The artificial-intelligence boom is increasingly becoming an electricity story. AI models need data centers, data centers need enormous amounts of electricity, and that electricity requires generation, transmission, cooling, power-management equipment, and digital infrastructure.

The International Energy Agency estimates that global data-center electricity consumption could more than double to roughly 945 TWh by 2030, while the United States could account for almost half of the growth in electricity demand over that period.

That creates an interesting investment theme: instead of buying only the companies building AI chips or software, investors can look at the infrastructure needed to power the AI economy.

Three ETFs stand out for different reasons:

• AIPO — Defiance AI & Power Infrastructure ETF: the most direct combination of AI, power, data centers, and electrical infrastructure.
• DTCR — Global X Data Center & Digital Infrastructure ETF: the strongest focus on data centers and digital infrastructure.
• UTES — Virtus Reaves Utilities ETF: a more traditional utility portfolio that has increasingly become an indirect way to participate in the AI-power buildout.

The important point is that these ETFs are not interchangeable. They give investors three different ways to approach the same long-term theme.

⚡ Why AI Infrastructure Is Becoming an Investment Theme

AI sounds like a software revolution, but behind every AI model is physical infrastructure.

👇A modern AI data center needs:

1. High-performance chips
2. Servers and networking equipment
3. Cooling systems
4. Backup power
5. Transformers and electrical equipment
6. Transmission and grid connections
7. Reliable electricity generation
8. Large physical facilities

The IEA estimates that servers account for roughly 60% of electricity demand in modern data centers, while cooling and other infrastructure also consume substantial amounts of power.

The scale is becoming especially important. The IEA notes that a traditional data center might require 10–25 MW, while a hyperscale AI facility can exceed 100 MW.

In simple terms:

More AI → more data centers → more electricity → more infrastructure investment.

That chain is what makes AIPO, DTCR, and UTES interesting.

🏆 The Three ETFs at a Glance

The fund data show an important difference in construction. AIPO has the broadest portfolio, while DTCR and UTES are considerably more concentrated. AIPO's expense ratio is also the highest of the three, while UTES has the lowest.

👇Expense ratios: AIPO vs. DTCR vs. UTES

🤖 1. AIPO: The Most Direct AI + Power Combination

Defiance AI & Power Infrastructure ETF (AIPO) is arguably the most straightforward choice for investors who want to combine the AI theme with the infrastructure required to support it.

The fund tracks the MarketVector U.S. Listed AI & Power Infrastructure Index. The index targets U.S.-listed companies deriving at least 50% of revenue from areas such as AI hardware, data centers, power infrastructure, or related industries.

As of October 1, 2026, AIPO had approximately $925.8 million in assets, 89 holdings, and a 0.69% gross expense ratio.

🔌 What Does AIPO Own?

Its portfolio demonstrates why the fund is different from a conventional technology ETF.

👇Recent major holdings included:

• Quanta Services — 8.64%
• Eaton — 8.61%
• GE Vernova — 8.49%
• Vertiv — 6.52%
• Bloom Energy — 5.36%
• Nvidia — 4.22%
• Alphabet — 4.18%
• Broadcom — 3.86%
• Cameco — 3.33%
• Constellation Energy — 3.09%

Notice something interesting: Nvidia is important, but it isn't the dominant holding.

Instead, AIPO combines electricity generation, grid equipment, data-center power systems, AI semiconductors, and technology companies.

That makes it a particularly interesting option for someone who believes the next stage of AI investment will expand beyond chips.

📊 AIPO Performance

Through September 30, 2026, AIPO reported a 27.59% YTD total return at NAV. However, investors should remember that AIPO only launched in July 2025, so its historical record is relatively short.

This is an important consideration.

A strong recent return can demonstrate momentum, but it cannot tell us how the ETF will behave through a complete economic cycle.

🏢 2. DTCR: A Direct Bet on Data Centers

If AIPO is about the whole AI-power ecosystem, DTCR is more specifically about the physical and digital infrastructure supporting data centers.

The Global X Data Center & Digital Infrastructure ETF (DTCR) seeks to track the Solactive Data Center REITs & Digital Infrastructure Index.

As of September 2026, DTCR had roughly $2.15 billion in net assets, 25 holdings, and a 0.50% expense ratio.

