AI-themed infrastructure exchange-traded funds ( ETFs ) are experiencing increased demand from Asian investors seeking a more focused but still diversified way of investing in the artificial intelligence trend.

At present, most AI-themed ETFs are heavily weighted towards mega-cap stocks, particularly the Magnificent 7 ( Apple, Microsoft, Google parent Alphabet, Amazon.com, Nvidia, Meta Platforms, and Tesla ), which dominate most of the major stock market indices, especially the S&P 500 and Nasdaq 100, due to their massive market capitalization and strong performance.

But the shifting performance of the Magnificent Seven in recent weeks has prompted investors in AI-themed ETFs to explore opportunities beyond these mega stocks to avoid concentration risk.

Another risk that comes with AI-themed ETFs is keeping updated with the very fast-paced developments in the AI sector as exemplified by DeepSeek, a Chinese tech start-up that has rapidly emerged as a significant player in the global AI landscape for developing cost-effective, high-performance AI models that challenge established industry leaders. DeepSeek is still an unlisted company and, as such, is not a direct holding in any ETF.

Regulatory risk

Yet another risk for investors in AI-themed ETFs is regulation. The European Union, for example, has come up with the AI Act, which seeks to ensure ethical and safe AI deployment. Given its broad scope and influence, the legal framework may introduce political risks stemming from increased compliance pressure, evolving regulatory landscapes, potential market access challenges, and geopolitical tensions. All these pressures can impact the performance and growth prospects of AI companies within these ETFs.

One way to invest in AI-themed stocks and avoid overconcentration on mega stocks is through ETFs that invest in AI infrastructure.

At present, there are already four AI-themed infrastructure ETFs available to Asian investors. These are the Tiger Global AI Infra Active ETF and Global X China Robotics and AI ETF, both offered by Mirae; the Kodex US AI Power Core Infrastructure ETF from Samsung Asset Management; and the Plus Global AI Infrastructure ETF offered by Hanwha Asset Management.

The growing demand for AI-themed infrastructure ETFs prompted Mirae to launch on February 10 the Global X AI Infrastructure ETF, which offers exposure to 30 AI infrastructure companies across North America, Asia, and the eurozone, focusing on sectors like energy infrastructure, data centre operations, and raw materials.

AI infrastructure spans multiple interconnected systems, including energy supply, thermal management, and network operations, such as data centres and utility companies that provide the energy systems needed to power the critical components of AI, according to Lizzie Liu, ETF research analyst covering AI ETF infrastructure for Mirae Asset Global Investments.

Diversified exposures

In terms of diversification, the Global X AI Infrastructure ETF’s asset allocation includes three categories of AI-themed infrastructure assets. These are data centre operation infrastructure, power and energy infrastructure, and raw material supply essential for AI technologies. The ETF tracks the Mirae Asset AI Infrastructure V2 Index.

“We have exposure to the physical infrastructure and we have diversified exposures to different sub-themes under the big AI infrastructure industry. Those themes not only benefit from AI development; they also have other growth drivers, like benefiting from other major global trends such as energy or green energy transition. For example, green energy transition could also benefit raw material sectors like the uranium in nuclear power,” Liu says.

To avoid concentration risks, the Global X AI Infrastructure ETF has a 5% cap for single securities. In terms of volatility risk, the annual average volatility is around 19.5%, which is relatively lower when compared to Mirae’s other AI-themed ETFs.

“The rapid advancement of artificial intelligence is driving the industry into a new era, with growing emphasis on the infrastructure needed to support AI development. It's the backbone of further AI development and innovation.  We believe that without the infrastructure part, it's hard to meet the requirements of AI, like power or energy, or data centre computing,” Liu says.