The Next Big Thing After the AI Boom: "Power Infrastructure Investment" — How Data Center Demand is Shaping the ETF Market

When discussing the growth of artificial intelligence, particularly generative AI, the focus has traditionally been on semiconductors, GPUs, cloud services, model performance, and training data. However, the focus of AI investment is gradually shifting. Market participants are now looking at the more fundamental conditions necessary to power AI, namely "electricity."

Training large language models, running image-generating AI, and keeping enterprise AI agents operational all require massive computational processing. This processing occurs in data centers, which consume electricity. AI is not only a story of software but also a story of massive industrial infrastructure.

An article from ETF Trends, "AI’s Exponential Power Demands Could Make This ETF a Winner," highlights the ALPS Electrification Infrastructure ETF, ticker ELFY, in this context. The article's argument is straightforward: if AI growth leads to a surge in power demand, ETFs investing in companies that support this demand—such as those in electrification infrastructure, power transmission and distribution, cooling, industrial equipment, and utilities—could benefit.

The Focus of AI Investment Shifts from "Processing Power" to "Power Supply"

In the early days of the AI boom, the clear winners were semiconductors. Companies that could supply high-performance GPUs, cloud companies that could procure large quantities of GPUs, and software companies that could build AI models attracted significant investment. However, as AI usage transitions from experimental to operational and becomes integrated into corporate activities and personal services, the bottleneck has shifted.

Running AI requires not just GPUs but also server racks, cooling equipment, transformer facilities, power grids, backup power, land, water, and regional power supply capacity. Thus, the expansion of AI does not conclude with the digital industry. Rather, the speed at which physical infrastructure can be developed is becoming the next competitive condition.

According to Gartner's forecast, global data center power consumption is expected to increase from 447 TWh in 2025 to 565 TWh in 2026, representing an approximately 26% year-over-year increase. Additionally, global data center power demand is projected to reach 132 GW in 2026 and could expand to 290 GW by 2030. The power consumption of AI-optimized servers is also becoming more significant, expected to account for 31% of data center power consumption in 2026 and surpass traditional servers by 2027.

These figures indicate that AI growth is not merely about "increased cloud usage." A wide range of industries, including power companies, power grids, transformer facilities, cooling systems, industrial electrical equipment manufacturers, construction companies, and energy management firms, play crucial roles behind the scenes of AI infrastructure.

What is ELFY?

ELFY is an ETF targeting large and mid-cap U.S.-listed stocks that are expected to benefit from the advancement of electrification. According to official information, the ETF aims to invest in companies related to the flow of "electrification," which involves powering machines and systems with electricity or the process of supplying, equipping, and operating with electricity.

The ETF Trends article notes that ELFY, with an asset size of approximately $200 million, is a relatively new ETF established in April 2025, yet it is introduced as one of the pioneers in the theme of electrification and power infrastructure investment in the AI era.

The unique aspect of ELFY is that it invests not directly in AI itself but in the foundation necessary to operate AI. Even after the stock prices of generative AI companies and semiconductor manufacturers have fluctuated significantly, the development of power infrastructure is likely to continue over the long term. As AI usage increases, more data centers will be built. As data centers increase, so will the demand for power supply, transmission and distribution, transformation, cooling, energy storage, and facility management. This investment chain is the fundamental idea behind ELFY.

Why Utility and Industrial Stocks Become Important

As the power demand of AI data centers expands, the most directly related sectors are utilities and industrials. Utilities supply electricity, while industrial companies provide transmission equipment, transformers, cooling equipment, construction services, and power control systems.

In traditional AI investments, utility and industrial stocks were not as flashy as semiconductor stocks. However, as data centers increasingly burden regional power grids, the importance of companies that can update power equipment, expand power grids, provide high-efficiency cooling, and build microgrids and distributed power sources is growing.

This scenario can be described as the "second act" of AI investment. In the first act, companies creating AI models and supplying AI chips were in the spotlight. In the second act, the infrastructure necessary to operate AI across society is questioned. Even if GPUs can be purchased, AI services cannot expand without sufficient power being delivered to data centers. Securing power is becoming a management challenge that affects the growth rate and profit margins of AI companies.