🏗️ DTCR's Portfolio Looks Different

👇Its largest positions include:

The portfolio is therefore heavily exposed to real estate, data centers, communications infrastructure, and digital networks.

As of August 31, 2026, Global X reported approximately 59.8% real-estate exposure and 37.9% information-technology exposure.

📈 DTCR's Recent Results

Global X reported a 29.10% YTD return for DTCR through October 2, 2026. Its one-year return was 33.86% in the same data set.

However, investors should not interpret that as proof that DTCR will outperform AIPO or UTES in the future.

The recent numbers are simply a snapshot of how the market has rewarded the theme.

⚡ 3. UTES: The Utility Route Into the AI Power Boom

At first glance, UTES may seem like the odd one out.

It isn't an AI ETF.

It is the Virtus Reaves Utilities ETF, an actively managed fund focused on utilities.

But this is exactly what makes it interesting.

As AI data centers consume more electricity, companies capable of producing and delivering reliable power may become increasingly important.

UTES held just 19 securities as of October 2, 2026, with an expense ratio of 0.49%. Its largest positions included Constellation Energy, Talen Energy, Vistra, Xcel Energy, and CenterPoint Energy.

The top three alone — Constellation, Talen, and Vistra — represented more than 30% of the portfolio.

☢️ Why Nuclear Power Matters

This is where UTES becomes particularly interesting for AI investors.

AI companies need reliable electricity around the clock. Solar and wind are important, but large data centers also require dependable power when the sun isn't shining or the wind isn't blowing.

Nuclear power can provide that steady generation.

And several major technology companies are already signing long-term agreements with utilities and power producers.

🏭 Real-World Example: Microsoft and Constellation

Consider the agreement between Microsoft and Constellation Energy.

In September 2024, Microsoft and Constellation announced a 20-year power purchase agreement designed to support the restart of Three Mile Island Unit 1 in Pennsylvania, which was renamed the Crane Clean Energy Center. The facility is expected to provide approximately 835 MW of carbon-free electricity to the grid.

This is more than an energy story.

It demonstrates the physical connection between:

AI/data centers → electricity demand → nuclear generation → utility investment.

For an investor holding UTES, Constellation is therefore not simply "another utility."

It can represent part of the infrastructure supporting the AI economy.

☢️ Another Example: Meta, Constellation and Nuclear Power

In 2025, Meta signed a 20-year agreement with Constellation for electricity from the Clinton Clean Energy Center in Illinois.

The agreement covers 1,121 MW of emissions-free nuclear energy and supports the continued operation of the facility.

Meta subsequently announced additional nuclear-energy initiatives in 2026 involving companies including Vistra, TerraPower and Oklo, citing the need for reliable electricity to support AI infrastructure.

This illustrates an important investment concept:

The AI infrastructure opportunity isn't limited to Nvidia or data-center operators. It can extend all the way upstream to electricity producers.

🔋 Another Real Example: Talen Energy and Amazon

Talen Energy provides another useful case.

Talen operates the Susquehanna nuclear facility in Pennsylvania, which has approximately 2.5 GW of gross capacity.

The company has developed a model in which nuclear power can directly support data-center operations. Talen says it entered an agreement with Amazon Web Services to provide power for an adjacent AWS data-center campus, and in 2025 expanded its relationship with Amazon to support AI and cloud operations.

This is precisely the kind of development that helps explain why utility and power-generation stocks can become part of the AI investment story.

📊 How Concentrated Are These ETFs?

Another major difference is concentration.

Using the latest published holdings, the largest five positions represented approximately:

These percentages are calculated from the published weights and can change as holdings move.

👇Approximate concentration of the top five holdings

This matters because concentration can amplify both gains and losses.

For example, if someone named John Miller bought DTCR because he expected data-center demand to accelerate, he would benefit if companies such as Digital Realty and Equinix performed strongly.

But if the data-center cycle weakened, interest rates rose sharply, or real-estate valuations compressed, the same concentration could work against him.

👨‍💼 A Simple Investor Example

Imagine three fictional American investors.

John Miller, age 45, believes AI spending will continue expanding across the technology and energy industries. He might find AIPO attractive because it covers multiple parts of the ecosystem.

Sarah Johnson, age 38, is specifically interested in the growth of data centers, cloud computing, and digital infrastructure. DTCR may fit her thesis more directly.

