NVIDIA Is the 8th Biggest Holding in the Most Popular AI ETF. What’s Number One?

Global X began trading the Global X LLM ETF (NASDAQ:LLMA) on Sept. 29. It is a fund built around large language models, the technology behind AI chatbots and it was one of five ETFs launched that day. According to its summary prospectus, the fund charges a 0.75% net expense ratio. That is the slice of your balance the fund subtracts each year to cover its costs, whether it gains or loses money. The gross and net fees are both 0.75%, so no temporary fee waiver is improving the price.

The launch lands with a twist. Global X’s existing AI fund — Global X’s AI ETF (NASDAQ:AIQ) — is dominated by memory chipmakers. As of May 31, its number one holding was South Korea’s SK hynix at 7.1% of net assets. NVIDIA (NASDAQ:NVDA | NVDA Price Prediction), the stock most people picture when they hear “AI,” ranked 13th at 2.7% ($293 million).

Memory Chipmakers Top the AI Fund Investors Already Own

Micron (NASDAQ:MU) sat second and was the fund’s largest U.S. company at 5.8% ($626 million). Samsung Electronics held 4.8%. Intel (NASDAQ:INTC) weighed in at 4.2% ($453 million) and networking veteran Cisco at 3.8% ($415 million). Further down, Palantir held 2.2% and IBM 2.0%, across 89 positions.

That tilt has paid off. AIQ is up 27.9% year to date, trading at $65.05 on Sept. 29. Micron shares have rose 274.22% this year and Intel 214.36%, while Nvidia gained 22.45%. Fund assets reached $10.85 billion by May 31, up from $7.8 billion at the end of February.

The memory story rests on a supply squeeze. Micron CEO Sanjay Mehrotra said in June that “DRAM and NAND industry demand continues to significantly exceed industry supply” and that he expects “tight conditions to persist beyond calendar 2027.” If that holds, AIQ’s top-heavy memory exposure keeps working. If chip pricing cools, the fund’s biggest weights take the hit first.

One caution: these holdings are four months old. Micron’s rally alone has likely changed the weights, so check Global X’s site or the latest quarterly holdings filing for current numbers.

What LLMA Holds Remains a Blank Page

LLMA has not yet published a portfolio, so its top holdings, and whether it follows an index or relies on managers picking stocks, cannot be confirmed from launch materials. What is clear: the best-known model builders remain private. CNBC ran a segment on the OpenAI versus Anthropic IPO race the same day LLMA launched, and Bloomberg ETF analyst Eric Balchunas weighed in on Anthropic’s pre-IPO positioning that morning. Until those companies list, the purest language-model businesses stay out of reach for ordinary brokerage accounts, and any public fund on the theme must work through listed companies instead.

A Crowded Field Raises the Bar for New AI Funds

Competition is fierce. 24/7 Wall St. AI analyst Eric Bleecker noted there are now more than 4,300 ETFs in the U.S. versus 4,200 stocks, with more than 1,000 new ETFs launched last year. His blunt take: “All too often, they charge a ton in fees, use bogus marketing, and don’t give you exposure to what you want from an ETF.” Compare LLMA’s fee against AIQ’s current expense ratio on Global X’s fund page before assuming either is the cheaper path to AI.

Who LLMA Targets and Where It Can Go Wrong

LLMA targets investors seeking exposure narrowed down to the language-model slice of AI rather than the broad hardware expansion that drives AIQ. The risks deserve weight:

• No track record: The fund has zero performance history to judge it by.
• Thin early trading: New ETFs often start small, with wide bid-ask spreads (the gap between buying and selling prices). Some funds close if they don’t attract enough assets.
• Thematic swings: Narrow AI funds move fast in both directions. AIQ gained 1.33% over the past week, while Micron dropped 0.29%.

Watch for LLMA’s first holdings disclosure, its asset growth over the next few months and whether its portfolio lives up to its label. Meanwhile, anyone holding an AI fund can run a quick audit of the top ten weights and the expense ratio to confirm they own what they think they own.

Contact [email protected] for any questions or corrections.