The artificial intelligence investment story is evolving, and ETF investors' transaction patterns in the first half of 2026 showed that the market is now looking beyond software giants and chip designers toward the physical infrastructure that powers AI.
Fresh first-half ETF flow data sourced by VettaFi show investors are shifting capital into themes such as memory chips, power infrastructure, cooling systems or HVAC, electrification, and data centers, signaling that the market is broadening from simply betting on AI adoption to investing in the ecosystem required to support it.
The trend comes despite macroeconomic headwinds, including sticky inflation, geopolitical uncertainty, and a shifting Federal Reserve policy stance, underscoring investors' conviction that AI infrastructure remains a long-term secular growth story.
The clearest example of this rotation has been the explosive rise of the Roundhill Memory ETF (DRAM). Launched earlier this year, the fund attracted $17 billion in second-quarter inflows alone, swelling to nearly $25 billion in assets and becoming the fastest-growing ETF ever. The surge reflects growing investor recognition that high-bandwidth memory has become one of the biggest bottlenecks in AI computing, making memory manufacturers a critical part of the next wave of AI investment.
The AI Supply Chain Is Becoming The New ETF Battleground
The momentum isn't limited to memory chips.
Semiconductor ETFs collectively gathered $23 billion in net inflows during the second quarter, while issuers rushed to capitalize on demand. The first half of the year saw 39 semiconductor ETFs launch, including 36 during the second quarter, highlighting fierce competition to offer increasingly specialized exposure across the AI chip ecosystem.
Rather than broad semiconductor funds alone, investors are increasingly seeking targeted exposure to different parts of the AI hardware stack, from chip manufacturing equipment and advanced packaging to memory and specialized processors.
Beyond semiconductors, ETF investors are also expanding into industries that enable AI's massive computing needs.
Utilities have benefited from expectations of surging electricity demand from hyperscale data centers, while nuclear energy ETFs continue to gain traction as investors look for reliable, carbon-free baseload power capable of supporting AI infrastructure.
Meanwhile, cooling technologies have emerged as another fast-growing investment theme, with products such as the AdvisorShares HVAC and Industrials ETF (HVAC) offering exposure to companies involved in heating, ventilation, air conditioning, refrigeration, and industrial cooling systems that are becoming increasingly essential as AI data centers consume more power and generate more heat.
Real estate has also joined the AI trade, with data center REITs drawing investor interest as cloud providers and hyperscalers race to expand computing capacity. Global X Data Center & Digital Infrastructure ETF (DTCR) and the iShares U.S. Digital Infrastructure and Real Estate ETF (IDGT) are examples of funds with exposure to data center REITs.
Infrastructure May Be The Next Phase Of The AI Investment Cycle
The latest ETF flow trends suggest the AI investment narrative is maturing from a software-led rally into an infrastructure buildout.
As hyperscalers continue investing billions in new data centers, the beneficiaries are expanding beyond AI application developers to companies supplying memory chips, power equipment, cooling technologies, industrial components, and energy infrastructure.