On Social Media: "The Constraint for AI is Not GPUs but Megawatts"

 

Interest in this theme is also rising on social media and investment communities. On Reddit's investment community, posts with the theme "The next constraint for AI is not GPUs but megawatts" can be found. This reflects the awareness that AI-related investments should not be considered solely in terms of semiconductors but should also include increased power demand, power grid constraints, and the physical limits of data center construction.

Another post discusses how to find investment opportunities in power supply companies, power generation, transmission, microgrids, and energy storage systems in response to the expansion of AI data centers. There are voices exploring ways to indirectly invest in power demand for AI data centers through individual stocks, ETFs, and indexes.

Additionally, in the data center operators' community, practical discussions are emerging on how to compensate for power shortages after 2026. Topics include gas turbines, fuel cells, solar power, energy storage, SMRs, small modular reactors, and the efficiency of AC/DC architecture, highlighting that power supply is being discussed not just as an investment theme but as a practical challenge at the field level.

On LinkedIn, posts discuss the impact of AI data center expansion on power grid management, energy transition, and regional infrastructure. It is emphasized that AI is not just a story of the technology industry but is linked to national energy policy, regional development, power grid investment, and even issues of water resources and environmental impact.

Expectations and Concerns are Spreading Simultaneously

Reactions on social media are not entirely optimistic. Some investors view AI's power demand as a new growth theme, seeing potential for funds to flow into utility stocks, industrial stocks, power generation, transmission equipment, and microgrid-related companies.

On the other hand, there are strong voices of concern about environmental impact and effects on local communities. AI data centers not only consume large amounts of electricity but may also use water for cooling. In some regions, issues such as residents' electricity rates, power supply stability, land use, noise, and lack of job creation can become problematic. In fact, there are increasing reports of local opposition and regulatory tightening regarding data center construction.

Thus, AI power demand is both an investment theme and a theme requiring social adjustment. While it may be a tailwind for power infrastructure companies due to demand expansion, if power grid development cannot keep up, data center construction itself may be delayed. Power shortages can be both an investment opportunity and a growth constraint for the entire AI industry.

IEA's Outlook Also Indicates "Structural Change in Power Demand"

The International Energy Agency (IEA) also considers the expansion of power demand due to data centers and AI as an important theme. According to IEA's outlook, global data center power consumption is expected to be about 415 TWh in 2024, accounting for approximately 1.5% of global power consumption. Furthermore, by 2030, it is expected to roughly double to about 950 TWh, accounting for about 3% of global power demand.

While 3% of the global total may seem small, the issue is not the average but the concentration in specific regions. Data centers tend to concentrate in areas where electricity is cheap, communication networks are well-developed, land is available, and cloud demand is nearby. If massive power demand arises in a short period in specific regions, the burden on power grids and power generation facilities will increase sharply.

Therefore, the power issue of AI data centers is not just about "whether there is enough power globally," but "whether power of the necessary quality can be delivered to the necessary places at the necessary times." This creates room for infrastructure investment in power grids, transformer facilities, energy storage, cooling, demand-supply management, and distributed power sources.

The Appeal and Risks of ELFY

The appeal of thematic ETFs like ELFY lies in the ability to broadly invest in the significant structural change of AI power demand without selecting individual companies. AI-related stocks tend to concentrate on semiconductors and major cloud providers, but electrification infrastructure ETFs allow for diversification into more foundational areas such as power transmission and distribution, construction, electrical equipment, and utilities.

Moreover, it overlaps with multiple long-term themes, including not only AI but also electric vehicles, electrification of manufacturing, renewable energy adoption, energy efficiency in industrial facilities, and the spread of energy storage batteries. Even if the AI boom fluctuates, electrification itself may continue as a structural change in the overall economy.

However, there are risks. First, thematic ETFs can easily lead to market expectations getting ahead. As AI power demand gains attention, related stocks' valuations may rise, potentially overpricing future growth. Second, utility and infrastructure-related companies are sensitive to interest rates. If interest rates remain high, the cost of capital investment and financing burdens will increase. Third, data center construction can be delayed due to regulations, local opposition, environmental standards, and power contract constraints.

