Utilities Were the #1 Bought Sector in Q2 13Fs. That Should Worry You.

The Q2 2026 13F season is over, and the headline everyone chased was Berkshire's sevenfold Alphabet increase. The more interesting number was quieter: across the roughly 8,800 managers who filed for the June 30 period, utilities was the single most-bought sector of the quarter.

That is not a sentence anyone wrote in 2021. Utilities were the bond proxy — the thing you owned when you wanted a dividend and didn't want to think about it. In Q2 2026 they were bought as a growth trade.

What Actually Changed

The mechanism is simple and by now well documented. U.S. data center electricity demand is on track to roughly triple between 2024 and 2030, eventually consuming close to a tenth of national power output. Utilities are responding with capital spending near a record $240 billion in 2026 alone. American Electric Power has laid out roughly $78 billion of investment through 2030. Meanwhile the hyperscalers — Meta, Microsoft, Amazon, Alphabet — are collectively pointed at something like $700 billion of AI build-out this year.

For a regulated utility, that demand growth converts almost mechanically into rate base, and rate base converts into earnings. For the merchant power names — Constellation at roughly 60 GW, Vistra at 45, NRG at 28, Talen at 13 — it converts into multi-decade power purchase agreements signed directly with tech buyers, plus capacity revenues already locked through 2028.

So institutions did the obvious thing. David Tepper's Appaloosa is the cleanest example: he trimmed Micron, exited SanDisk entirely, and rotated the proceeds into Vistra and NRG, both of which were still trading under 16x forward earnings when he bought. That is not a defensive allocation. That is the same AI thesis, expressed one layer further down the stack, at a third of the multiple.

You can trace this rotation yourself on the InvestorLens flow page, which shows quarter-over-quarter position changes across every tracked manager rather than the single-quarter snapshot most 13F sites default to.

The Part That Should Give You Pause

Here is the problem with a sector being the most bought thing in a quarter: by definition, the buying already happened. Utilities are up roughly 8% on the AI power thesis. The story is no longer a secret, and 13F data is a 45-day-lagged rearview mirror. Every position you're reading about was established between April and June.

We have written before about what happens when a thesis becomes consensus. Our overlap tool exists specifically to measure how many top managers hold the same name, because crowding is a risk factor that never appears in a fund's own filing. A stock that eleven elite managers own has already had eleven elite managers buy it. The marginal buyer has to come from somewhere else.

Utilities are now approaching that condition. Not at semiconductor levels of crowding — the July chip unwind showed what that looks like — but the direction is unambiguous.

The Risk Nobody Is Modeling

The second issue is political, and it is the one most 13F-driven analysis misses entirely.

AI power demand is pushing electricity prices up for ordinary ratepayers. Goldman Sachs estimates higher power prices will add roughly 0.1% to core inflation in both 2026 and 2027. That is a small number in a macro model and a very large number in a state utility commission hearing.

Utilities are regulated businesses whose earnings depend on regulators approving rate increases. The entire bull case assumes commissions keep granting the returns needed to fund data-center-driven capex. That assumption holds until residential bills rise enough that it becomes a local election issue — and in several states, it already has. Municipal pushback against new data center construction is now routine rather than exceptional.

Merchant power names like Vistra and NRG are somewhat insulated, since they sell into wholesale markets rather than negotiating residential rates. That distinction may be why Tepper picked the merchants specifically. It's a distinction worth checking before you assume "utilities exposure" means one thing.

How to Use This

A 13F tells you what a manager owned on June 30. It doesn't tell you why, at what price, or whether they still own it. Treat the utilities rotation as evidence that serious capital found the AI power thesis credible enough to underwrite — not as a signal to buy the sector two months after the fact at a higher price.

The more useful questions: which managers bought regulated versus merchant power, and what does that say about who they think bears the political risk? Are the same funds that bought utilities also still holding the chip names, or was this a genuine rotation? You can compare positioning across managers on the investors page, and see how the sector fits into broader institutional macro positioning at macro consensus.

The AI trade didn't end in Q2. It moved. Knowing where it moved to is only useful if you also know who else already got there.

---

Data sourced from public SEC 13F filings. Educational research only — not investment advice.