5 questions that decide the best AI ETF, from the AIQ ETF to IVES

On July 1, Dan Ives walked out of Wedbush after eight years to start his own firm. That would be a career-page footnote, except there is an exchange-traded fund whose ticker is literally his surname. The Dan Ives Wedbush AI Revolution ETF, ticker IVES, held over $1.1 billion the day after he left, and its benchmark was renamed the Solactive Dan IVES Wedbush AI Revolution Index back in April. So the question a lot of us have typed into a search bar, what is the best AI ETF, has a hidden second half. Best at holding what, exactly, and decided by whom?

What Dan Ives leaving Wedbush revealed about the AIQ-to-IVES shelf

Morningstar counted 48 funds carrying AI or artificial intelligence in their names as of mid-April, and they do not hold the same things. Some own the large language models, some own the power and data centers behind them, and some simply use AI to pick otherwise ordinary stocks. Pull the IVES holdings and the surprise is how ordinary they are: 31 names, close to equal weight, led by Apple at 5.2 percent, then Taiwan Semiconductor, Amazon, Microsoft, Meta, Alphabet, and Nvidia. The top ten come to 47.7 percent. A fund named after a technology analyst is, under the hood, mostly the same seven companies already at the top of the S&P 500.

VettaFi’s Todd Rosenbluth called it key person risk, a measured way of saying the fund’s idea of AI walked out the door attached to a person.

Ives' departure is a poster child example of key person risk that can happen with certain ETFs.

The 80 percent rule is measured against the fund’s own definition of AI

Here is the mechanism, and it is not obvious. When the SEC amended its fund Names Rule (Rule 35d-1) in 2023, it required any fund whose name suggests a focus, including a theme like AI, to keep at least 80 percent of assets invested in line with that name. Eighty percent sounds like a hard floor. The catch is that the fund gets to define the term in its own prospectus, and the 80 percent is measured against that self-written definition. There is no standard dictionary of what counts as an AI company.

So the definition is usually outsourced to an index, and who writes the index tells us most of what matters. AIQ tracks the Indxx Artificial Intelligence and Big Data Index, built by a third party, holding about 84 names with its top ten near 33 percent. THNQ follows an index from VettaFi, which also collects the licensing fee, so the same house that flags key person risk designs and monetizes an AI benchmark. IVES follows an index branded after one analyst. Three funds, three definitions of the same two-letter word, and each 80 percent floor is doing exactly what its sponsor wrote it to do.

Why a Vanguard or Fidelity AI ETF can overlap an index fund by 70 percent

This is where the label costs real money. Broad AI ETFs overlap standard index funds by roughly 60 to 70 percent, because the biggest AI beneficiaries are the same mega-caps that already dominate the S&P 500. A Vanguard, Fidelity, or BlackRock AI ETF bolted on top of an index fund can quietly double a position in Nvidia and Microsoft while charging more to do it. BOTZ runs a 0.68 percent expense ratio; a plain Nasdaq-100 fund like Invesco’s QQQ runs about 0.20 percent, and a total-market index fund runs closer to 0.03 percent.

The AI and big-data segment, 23 funds by one screen, pulled in roughly $8.5 billion of new money over the past year and now runs about $19.6 billion, so this is not a niche rounding error. Paying a thematic fee for exposure we already own is the quiet cost of buying the name instead of the holdings.

Five questions that separate the best AI ETF from an AI label

1. Read the prospectus definition, not the name.

   The 80 percent floor is measured against how the fund itself defines AI, so that one paragraph is the actual product. A vague definition lets the holdings drift anywhere.

2. Find out who writes the index.

   A third party like Indxx, an affiliate that also earns the licensing fee like VettaFi, or one named person like Dan Ives each carry a different conflict and failure mode.

3. Measure the overlap with what we already own.

   If the top ten are Apple, Microsoft, Nvidia, Alphabet, and Meta, the fund is mostly a plain index fund in a costume, and the extra fee buys concentration, not diversification.

4. Read the concentration, not the trailing return.

   A top-ten weight near 48 percent (IVES) versus about 33 percent (AIQ) is a genuinely different bet, whatever last year's number was.

5. Price the methodology risk.

   An index rename, a key person exit, or a reconstitution can change the holdings without changing the ticker, and none of it shows on the fund's front page.

The honest answer to what is the best AI ETF is that the label is marketing and the prospectus is the product. Buying the name means renting a definition we did not write and often did not read, and a portfolio copilot asked the same question will match the ticker to the label, not to the prospectus. The question is not which AI ETF is best. It is whose definition of AI we are comfortable renting, and for how long.