AI-centric ETFs and the traders who love them…

“The measure of intelligence is the ability to change.” — Albert Einstein

There was a great scene in the film Jerry Maguire in which Tom Cruise plays the titular role of a sports agent. He’s in a tight spot and desperately needs to sign a mercurial wide receiver if he wants to stay in business. The player makes Cruise’s character yell “show me the money” an absurd number of times as a way to prove he’s all in for him.

Traders are willing to shout out similar platitudes in endless fashion, as long as the financial vehicle that’s the object of their affections keeps delivering bottom-line black numbers, preferably in the deep double digits.

Which brings us to AI.

For the last several years, both retail and institutional investors have followed artificial intelligence with an enthusiasm that borders on the devotional. Whether that devotion is fully warranted, partially warranted, or a financial fever dream of the first order remains an open question, at least for the moment. What is considerably less arguable is that AI-centric ETFs have arrived, multiplied, and in several cases delivered returns that have made early adopters quite happy with their investment life choices.

However the current debate over artificial intelligence ultimately resolves itself — genuine paradigm shift or a collective miscalculation that ultimately dissipates into some sort of tech version of a will-o’-the-wisp — one thing seems reasonably clear: the investment community has already voted, and the vote was not even close.

The first ETF explicitly targeting artificial intelligence themes was AIQ (Global X Artificial Intelligence & Technology ETF), which launched in May of 2018. Since then, AIQ has posted an average annual return of approximately 19.6%, and its trailing twelve-month return currently sits somewhere in the neighborhood of 49–52%. For a fund that arrived on the scene before most investors had strong opinions about large language models, that can only be viewed as a fairly impressive report card.

The broader AI ETF landscape has expanded considerably in the years since AIQ’s debut. The dominant players by AUM remain AIQ, now approaching nearly $10 billion, and BOTZ (Global X Robotics & Artificial Intelligence ETF), at approximately $3.8 billion. Both carry an expense ratio of 0.68% and have delivered trailing twelve-month returns that compare favorably to the benchmarks — AIQ’s 49–52% clearing QQQ’s approximately 37% and SPY’s approximately 27% over the same period. Pretty sweet for those who managed to catch the wave while it was still just a relative swell, though the high levels of concentration risk in AI-centric funds tend to make those comparisons look better in calmer markets than turbulent ones. BOTZ, it should be noted, has seen its own trailing twelve-month return compress more recently, currently running in the range of approximately 11–30% depending on the measurement window — a reminder that not all AI-themed funds move in lockstep, even when they share a general thesis.

A look under the hood of AIQ reveals approximately 90 holdings spread across semiconductors, software, and cloud infrastructure, with the top five positions currently led by SK Hynix at 6.14%, Micron Technology at 4.74%, Samsung Electronics at 4.52%, Intel at 4.37%, and Advanced Micro Devices at 4.35%. The top ten holdings account for roughly 41% of total assets — still meaningfully below the technology category average of 53%, which gives AIQ a somewhat broader base than many of its peers. Nearly 30% of the portfolio sits outside the U.S., with Korean and Taiwanese chipmakers occupying a central role.

BOTZ takes a different angle on the same underlying thesis, leaning heavily toward industrial robotics and automation hardware, with Keyence — not Nvidia, as many assume — currently its largest holding at around 9.38%, followed by ABB, FANUC, Nvidia, and Intuitive Surgical rounding out the top five.

The bull case for AI as a macroeconomic force is not hard to make. Capital expenditure commitments from the major hyperscalers are running into the hundreds of billions annually, semiconductor demand shows no signs of abating, and the infrastructure build-out powering the current AI cycle is rippling through supply chains in ways that are still only beginning to show up in earnings reports. The productivity argument — that AI will ultimately do for knowledge work what industrial automation did for manufacturing — remains more thesis than demonstrated reality, but the market has generally decided not to wait for the proof before pricing it in.

Bubble-ish, one might even say, provided there’s some Kool-Aid left in the glass of the drinker. And that’s where the bear case makes its way to the fore.

The bearish argument is straightforward enough: the spending has been front-loaded in spectacular fashion, while the monetization of all that investment remains, for the most part, a work in progress. Enterprise adoption has been slower and more uneven than the promotional materials would suggest, and the revenue that might justify the scale of capital being deployed is still largely a projection rather than a result.

When the gap between spending and return gets wide enough, the word that tends to surface — quietly at first — is, and here’s that concept again, “bubble.” The dot-com era gets invoked here with some regularity and with some justification, not because AI will necessarily end the same way, but because the pattern of front-running transformative technology with capital that outruns demonstrated value is one Wall Street has visited before. Though, admittedly, and by most valid metrics, there has never been a tech product with more disruptive potential than AI.

There is also the concentration problem. The AI thesis, as currently priced into the market, depends heavily on a small number of mega-cap companies continuing to deliver both the infrastructure and the narrative. Should that narrative develop any meaningful cracks — an earnings disappointment, a regulatory intervention with real teeth, or simply a collective investor decision that the wait for commercial payoff exceeds their patience — the unwinding could move considerably faster than the build-up.

On a practical level, then, the valuation questions are worth factoring into any trader’s portfolio, and the case for measured rather than maximal exposure seems, at the very least, worth noting and putting on a continual re-evaluation loop.

Bottom line: for investors who find the AI thesis compelling but prefer not to bet the entire portfolio on it, AIQ and BOTZ remain the most liquid and battle-tested vehicles for getting exposure to the theme without the white-knuckle volatility of individual stock selection. For those with a genuine appetite for risk, leveraged plays like GDXU and NUGT are available — though as with any leveraged product, it’s worth recalling a catchphrase from another pre-2000 film: Ferris Bueller’s Day Off; “Buckle up, buttercup.”