After Following Generative AI Every Single Day These 3 ETFs Keep Rising to the Top of My Research
After Following Generative AI Every Single Day These 3 ETFs Keep Rising to the Top of My Research
John SeetooSun, May 31, 2026 at 10:41 AM UTC
0
24/7 Wall St.Quick Read -
Invesco AI and Next Gen Software ETF (IGPT) heavily weights memory suppliers like SK hynix and Micron, capturing AI’s critical infrastructure bottleneck.
Themes Generative Artificial Intelligence ETF (WISE) offers the lowest fees at 0.4% but trails peers with smaller assets and concentrated pure-play exposure.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Roundhill Generative AI & Technology ETF didn't make the cut. Grab the names FREE today.
Tracking generative AI as a daily beat means watching dozens of funds slap the "AI" label on portfolios that look suspiciously like a Nasdaq-100 mirror. After cutting through that noise, three names keep surfacing as genuinely thoughtful ways to own the theme: the Roundhill Generative AI & Technology ETF (NYSEARCA:CHAT), the Invesco AI and Next Gen Software ETF (NASDAQ:IGPT), and the Themes Generative Artificial Intelligence ETF (NASDAQ:WISE).
Each one approaches generative AI through a different lens. CHAT is the actively managed pure-play with the largest asset base. IGPT leans into the global semiconductor and software stack through a rules-based index. WISE is the scrappy low-cost upstart trying to undercut both on fees. The performance gap between them over the past year is wide enough that the choice actually matters.
Why generative AI still earns its own ETF allocation
Broad tech funds give you generative AI exposure in name only. The Magnificent Seven account for so much of the Nasdaq-100 that buying QQQ is more a bet on mega-cap concentration than on the picks-and-shovels build-out happening in foundation models, inference chips, memory, and orchestration software. A dedicated generative AI ETF tilts the portfolio toward the names actually pulling the cart, including memory makers and smaller infrastructure plays that get diluted in a generalist fund.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Roundhill Generative AI & Technology ETF didn't make the cut. Grab the names FREE today.
The catch is that "generative AI ETF" is now a marketing term, not a category with rules. Holdings vary wildly between funds, and so do the returns. That is why the methodology behind each pick matters more than the label on the front of the prospectus.
CHAT: the active pure-play that has dominated the category
Roundhill launched CHAT on May 18, 2023, positioning it as the first ETF built specifically around generative AI as an investable theme. The fund is actively managed, with an investment committee sizing positions by conviction rather than waiting for a quarterly index reconstitution. In a field where the leadership board can flip in a single earnings cycle, that flexibility has earned its keep.
The mechanism connecting CHAT to the theme is straightforward: the portfolio targets companies whose revenue is materially exposed to building, training, or deploying generative AI models. Top holdings include Alphabet, NVIDIA, AMD, and SK hynix, which together cover the model developer, the accelerator vendor, the challenger silicon, and the high-bandwidth memory supplier feeding every training run. That is the entire stack in four names.
Performance has rewarded the approach. CHAT has returned roughly 106% over the past year and is up about 42% year to date. Assets have followed the returns, with the fund now sitting at $1.43 billion in AUM, making it the heavyweight of the group.
The tradeoff is cost. CHAT charges roughly 0.8%, which is the highest fee of the three and meaningfully above what passive thematic funds typically run. Active management is doing real work here, but investors are paying for it. Anyone uncomfortable with a manager-driven approach in a fast-moving theme should look elsewhere.
IGPT: the global infrastructure tilt with a real track record
IGPT is the most misunderstood fund on this list. Its June 23, 2005 inception date is a tell that this is a repurposed vehicle, originally a software-focused ETF that Invesco rebranded toward AI as the theme took off. That history actually works in its favor, because the fund has a longer NAV record than any other dedicated AI product and has compounded about 551% over the past decade through its various incarnations.
Advertisement
The current portfolio tracks the STOXX World AC NexGen Software Development Index, with the fund committing to invest at least 90% of assets in index constituents. Top holdings skew hard toward the hardware layer: SK hynix, Micron, Alphabet, Intel, and NVIDIA. The presence of SK hynix and Micron near the top is the differentiator. Memory is the bottleneck in AI training, and IGPT weights that bottleneck more aggressively than its peers.
The fund has been on a tear, gaining about 89% over the past year and roughly 45% year to date. Assets sit at about $875 million, and the expense ratio of about 0.6% splits the difference between CHAT and WISE.
The tradeoff is index discipline. If the AI narrative pivots away from memory and toward, say, custom inference silicon or vertical software, IGPT will be slower to reposition than an actively managed peer. Investors get rules-based transparency at the cost of agility.
WISE: the low-cost contrarian with a concentrated portfolio
WISE is the overlooked option, and that is exactly why it belongs on this list. Themes ETFs launched the fund on December 8, 2023 with a single competitive lever: price. The expense ratio is roughly 0.4%, less than half of what CHAT charges. For a buy-and-hold investor planning to own the theme for years, that fee gap compounds into real money.
The portfolio tracks the Solactive Generative Artificial Intelligence Index and screens for companies with direct revenue exposure to generative AI rather than tangential beneficiaries. Top holdings include AMD, QuickLogic, NVIDIA, and BZAI. That mix of mega-cap silicon alongside small-cap names like QuickLogic and BZAI gives WISE a barbell character that the larger funds lack.
Performance has been the wart on the story. WISE returned about 21% over the past year and is down roughly 3% year to date, a stark gap versus the other two. The small-cap exposure that defines the fund has been a drag while capital has flowed to the established hyperscalers. Assets remain modest at about $31 million.
The tradeoff is concentration and liquidity. With a portfolio that leans toward smaller pure-plays and an AUM still under $50 million, WISE will be more volatile than its peers and could face wider bid-ask spreads during stressed sessions. The thesis only works if the smaller AI names eventually catch a bid.
Picking the one that fits
The choice between these three comes down to how an investor wants to express the generative AI view. CHAT is the right answer for someone who wants a manager actively rotating through the theme and is willing to pay roughly 0.8% for that discretion. Its returns over the past year justify the fee so far.
IGPT fits the investor who believes the real money in AI flows to the hardware and memory layer, and who wants a rules-based index with a long operational history rather than a fund launched two years ago. The Micron and SK hynix overweights are the entire point of owning this fund.
WISE is the contrarian pick. It is cheaper, smaller, more concentrated in pure-plays, and has badly trailed the others. That is precisely what makes it interesting for anyone who thinks the generative AI rally has been too narrow and that the next leg should broaden into smaller names. It is the fund for investors who would rather pay roughly 0.4% for a portfolio that does not look like every other AI ETF on the shelf.
Act now: the analyst who called NVIDIA in 2010 just named his top 10 AI stocks — and Roundhill Generative AI & Technology ETF didn't make the cut. Grab the names FREE today.
Source: “AOL Money”
