Published on: 2026-07-17
Updated on: 2026-07-17
AIPO invests in the power systems, grid equipment, data centres and chips supporting the AI buildout. Its latest holdings reveal a concentrated mix of industrial leaders, semiconductor stocks, nuclear-energy companies and several higher-risk infrastructure names.
Artificial intelligence is driving rapid growth in data-centre electricity demand, although forecasts still vary widely over how large that demand will become. The harder question for an investor is more specific: if you buy a fund built around that idea, what are you actually holding? The Defiance AI & Power Infrastructure ETF (AIPO) is one of the most direct attempts to package the “AI needs power” thesis into a single ticker.
But its portfolio is broader, and more volatile, than the calm-sounding “power infrastructure” label suggests. This article explains what AIPO owns, what exposure you are really buying, and where it fits.
AIPO gives targeted exposure to the physical infrastructure behind AI: grid equipment, power generation, data centres and computing hardware, not just the usual mega-cap tech names.
The four largest holdings (GE Vernova, Eaton, Vertiv and Quanta Services) make up about 34% of the fund, and the top 10 roughly 56%, so it is concentrated despite holding 83 names.
It is not a defensive utility fund. Industrials dominate the portfolio, and chip holdings mean it can fall alongside technology stocks.
Returns have been strong (about 51% year-to-date and 67% since inception on a NAV basis through 30 June 2026), but the fund launched only in July 2025 and has no full-cycle record.
The 0.69% expense ratio is high relative to broad index, utility and infrastructure ETFs.
Best viewed as a higher-risk satellite holding for investors who already own a diversified core, not as a core position.
AIPO is a thematic exchange-traded fund from Defiance ETFs that launched on 24 July 2025 and trades on the Nasdaq. AIPO is passively managed and legally classified as non-diversified, meaning it is permitted to place more of its assets in a smaller number of issuers than a diversified fund. It reports 83 holdings, including cash equivalents, and net assets of roughly $938 million. The expense ratio is 0.69%.
The pitch is straightforward: rather than betting only on the software and chip companies that dominate most AI funds, AIPO targets the physical backbone that makes large-scale AI possible: the electrical grid, power generation, data centres and the hardware inside them.
In practice that means it sits somewhere between a technology fund and an industrials-and-utilities fund, without being a clean example of either.
The index divides eligible US-listed companies into four groups covering power and grid equipment, construction and engineering, utilities and power producers, and data centres and AI hardware.
New candidates generally need at least 50% revenue exposure to a qualifying activity, while existing constituents can remain eligible at a 25% threshold. Lower-exposure holdings are capped collectively at 20% of the index.
Constituents are placed into four thematic tiers and weighted using a modified free-float market-cap system. Half of the index is allocated to power generation and grid equipment, with 15% each assigned to construction and utilities and 20% to data centres and AI hardware.
Position caps apply at each quarterly rebalance, although weights can drift higher between rebalances.
The fund’s largest positions, per the issuer’s portfolio dated 17 July 2026:
| Rank | Holding | Ticker | Weight |
|---|---|---|---|
| 1 | GE Vernova | GEV | 9.65% |
| 2 | Eaton | ETN | 8.52% |
| 3 | Vertiv | VRT | 8.19% |
| 4 | Quanta Services | PWR | 7.82% |
| 5 | Nvidia | NVDA | 4.18% |
| 6 | Broadcom | AVGO | 4.06% |
| 7 | Bloom Energy | BE | 3.97% |
| 8 | Constellation Energy | CEG | 3.76% |
| 9 | Cameco | CCJ | 3.51% |
| 10 | AMD | AMD | 2.46% |
The four largest holdings account for about 34% of the fund and the top 10 for roughly 56%. So while the portfolio contains more than 80 company positions, its results depend heavily on a small group of large positions.
The long tail, made up of smaller utilities, nuclear developers, data-centre operators and former crypto miners, adds breadth on paper but may provide limited ballast when the top holdings move together.
The portfolio is best understood as four overlapping trades bundled into one fund.
GE Vernova, Eaton, Vertiv and Quanta Services dominate, together making up more than a third of assets. These companies supply the power systems, cooling, electrical components and engineering work needed to build and connect data centres. This is the most direct expression of AIPO’s stated infrastructure thesis.
Nvidia, Broadcom and AMD give direct exposure to AI chips. That is a feature, not a bug, but it means AIPO can behave like a technology fund when semiconductors sell off. It is not a defensive alternative to the Nasdaq or a chip ETF; it shares some of the same risk.
