VSTVistra Corp

Vistra Corp is the purest traded expression of the AI power bottleneck: existing, licensed nuclear megawatts that hyperscalers literally cannot build for themselves on their own timeline. This is not a defensive utility story — it is a scarcity-asset story with named hyperscaler contracts already on the books.

Investment Thesis

Vistra is one of the largest US merchant power generators and retailers, with nuclear, natural gas, solar, and storage assets. The Physical AI thesis is not generic utility exposure: it is scarce, licensed, dispatchable nuclear capacity sold into long-duration hyperscaler contracts. Vistra owns six nuclear units totaling 6,448 MW, including Comanche Peak's two 1,200 MW units licensed to 2050 and 2053. In February 2026 the company announced an 800 MW nuclear PPA with Amazon Web Services at Comanche Peak, subsequently disclosed in the 2025 10-K as a 20-year agreement for up to 1,200 MW with deliveries beginning in Q4 2027 and ramping to full capacity by 2032. It also signed nuclear energy, capacity, and uprates with Meta at its PJM facilities. Combined, the named AWS and Meta tranches total ~3,809 MW — roughly 59% of the nuclear fleet pre-contracted to hyperscalers. The financials confirm scale: Q1 2026 revenue was $5.001B (+17.7% YoY), operating income swung to $1.499B from a $120M loss a year earlier, and net income was $1.029B. FY2025 ongoing-operations adjusted EBITDA was $5.912B.

Physical AI / Value-Chain Relevance

Vistra maps to the Grid, Power & Thermal Infrastructure layer — the foundation of the Physical AI stack. AI data-center load growth is the demand shock: the Q1 2026 10-Q explicitly cites data centers and other large-scale electricity consumers as contributors to fast-paced load growth in the regions Vistra serves. The moat is not brand; it is operating licenses, interconnection, and asset scarcity. New nuclear build takes 15+ years, so hyperscalers cannot self-supply equivalent baseload on their own timeline. The quantity math is stark: 1,200 MW x 8,760 hours equals ~10.5 TWh per year at nameplate, roughly 9.5 TWh at a 90% capacity factor. Every $1/MWh change in realized pricing on that tranche moves annual revenue by ~$10.5M at full output; a $5/MWh swing is a ~$47-53M annual swing. Gas acquisitions — Cogentrix (~5,500 MW, $3.45B, closed mid-2026) and Lotus (~2,600 MW) — broaden the dispatchable stack around the nuclear core.

Catalysts

• Q2 2026 earnings on August 6, 2026 — the first print since the Helix deal, with guidance updates and Cogentrix close commentary
• Helix Digital Infrastructure, the $10B+ KKR/Nvidia vehicle, converting partnership structure into signed PPAs
• AWS delivery start in Q4 2027 and ramp to full capacity by 2032
• Meta nuclear uprates at PJM
• 2027 power-curve strength and any additional hyperscaler contracts

Positioning / What the Market May Be Missing

At a forward P/E near 14.8x versus 15-20x for nuclear peers, the market still appears to be pricing merchant volatility rather than the contracted cash-flow transition. Institutional ownership is 91%. The stock (~$140.58) sits below both its 50-day ($156.15) and 200-day ($163.79) averages and below the tracked entry range ($147.38-$171.94), with a research target of $142.98 — the market has de-rated the name even as the underlying contracts de-risk it. Each signed hyperscaler tranche converts volatile merchant megawatts into long-dated contracted revenue; with 59% of the nuclear fleet already contracted, the remaining uncontracted capacity is scarce optionality, not commodity supply. Honest caveat: the machine-level thesis status is currently NEEDS_MORE_DATA with conviction ~3.4/5 — a research-backlog artifact (last formal review 2026-06-29) rather than an evidence downgrade, and the vault research verdicts are consistently STRENGTHENED. The stock's YELLOW crowding flag reflects that the AI-power trade is consensus; that is a valuation-timing risk, not a thesis risk.

Risks and What Invalidates the Thesis

• Earnings miss: Q1 2026 EPS of $1.35 missed the $1.50 consensus (-10%) on lower ERCOT realized prices and higher nuclear O&M — merchant noise, but real.
• Merchant volatility: wholesale power pricing and hedge mark-to-market swings can distort reported results.
• Leverage: long-term debt of ~$19.2B makes the story sensitive to rates and EBITDA delivery.
• Contract conversion: Helix must convert from aspiration into signed, duration-rich PPAs; AWS must start deliveries in Q4 2027 and ramp by 2032.
• Power-curve softening: a materially weaker 2027 curve would erode the scarcity premium.
• Operations: nuclear outage, refueling, or regulatory events would reduce deliverable MW.

The thesis is invalidated if 2027 power curves soften significantly, if Helix fails to produce contracted revenue, or if the balance sheet becomes untenable under elevated rates.

What to Watch Next

The Q2 2026 print on August 6 — guidance, Cogentrix commentary, and any Helix PPA signings; PPA pricing and escalation disclosures; the Q4 2027 AWS delivery start; Meta uprate approvals; ERCOT/PJM forward curves; and the leverage trajectory. A confirmed Helix PPA would be the single strongest conviction-upgrade signal available.