Hynix does not have a US-listed ADR. Samsung trades thinly on OTC markets. You cannot get either through QQQ, you cannot get either through SPY, and you cannot get either through a typical S&P 500 fund because these indexes only hold American companies. AIQ does what an individual investor essentially cannot: it walks into Seoul and Taipei and buys the memory and foundry companies that make NVIDIA chips physically possible.
What Is Actually Inside the Fund
The top two holdings are Samsung Electronics at roughly 5% and SK Hynix just behind it. Taiwan Semiconductor sits near 4%. NVIDIA, the stock everyone obsesses over, is only about 3% of the portfolio. Broadcom, Apple, and Cisco each sit around 3%.
The whole portfolio runs 97 positions, with roughly 40% in international names spanning South Korea, Taiwan, China, Japan, and Europe. You also get smaller weights to names that broad-market funds barely touch: Palantir, AMD, Micron, Oracle, Tencent, Alibaba, SAP, Siemens, and a sliver of speculative AI like CoreWeave, C3.ai, Pony AI, and SoundHound. Net assets sit at nearly $7 billion, and the expense ratio is 0.68%. Not cheap. Not absurd either, given what you are getting.
The Problem With Just Buying QQQ
QQQ is a fine fund. It is also a concentrated bet on US mega-cap tech in a costume. NVIDIA alone is nearly 10% of QQQ. The top 10 holdings combine to 52% of net assets. If you own QQQ, you own a tracker where one in every ten dollars rides on NVDA, and roughly half rides on ten US-listed names.
That worked beautifully in 2023 and 2024. It became less beautiful in May 2026 when NVDA dropped 9% in a month and Broadcom dropped 8% alongside it. Same month, Taiwan Semiconductor was up 6%. The supply chain does not move in lockstep with the customer. AIQ owns more of the supply chain.
The Case for Owning the Basket Over the Chip
I have owned NVDA for over 15 years and I am not selling. But the structural argument for owning AIQ alongside it, or instead of a fresh NVDA position at these levels, is straightforward.
NVIDIA just printed $81.6 billion in Q1 FY27 revenue, up 85% year over year, with Data Center revenue at $75.2 billion and Data Center Networking up 199%. Jensen Huang called it “the largest infrastructure expansion in human history.”
That growth is already priced in. NVDA carries a market cap near $5 trillion, ships zero H20 product to China after the policy changes, and has $119 billion in supply commitments riding on demand staying exactly where it is.
Broadcom is the cleaner second-derivative play. AI semiconductor revenue hit $10.8 billion last quarter, up 143%, with Hock Tan guiding to $16 billion next quarter, more than 200% growth. Taiwan Semiconductor is the only one of these companies that gets paid whether NVIDIA, AMD, or Broadcom wins the next round, and TSM is up 99% over the past year. AIQ holds all of them, plus the Korean memory duopoly that feeds the entire system. That is the picks-and-shovels logic, executed at the index level.
The Tradeoffs
This is a more general technology fund than its name suggests. If you want a pure-play AI infrastructure bet on data center power, cooling, and grid build-out, AIQ will not give it to you. The 0.68% expense ratio is higher than QQQ. The international exposure introduces currency and geopolitical risk, particularly with the China holdings (Alibaba, Tencent, Baidu) and the Taiwan exposure.
The other thing worth knowing: AIQ has also outpaced the more specialized robotics-and-automation funds in the category, which have lagged because their industrial robot holdings have not participated in the AI rerating the way semiconductors have. AIQ’s heavier chip weighting is exactly why it pulled away.
Who Should Actually Own This
If you already hold NVDA and want diversification without giving up AI exposure, AIQ is the cleanest tool I have seen. If you own QQQ and are looking for something more targeted that still gives you mega-cap tech, AIQ overlaps enough to feel familiar while adding the Korean and Taiwanese semiconductor names that have driven the recent outperformance.
If you want a pure NVIDIA bet, just buy NVIDIA. But the entire point of the past year is that the AI trade is no longer one stock. It is a global supply chain, and the foreign names anchoring that chain are not in the index most Americans default to. AIQ owns them. QQQ does not. AIQ owns those names. QQQ does not. The result is 17 percentage points of one-year returns hiding inside an ETF most investors have never heard of.
Top Gaining Stocks
Top Losing Stocks