Snapshot

  • Ticker: VST (NYSE)
  • Bucket: Power & Physical Infrastructure
  • Q1 2026 position: 523,788 shares, $78.7M, 1.57% of 13F — borderline top-20 (~#20 by issuer value)
  • HQ: Irving, Texas
  • What it does: Largest competitive (non-regulated) power generator and retailer in the US — a ~41 GW (approx.) nuclear, gas, coal, solar and storage fleet plus a large retail electricity business (TXU Energy).

Business Overview

Vistra is an independent power producer (IPP), meaning it sells generation into wholesale markets and through retail contracts rather than earning a regulated return — so it has direct torque to rising power prices and tightening supply. Its fleet spans the second-largest competitive nuclear portfolio in the US (Comanche Peak, plus the 2024 Energy Harbor acquisition’s Beaver Valley, Davis-Besse and Perry), a large gas fleet (expanded in 2025 by acquiring ~2,600 MW across seven plants from Lotus Infrastructure Partners), legacy coal, and growing solar/storage, integrated with TXU’s retail book in Texas.

The AI-datacenter demand wave is the story. Vistra has signed power purchase agreements with Amazon Web Services covering ~3,800 MW of nuclear capacity, including a 20-year PPA for up to 1,200 MW of carbon-free power at Comanche Peak, and management is targeting long-term contracting of roughly 3.2 GW of remaining nuclear capacity at Beaver Valley and Comanche Peak. It is also building new gas — two units totaling 860 MW announced in West Texas — into a market where new entry is slow.

Financially: 2025 ongoing-operations adjusted EBITDA was guided to 1,494M on strong realized prices and higher PJM capacity revenue, and management reaffirmed 2026 guidance of 3.925–4.725B adjusted free cash flow before growth — implying a double-digit FCF yield at the ~$150/share price where Atreides’ Q1 2026 stake was marked.

Financial Trajectory

PeriodOngoing ops adj. EBITDANotes
FY2024~$5.6BEnergy Harbor nuclear acquisition closed
FY2025$5.7–5.9B (guided, narrowed Nov 2025)Described by company as record
Q1 2026$1,494MRecord Q1; PJM capacity tailwind
FY2026 guide$6.8–7.6BAdj. FCFbG $3.925–4.725B

Why Atreides Owns It

Vistra is the most literal expression of the “watts” half of Baker’s “watts and wafers” thesis (Invest Like the Best EP.473, May 2026): the binding constraints on the AI buildout are electric power and wafer capacity, and scarcity pricing accrues to whoever owns the constrained resource. An unregulated IPP with contracted nuclear, a deep gas fleet in ERCOT/PJM, and signed hyperscaler PPAs is the cleanest large-cap way to own that scarcity — datacenter demand growth meets a grid that takes 5+ years to add firm capacity.

Notably, Atreides expressed this thesis almost everywhere else first — connectivity, optics, memory, even space power — and only bought the direct generation play in Q1 2026, after VST had already re-rated through 2024–2025 and then corrected. The new ~$79M position alongside EchoStar gives the Power & Physical Infrastructure bucket both a terrestrial and an orbital leg. The 20-year AWS Comanche Peak PPA structure also fits Baker’s preference for businesses converting commodity exposure into contracted, utility-like cash flows at premium prices (the same logic as his HBM/memory argument).

Position History

QuarterTypeShares/NotionalValue% of 13F
Q4 2024not held
Q1 2025not held
Q2 2025not held
Q3 2025not held
Q4 2025not held
Q1 2026Common523,788$78,741,0501.57%

A brand-new Q1 2026 position with no prior history — initiated in the same quarter the fund cut its QQQ put by 60% and rotated into power, connectivity re-adds and housing names. At 1.57% it is a starter-to-mid-size holding; whether it gets built like Ciena/Micron or cycled like the software names will be visible in Q2.

Risks

  • Late to the trade: VST was one of the best-performing S&P 500 stocks of 2024–25; much of the datacenter-demand story is already in the multiple, and Atreides bought after the re-rating.
  • Power-price reversion: an AI-capex pause, faster grid/generation buildout, or demand-response efficiency gains would deflate scarcity pricing — the exact scenario the fund’s own QQQ put hedges against, making this position pro-cyclical with the rest of the book.
  • Commodity and weather exposure: unhedged spark spreads, ERCOT volatility, and gas-price swings drive earnings dispersion around guidance.
  • Regulatory/political risk: co-location and behind-the-meter datacenter deals face FERC scrutiny; political pressure on consumer power bills could cap wholesale upside.
  • Nuclear operational risk: the contracted-nuclear thesis concentrates value in a handful of plants where outages are costly.
  • 2024’s Q4 EBITDA miss (followed by the Q1 2026 rebound) shows quarter-to-quarter results are noisy even when the structural story is intact.

Sources