David Brown, age 55, prefers established power producers and utilities but still wants some exposure to rising electricity demand from AI. UTES may provide a more utility-focused approach.

None of these choices is automatically "best."

The right choice depends on the investor's risk tolerance, time horizon, existing portfolio, and specific view of the AI infrastructure cycle.

⚖️ AIPO vs. DTCR vs. UTES: Which One Stands Out?

These ratings are qualitative comparisons, not investment recommendations.

⚠️ The Risks Investors Should Not Ignore

The AI-power theme is exciting, but excitement can create expensive valuations.

1. 📉 Valuation Risk

If investors price in years of AI growth today, disappointing growth could lead to sharp corrections.

2. 💰 Interest-Rate Risk

DTCR in particular has meaningful real-estate exposure. Higher interest rates can increase financing costs and pressure valuations for infrastructure and REIT-related companies.

3. ⚡ Regulatory Risk

Utilities operate in heavily regulated markets. Changes in electricity pricing, environmental rules, permitting, or nuclear regulation can affect returns.

4. 🏗️ Construction Risk

Power plants, transmission lines, and data centers take years to build. Delays can significantly change expected economics.

5. 🤖 AI Efficiency Risk

AI hardware and software may become dramatically more energy-efficient.

That would be good for society, but it could reduce the amount of electricity required for each AI workload.

6. 🎯 Thematic Concentration

None of these ETFs should automatically be considered a complete portfolio.

A thematic ETF can complement a diversified portfolio, but relying entirely on one investment theme can expose investors to unnecessary risk.

🧭 So, Which ETF Is the Best?

There is no universal winner.

🥇 AIPO — Best for a broad AI-power thesis

AIPO is probably the most interesting choice for an investor who believes AI, electricity, grid modernization, data centers, and semiconductors will all grow together.

Its broader portfolio is its biggest advantage.

The trade-off is a higher 0.69% expense ratio and a relatively short operating history.

🥈 DTCR — Best for a data-center infrastructure thesis

DTCR is more focused.

If your central belief is that data centers, cloud computing, towers, and digital infrastructure will expand dramatically, DTCR offers a direct way to express that view.

Its 0.50% expense ratio is also lower than AIPO's.

🥉 UTES — Best for the electricity-generation angle

UTES offers a different path.

Rather than betting directly on data centers, investors are buying companies that produce and distribute electricity — including several companies increasingly connected to AI-related power demand.

Its 0.49% expense ratio is the lowest among the three, although its portfolio is more concentrated.

📝 Final Summary

The AI revolution is gradually becoming an infrastructure revolution.

AI models require enormous computing power, and computing power requires data centers. Data centers require electricity, cooling, networking, transformers, transmission infrastructure, and reliable power generation.

That creates three different investment approaches:

AIPO offers the broadest combination of AI technology and power infrastructure. It is the most "all-in-one" expression of the AI-power theme.

DTCR provides a more concentrated bet on data centers and digital infrastructure. It may appeal most to investors who believe the physical footprint of cloud and AI computing will expand rapidly.

UTES takes the electricity route. It is not an AI ETF, but its holdings include major power producers that can benefit if AI-driven electricity demand continues to rise.

The most important lesson is that AI infrastructure is much larger than semiconductors.

The companies generating electricity, building grids, operating data centers, supplying cooling and electrical equipment, and maintaining digital networks may all participate in the next phase of AI investment.

At the same time, investors should remain disciplined. High growth expectations can already be reflected in stock prices, interest rates can affect infrastructure valuations, and technological improvements can change how much electricity AI ultimately requires.

For a diversified investor, the question may therefore be less about finding one "winning" ETF and more about deciding which part of the AI infrastructure chain makes the most sense for the portfolio.

📚 Sources & References

• Defiance ETFs — AIPO official fund page — fund objective, holdings, expense ratio and performance.
• Global X — DTCR official fund page — holdings, portfolio exposure, expense ratio and performance.
• Virtus — Reaves Utilities ETF information — UTES strategy and utility-sector focus.
• International Energy Agency — Energy and AI — data-center electricity demand and AI energy outlook.
• Talen Energy — Powering Data — Susquehanna/AWS data-center power example.
• Constellation Energy — Microsoft/Crane Clean Energy Center announcement — Microsoft nuclear PPA example.
• Meta — Meta and Constellation clean-energy agreement — AI-related nuclear power example.