Beyond Chips: Spotting the AI-Power Trade in 13Fs

For two years, the easiest way to play artificial intelligence was to own the chipmakers. But institutional investors have been quietly broadening the trade — and you can see it in 13F filings. Increasingly, funds are buying not just the semiconductors that train AI models, but the power generation, grid equipment, and physical infrastructure that the data centers behind AI consume. This guide explains the AI-and-power trade and how to spot it in institutional holdings.

The thesis in plain terms

AI data centers are enormous consumers of electricity. Training and running large models requires racks of power-hungry chips, and those racks need reliable, large-scale power plus the land, cooling, and connectivity to support them. That demand flows downstream to the companies that generate electricity, build grid and turbine equipment, supply natural gas, and own the physical infrastructure — including the towers and real estate that carry digital traffic.

So the "AI trade" is no longer just chips. It is a chain: semiconductors at one end, and the power and infrastructure that feed them at the other.

What the chain looks like in a 13F

You can recognize a fund leaning into this theme when its buying spans both ends of the chain in the same quarter:

• Compute — semiconductor leaders like Nvidia (NVDA) and Broadcom (AVGO).
• Power generation and equipment — names like GE Vernova (GEV), which makes gas turbines and grid hardware, and energy majors such as Chevron (CVX).
• Physical infrastructure — communications real estate like cell-tower REITs American Tower (AMT) and Crown Castle (CCI).

When several of these move up together in one fund's filing, you are likely looking at a deliberate AI-infrastructure thesis rather than a coincidence.

Two real examples

Federated Hermes showed the pattern clearly in a recent quarter, adding to Nvidia and Broadcom on the compute side while raising GE Vernova by 22% and Chevron by 14% on the power side — buying the full chain at once, as we detailed in Federated's AI-and-power buying.

Separately, Lazard Asset Management raised both Nvidia and the cell-tower REITs American Tower and Crown Castle in the same quarter — barbelling AI growth with rate-sensitive infrastructure, a move we covered in Lazard's broad buying. Different managers, same underlying logic.

Why investors read the trade this way

Spreading exposure across the chain has a clear rationale. The chipmakers capture the most visible growth but trade at high valuations and concentrate risk in one industry. The power and infrastructure names offer a more diversified way to participate: they benefit from the same demand wave while often carrying steadier, more rate-sensitive cash flows. Buying both is a way to own the theme without betting everything on a single link.

How to use this when reading 13Fs

When you scan a fund's quarterly changes, do not stop at the chip names. Look for adds in power generation, grid equipment, energy, and communications infrastructure in the same filing. A fund buying only Nvidia is making a semiconductor bet; a fund buying Nvidia and turbines and towers is expressing a view on the entire AI build-out. Recognizing the difference tells you how broadly — and how confidently — a manager is positioned for the theme. Track these names and their institutional holders on their stock pages to see who else is building the same chain.

FAQ

What is the AI-and-power trade?

It is the strategy of investing not only in AI chipmakers but also in the electricity generation, grid equipment, and physical infrastructure that AI data centers require. The thesis treats AI as a chain from semiconductors to power.

Why does AI drive demand for power and infrastructure?

AI data centers consume large amounts of electricity and require turbines, grid hardware, cooling, and communications infrastructure. That demand flows to power generators, equipment makers, energy suppliers, and infrastructure REITs.

How can I spot the AI-power trade in a 13F?

Look for a fund adding across the chain in one quarter — semiconductors like Nvidia and Broadcom alongside power names like GE Vernova or Chevron and infrastructure names like American Tower or Crown Castle.

Which stocks represent the power side of the AI trade?

Power-generation and equipment names such as GE Vernova, energy majors like Chevron, and communications-infrastructure REITs like American Tower and Crown Castle are common ways funds play the power side.

Why do funds buy power and infrastructure instead of just chips?

Chipmakers offer the most visible growth but high valuations and concentrated risk. Power and infrastructure names participate in the same demand wave with steadier, more diversified cash flows, so owning both spreads the bet.

Does buying Nvidia alone count as the AI-power trade?

No. Buying only chip names is a semiconductor bet. The AI-power trade specifically spreads across the chain — compute plus the power and physical infrastructure that support it.