Cheapest AI ETF: Low Cost AI ETFs Ranked and the Alternative

The cheapest AI ETF with an artificial intelligence mandate is the WisdomTree Artificial Intelligence and Innovation Fund (WTAI) at 0.45% a year, followed by the iShares Future AI & Tech ETF (ARTY) at 0.47%. The best-known names cost more: Global X BOTZ and AIQ charge 0.68%, and Roundhill CHAT, ARK ARKQ and the Dan IVES Wedbush AI Revolution ETF charge 0.75%. If you only want the companies that dominate AI, broad funds are far cheaper: SOXX 0.34%, SMH 0.35% and Vanguard's VGT 0.09%. Cheapest of all is owning the stocks directly, where there is no expense ratio.

Fee is only half the comparison, and the funds make that hard to see. Two ETFs with AI in the name can have almost nothing in common, so the cheapest one is only a bargain if it holds what you meant to buy. Below, every major US-listed AI ETF is ranked by its expense ratio, with what that costs you in dollars and what the fund actually owns, as its issuer published it.

Expense ratios, holdings counts and top positions come from each issuer's fund page, holdings file or fact sheet, read on September 27, 2026. We make index construction and backtesting software; we do not sell funds or give investment advice.

Cheapest AI ETFs ranked by expense ratio

And the three cheaper funds people use as a stand-in for AI:

What is the cheapest AI ETF?

WTAI, at 0.45%. It tracks the WisdomTree Artificial Intelligence and Innovation Index across 57 holdings, and at the end of June its top five were Micron, Samsung, Sandisk, Kioxia and NVIDIA. That is a heavy tilt toward memory and storage chips, which is a specific bet on AI hardware demand. ARTY, two basis points more expensive, is the closest substitute: 49 holdings on the Morningstar Global AI Select Index, led in late August by TSMC, NVIDIA, Micron, AMD and Super Micro, each between 4% and 5%. It used to be the iShares Robotics and Artificial Intelligence ETF, ticker IRBO, before iShares repurposed it.

If you are willing to call a semiconductor fund an AI fund, SOXX at 0.34% and SMH at 0.35% undercut both. They are also much larger funds: SMH held about $75 billion on September 25, against about $670 million in WTAI at the end of June. The catch is concentration: SMH had 19.3% in NVIDIA alone on September 24.

Are cheap AI ETFs as good as the expensive ones?

On fee, yes; on holdings, they are simply different funds. The 0.75% products are not charging more for a better version of the same thing. CHAT and ARKQ are actively managed, so part of the fee pays a manager to pick. IVES follows a 30-stock index built around one analyst's AI list. BOTZ, at 0.68%, is the one most likely to surprise a buyer: on September 25 its four largest positions were Intuitive Surgical, Fanuc, Keyence and ABB, all robotics and automation companies, with NVIDIA fifth at 7.8%. AIQ, the other 0.68% Global X fund, owns Palantir, Microsoft, SpaceX, Meta and Intel at the top, each between 3% and 4%.

So paying less is never the mistake. Buying a fund whose definition of AI does not match yours is. Put the holdings next to each other before comparing fees; the full table with every fund's five largest positions is on our AI stocks index builder page.

How much do AI ETF fees cost over 10 years?

More than the yearly figure suggests, because the fee compounds. Take $25,000 growing 8% a year before fees. After ten years at 0.45% you would have about $51,600. At 0.75% you would have about $50,060. The difference is roughly $1,530, for funds that differ by 0.30% a year. At VGT's 0.09% the same money grows to about $53,490, nearly $3,400 more than the most expensive AI funds. Those are illustrations of fee drag, not forecasts of returns.

Does Vanguard have an AI ETF?

No. Vanguard does not run a fund dedicated to artificial intelligence. VGT is the one people reach for: 321 technology stocks for 0.09%, with NVIDIA, Apple and Microsoft making up about 39% of the fund at the end of June. It gives you the AI leaders at the lowest fee in this article, but it is a tech sector fund, and you also get everything in tech that has nothing to do with AI. If you already hold an S&P 500 or Nasdaq-100 fund, much of VGT's top end is already in your portfolio.

The cheapest alternative: build the AI basket yourself

Owning the stocks directly has no expense ratio at all. At a broker with fractional shares you can buy 20 or 30 AI names at exact weights with a few hundred dollars. Fidelity Basket Portfolios is built for exactly this: baskets of up to 50 stocks or ETFs, from $1 per position, traded in one order for a flat $4.99 a month. That is $59.88 a year however large the basket gets. Against WTAI's 0.45%, the flat fee is cheaper once the basket passes about $13,300. Against a 0.75% fund, the break-even is about $8,000.

The bigger reason to build is control, not cost. You decide which layers count as AI (chips, cloud, software, or the power and data center build-out), whether the mega caps get capped, and which names stay out. Before adding a company, it is worth taking a few minutes to research each AI stock's fundamentals rather than buying it because it shows up on a list. Then there is tax: each stock in a basket has its own cost basis, so a single AI name that falls 30% can be sold for a loss while the rest of the basket is up. An ETF hides that loss inside the fund.

The work you take on is design and upkeep. That is where we come in: you can build your own AI index fund in Indexes, set equal, market cap, inverse volatility or custom weights, and backtest it against the S&P 500 and the Nasdaq-100. To see whether a cheap ETF would have done the job, build a second index holding just that fund and compare the two over the same years, including 2022, when AI and chip stocks fell hard. If the ETF tracks your design closely, buy the ETF and skip the upkeep. If not, you know exactly what you are paying 0.45% to 0.75% to miss.

Which cheap AI ETF fits which view?

• You think the money lands with chipmakers: SMH or SOXX at about 0.35%. Pick SOXX if you want less NVIDIA concentration.
• You want a dedicated AI fund at the lowest fee: WTAI at 0.45% for a hardware and memory tilt, ARTY at 0.47% for a more even spread across chips.
• You want the AI leaders and do not mind broad tech: VGT at 0.09%, as long as you check the overlap with what you already own.
• You want robotics and automation: BOTZ is honest about that in its name, at 0.68%.
• None of them holds your mix: build it, backtest it, and hold the stocks at your broker. The same approach works for any theme, as our thematic ETFs vs building your own basket comparison lays out.

Expense ratios change and funds get renamed, as ARTY shows, so confirm the current figure on the issuer's page before you buy. Backtested results are hypothetical, and nothing here is a recommendation to buy any fund or stock. For broader themes than AI, the thematic investing builder works the same way.