Names such as Bloom Energy, Constellation Energy, Cameco, Vistra and a cluster of nuclear and uranium plays express the bet that data-centre demand drives investment in gas, nuclear, fuel cells and uranium. This is the part of the fund most tied to energy and commodity cycles.
The fund also owns data-centre operators like Equinix and Digital Realty alongside a set of smaller, more speculative names, including companies historically associated with crypto mining and high-performance computing. This tail raises potential upside but adds funding, execution and commodity-price risk.
AIPO’s short record has been strong. On a NAV basis it returned 51.23% year-to-date through 30 June 2026 and 67.06% since inception. Its $938 million asset base and 0.06% median spread indicate relatively tight secondary-market trading for a young thematic ETF.
Two caveats matter. First, those returns cover well under a year of live trading, so they say little about how the fund behaves across a full cycle. Second, the 0.69% expense ratio is on the high side: a $10,000 position costs roughly $69 a year in fund fees before trading costs, notably more than broad index, utility or infrastructure ETFs.
AIPO fell 4.1% on 16 July during a technology- and semiconductor-led selloff, illustrating how quickly its chip and growth-infrastructure exposure can overwhelm its utility holdings.
The strongest argument for AIPO is genuine differentiation. Its four biggest holdings are industrial and infrastructure companies rather than the familiar mega-cap tech names, so its exposure differs from broad-market and technology-heavy indexes.
It targets a real and widely discussed constraint, the power and grid capacity needed to run AI, rather than the AI applications themselves. And it has achieved the scale and tradability that many thematic launches never reach.
The weaknesses are the mirror image of the appeal. Despite more than 80 company positions, the fund is concentrated: if GE Vernova, Eaton, Vertiv and Quanta Services fall together, the smaller positions may provide limited protection. It is not a defensive power play.
AIPO should not be confused with a conventional utility ETF. Its index begins with a 50% allocation to power and grid equipment, 15% to construction and engineering, 15% to utilities and power producers, and 20% to data centres and AI hardware. The portfolio therefore carries substantial industrial, capital-spending and technology sensitivity.
Some holdings are outright speculative, including small nuclear developers and former crypto miners prone to sharp swings and funding pressure. And the operating history is short: AIPO has not yet been tested through a full downturn, and its own prospectus flags new-fund, sector-concentration, technology, energy-infrastructure and passive-index risks.
Comparing AIPO with the funds investors often consider alongside it clarifies its role.
A semiconductor fund such as SMH provides more direct exposure to semiconductor producers and equipment companies, without AIPO’s grid, construction and power-generation holdings.
XLU focuses on established S&P 500 utility companies and has less direct exposure to semiconductor valuations, small nuclear developers and data-centre construction.
PAVE offers broader US infrastructure exposure across construction, industrial transportation, materials and equipment, but it does not specifically target the AI-power supply chain.
Broad infrastructure ETFs spread across roads, pipelines and transport but capture little of the specific AI-power theme.
AIPO’s distinction is that it blends chips, grid equipment, power generation and data centres into one vehicle. That blend is its selling point and its risk: the pieces can rise together, and they can fall together.
The honest answer is that it depends on the job you want it to do. AIPO is well constructed for a specific theme. Its holdings provide differentiated exposure to grid equipment, data-centre power and AI infrastructure compared with broad-market and technology-heavy indexes.
But its fee, concentration and volatility make it a poor fit as a broad core holding. It looks most sensible as a satellite position for an investor who already owns a diversified core, wants targeted exposure to the AI buildout’s physical layer, and can tolerate large short-term drawdowns.
The central thing to keep in mind is that AIPO is marketed as an AI-power fund, but its performance is really driven by three separate cycles: AI hardware spending, grid and data-centre construction, and power-generation investment. Those cycles can reinforce one another on the way up and reverse together on the way down.
Neither, exactly. It is dominated by industrials and grid-equipment companies, with meaningful chip and power exposure layered on top.
AIPO can make distributions, but it is not designed as an income ETF. It paid only a token distribution in December 2025, and its 30-day SEC yield was −0.16% as of 30 June 2026.
An investor with a diversified core who wants a targeted, higher-risk allocation to AI infrastructure and is comfortable monitoring holdings and quarterly rebalances.
AIPO delivers on its promise of exposure to the physical infrastructure behind AI, and it does so with more originality than a typical AI fund. But “differentiated” is not the same as “safe.” Its concentration, above-average fee, speculative tail and short track record all argue for treating it as a focused satellite bet rather than a foundation for a portfolio.