Furthermore, there is uncertainty in AI power demand forecasts themselves. If model efficiency, chip power savings, and data center operation improvements progress, power consumption per unit calculation may decrease. On the other hand, if AI usage explodes, demand growth may exceed savings from efficiency improvements. Investors need to consider not only demand expansion scenarios but also technological advancements and regulatory impacts.

The Winners of the AI Era Might Be "Unassuming Companies"

In the AI boom, attention tends to focus on glamorous companies. Companies developing models, making AI chips, and providing cloud services easily make headlines. However, those actually supporting AI society include companies that lay power lines, install transformers, supply cooling equipment, and stabilize power grids.

Data centers are symbols of futuristic software and are also very realistic power-consuming facilities. Servers generate heat, require cooling, cannot afford power outages, and have massive power contracts. The more AI demand increases, the more the value of the power infrastructure behind it rises.

The significance of the ETF Trends article highlighting ELFY goes beyond simply introducing one ETF. It indicates that the perspective of AI investment is expanding from semiconductors to infrastructure, from software to power grids. If you believe in AI growth, you also need to focus on the foundation that enables that growth.

Conclusion: The Future of AI Lies on the Power Grid

AI does not grow solely in virtual space. It is built on massive data centers, power plants, power grids, transformer facilities, cooling systems, construction sites, and the consensus of local communities. Therefore, the exponential growth of AI also exerts exponential pressure on power infrastructure.

Electrification infrastructure ETFs like ELFY are one way to capture this change as an investment theme. The idea of investing not in AI itself but in the power and equipment necessary to operate AI may gain further attention in the future.

However, this theme is not a simple "buy because it's AI-related" story. Many factors, including increased power demand, power grid constraints, interest rates, regulations, local opposition, environmental impact, and technological efficiency improvements, are intertwined. What is required of investors is not to ride the wave of AI boom enthusiasm but to calmly observe the real infrastructure supporting that enthusiasm.

The next winners in AI may not only be companies creating cutting-edge models. Companies delivering electricity, making equipment, providing cooling, and strengthening power grids—these seemingly modest entities might become the real foundation of the AI era.

※This article is intended for informational purposes only and does not recommend buying or selling specific ETFs or individual stocks.

---

Source URL

ETF Trends. Main reference source on the investment theme of AI power demand expansion and ALPS Electrification Infrastructure ETF, ELFY.
https://www.etftrends.com/etf-building-blocks-content-hub/ais-exponential-power-demands-make-etf-winner/

Gartner's June 2026 release. Reference source for data center power consumption expected to increase to 565 TWh in 2026 and the power consumption ratio of AI-optimized servers.
https://www.gartner.com/en/newsroom/press-releases/2026-06-10-gartner-says-data-center-electricity-demand-to-grow-26-percent-in-2026

Official ALPS Funds page introducing ELFY. Reference source for ELFY's investment objectives, exposure to electrification infrastructure, and fund overview.
https://www.alpsfunds.com/exchange-traded-funds/elfy

IEA's "Energy demand from AI." Reference source for global data center power consumption and background data on AI and energy demand.
https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai

IEA's "Key Questions on Energy and AI" executive summary. Reference source for the outlook that data center power consumption will expand to about 950 TWh by 2030.
https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary

U.S. Department of Energy announcement on data center power demand. Reference source for supplementary information on the expansion of power load in U.S. data centers.
https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers

Reddit investment community post. Reference source for SNS reactions indicating interest in stocks, ETFs, and indexes related to power supply for AI data centers.
https://www.reddit.com/r/investing/comments/1mb4k2u/what_are_some_stocks_funds_or_indexes_for/

Reddit investment community post. Reference source for SNS reactions indicating the argument that "AI's next constraint is megawatts, not GPUs."
https://www.reddit.com/r/investing/comments/1op3l15/ais_next_constraint_is_megawatts_not_gpus_iea/

Reddit data center-related community post. Reference source for practical discussions on power shortages for AI data centers after 2026, alternative power sources, fuel cells, solar power, SMRs, etc.
https