AI Needs Power. This Soaring ETF Has the Right Stuff.

The Defiance AI & Power Infrastructure ETF (NASDAQ: AIPO) has grown to just over $665 million in assets under management only 10 months after launch, with nearly two-thirds of that AUM arriving this year. The fund is up nearly 46% year to date and holds 77 stocks, with about 20% in AI-linked semis/data infrastructure and roughly 80% in power infrastructure, including Broadcom, Nvidia, GE Vernova, and Eaton among top holdings. The article argues the ETF is well positioned to benefit from AI-driven data center and grid spending, but the piece is largely promotional rather than news-driven.

Analysis

The key takeaway is that this is less a pure AI trade than a levered call on the industrial capex supercycle behind AI. The market is increasingly paying up for the companies that convert GPU demand into grid spend, which means the second-order winners are the electrical equipment, power management, and EPC names with backlog duration rather than the chipmakers themselves. That helps explain why the strongest exposure is in the power sleeve: the bottleneck is no longer model quality, but interconnect, switchgear, transformers, and utility approvals.

GEV looks like the cleanest expression of that bottleneck because it sits at the intersection of generation, gas turbines, and grid modernization, where pricing power can persist for multiple budget cycles. AVGO and NVDA remain important sentiment anchors, but the marginal upside may now be constrained by elevated expectations and valuation; they are still beneficiaries, just not the highest-beta way to play incremental AI infrastructure spend. INTC is the odd one out and acts more as a thematic placeholder than a direct beneficiary, so its inclusion may dilute the ETF’s purest exposure.

The main risk is that the AI power narrative remains a multi-year story while the stock move has already front-run a lot of the near-term capex visibility. If hyperscaler spending pauses, utility rate cases slip, or transmission permitting slows, the market could de-rate the whole basket even if long-term demand stays intact. On the other hand, if electricity demand growth forces faster regulatory approvals, the upside becomes more convex for the grid-equipment names than for semis.

Consensus is underestimating how narrow the bottleneck is: the trade is not just ‘more power,’ it is specific scarcity in grid hardware and engineering capacity. That means the right way to express the theme is through the bottleneck owners, not the broad ETF wrapper, because the ETF drags in lower-quality utility exposure and some weaker AI adjacency. In our view the move is directionally right but probably over-bundled, which creates an opportunity to own the best-in-class names outright and fade the weaker constituents through relative-value structures.

Key Decisions for Investors

• Long GEV vs short a basket of lower-beta utilities for 3-6 months: target outperformance if AI-driven grid capex keeps reaccelerating; risk is a bond-yield spike that compresses all utility multiples together.
• Buy AVGO on 5-10% pullbacks over the next 1-2 months rather than chasing breakouts; use it as the higher-quality AI infrastructure proxy with better duration than pure semis, but cap upside expectations given crowded positioning.
• Use NVDA call spreads 6-12 months out instead of outright equity if you want AI exposure with defined premium risk; this captures further upside from infrastructure spend while acknowledging valuation and positioning risk.
• Avoid chasing AIPO after the recent flow surge; if used at all, keep it as a tactical basket substitute only on market weakness, since the 0.69% fee and lower-conviction holdings likely drag relative returns over a full cycle.
• If you want a contrarian pair, long GEV / short INTC for 3-6 months: one is a direct beneficiary of the power bottleneck, the other is mostly incidental to the theme; the pair isolates the real winner from thematic noise.