The AI Power Play: Why Hedge Funds Are Loading Up on Energy Stocks

The Trade Everyone Is Talking About

When you scan Q1 2026 13F filings, one theme cuts across fund after fund: energy. Not oil and gas in the traditional sense, but power generation — nuclear plants, natural gas turbines, and grid infrastructure that feeds the insatiable appetite of AI data centers.

David Tepper doubled his Vistra Energy ($VST) position at Appaloosa Management. Leopold Aschenbrenner's Situational Awareness LP — now a staggering $13.68 billion in U.S. equity exposure, more than double its year-end 2025 size — is built on a thesis that the most valuable assets in the AI era are kilowatt-hours, not model weights. T. Rowe Price added nearly 2.5 million shares of Constellation Energy ($CEG) in Q1 alone, an 85% increase.

This isn't a coincidence. It's a convergence.

Why Energy? Why Now?

The logic is straightforward once you see it. Training a frontier AI model consumes roughly the same electricity as tens of thousands of homes. Inference at scale — serving billions of queries daily — is even more demanding. Microsoft, Google, Amazon, and Meta are all in a race to build data center capacity, and every new megawatt of compute needs a megawatt of reliable, around-the-clock power.

The bottleneck isn't GPUs anymore. Nvidia can manufacture chips; TSMC can fab them. But you can't spin up a nuclear reactor in 18 months. Power generation at the scale hyperscalers need takes years to permit, finance, and build. That constraint is exactly what hedge funds are pricing in.

Vistra Energy, which operates a large fleet of natural gas and nuclear plants, sits squarely in this thesis. Constellation Energy is the largest nuclear operator in the United States and has already signed direct power purchase agreements with Microsoft for dedicated clean energy supply. Both companies were once boring regulated utilities. In 2026, they're being valued like growth stocks.

Aschenbrenner's Contrarian Angle

Situational Awareness LP's positioning is particularly striking. Former OpenAI researcher Leopold Aschenbrenner built a fund around the premise that superintelligence is coming — and that the physical infrastructure required to get there is dramatically undervalued relative to AI software companies.

His fund is reportedly long on power generation and data center operators while maintaining short positions on chipmakers. The bet: semiconductor companies will face margin pressure from commoditization and geopolitical risk, while the entities that own and operate the actual physical capacity to run AI will capture durable, long-term economic value.

It's a deeply contrarian view in a market that has spent three years rewarding Nvidia above almost everything else. But when Situational Awareness LP doubled in size in a single quarter, it's clear institutional capital is listening.

You can track how Aschenbrenner's positions compare with other top funds on the InvestorLens overlap tool.

Tepper's Conviction Play

David Tepper is no stranger to conviction bets. His Appaloosa Management portfolio runs concentrated — 31 holdings across $5.93 billion — and doubling a position is a meaningful signal. Vistra was already a top-ten holding; after Q1, the position is even more dominant.

Tepper's move fits his broader playbook: identify structural dislocations early, size up before consensus forms, and hold through volatility. He exited airline stocks entirely in Q1 (a sector he'd held through the post-COVID recovery) and rotated further into AI-adjacent infrastructure. That's a clear statement about where he sees the next decade of returns.

Follow the full picture of Tepper's portfolio and how it compares to peers at investorlens.capital/investors.

What 13F Flow Data Reveals

Aggregating across Q1 13F filings, the sector flow tells a clear story: energy utilities tied to nuclear and gas-fired baseload power are absorbing institutional capital at a rate not seen in years. Meanwhile, traditional software SaaS names — once the darlings of growth investors — are being trimmed.

The shift reflects a maturing view of the AI trade. Early positioning was about who builds the models (Microsoft, Google, Amazon) and who provides the chips (Nvidia). The 2026 positioning is about who powers the models. See the latest institutional flow data at InvestorLens.

The Risks Worth Watching

The energy thesis isn't without friction. Utility stocks trade at compressed multiples for a reason — regulation, capital intensity, and slow growth have historically capped upside. The bull case assumes hyperscaler demand growth overwhelms those constraints, which is plausible but not guaranteed.

Permitting risk is real. Several proposed nuclear restarts and new gas facilities face regulatory and community opposition. If AI capex growth slows — due to a model capability plateau or economic downturn — power demand projections could overshoot reality, leaving utilities overvalued relative to actual offtake agreements.

And there's the classic hedge fund timing risk: being right about the thesis but early on the trade. Vistra is already up significantly from its 2024 lows. Much of the easy money may be made.

Reading the Signal

What makes the energy trade compelling as a market signal isn't any single fund's positioning. It's the convergence. When Tepper, Aschenbrenner, and institutional asset managers like T. Rowe Price all arrive at the same sector through different analytical frameworks, it suggests the thesis has real structural underpinning — not just momentum chasing.

For retail investors, the 13F data offers a lagging but powerful window into where sophisticated capital is flowing. The InvestorLens macro consensus page aggregates these signals across dozens of top funds to surface where the weight of institutional opinion is landing.

Right now, it's landing on electricity.

---

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