Vistra

Phase A — Understand the business

Lens 1 · Company Overview

Vistra is an integrated competitive retail-electricity and power-generation company operating across 18 states + DC, in every major competitive US wholesale market. Two halves bolted together:

• Generation — ~44,000 MW of capacity across nuclear, natural gas, coal, solar, and battery storage. The fleet is the asset base.
• Retail — ~5 million residential/commercial/industrial customers (≈2.6M in Texas alone) under TXU Energy, Ambit, Dynegy, Homefield, Energy Harbor, Public Power, TriEagle, 4Change, and others.

Why the integration matters (the actual business model): the generation fleet produces power; the retail book consumes it to serve load. When power prices spike, generation wins and retail's procurement cost rises (and vice-versa) — the two natural hedges partly offset, which management argues "mitigates the impact of commodity price fluctuations and enhances the stability and predictability of our cash flows". This is the core competitive claim vs. a stand-alone merchant generator or a stand-alone REP.

Contract structure is shifting from short-cycle merchant + retail churn toward long-dated, take-or-pay-style PPAs. The marquee 2025–26 development: 20-year PPAs to supply carbon-free nuclear power to AWS (1,200 MW from Comanche Peak, Texas) and Meta (2,609 MW from PJM nuclear, incl. 433 MW of uprates). These "underwrite higher base profitability in the future" and convert a slice of the merchant P&L into contracted cash flow.

Five reportable segments: Retail · Texas (ERCOT generation) · East (PJM/ISO-NE/MISO/NYISO generation) · West (CAISO generation) · Asset Closure (decommissioning). Generation capacity splits Texas 19,858 MW (46%) / East 22,254 MW (51%) / West 1,529 MW (3%).

Lens 2 · Supply Chain

Map: fuel & equipment upstream → Vistra generation → ISO/RTO grid → retail book + wholesale buyers → end customers (incl. AI/data-center load). Named stakeholders along the chain:

• Fuel inputs: natural gas (third-party intrastate/interstate pipelines), nuclear fuel (uranium/enrichment — Vistra capitalizes nuclear fuel as capex, $133M Texas + $354M East amortization in 2025), lignite/coal (Texas mines under RCT oversight), solar/battery (LG Energy Solution supplied the Moss Landing batteries — now a co-defendant in the fire litigation ``).
• Markets / grid operators (chokepoints): ERCOT (~90% of Texas load, ~83,707 MW 2025 peak), PJM (~160,709 MW peak), MISO, ISO-NE, NYISO, CAISO. The ISO/RTO dispatch + price-formation rules are the single biggest external chokepoint — Vistra's revenue is set by marginal-unit clearing prices it does not control.
• End customers / off-takers (the new layer): Amazon Web Services (1,200 MW Comanche Peak), Meta Platforms (2,609 MW PJM nuclear). These two hyperscalers are now strategically load-bearing counterparties — the supply chain now runs forward into the AI buildout.
• Acquired-fleet sellers: Energy Harbor (4,048 MW nuclear, 2024), Lotus (2,557 MW gas, 2025), and the pending Cogentrix (5,500 MW gas, 10 plants, $4.7B, H2 2026 close) ``.

Single-source / concentration dependencies: nuclear is the crown jewel and the bottleneck — the AWS/Meta PPAs depend on Comanche Peak + the PJM nuclear plants (Perry, Davis-Besse, Beaver Valley) plus the 433 MW of not-yet-built uprates. Execution risk concentrates there. Fuel-wise, ERCOT's high wind/solar penetration makes Vistra's gas/nuclear baseload more valuable in scarcity but exposes the whole market to weather-driven supply shocks (Winter Storm Uri is the cautionary tale, Lens 10).

Lens 3 · Competitive Advantages (moats)

1. Integrated retail + generation scale (the headline moat). Vistra is one of the largest competitive residential REPs and a top-tier merchant generator. The integration smooths cash flow and lets it "structure products and contracts in a way that offers significant value compared to stand-alone retail electric providers". The TXU brand has 20+ years of trademark-protected recognition in ERCOT.
2. Irreplaceable nuclear + dispatchable fleet. ~6.4 GW of nuclear (Comanche Peak + the PJM plants) is carbon-free, 24/7, and effectively un-buildable today — it is exactly what hyperscalers will pay 20-year premiums for. This is the moat the AWS/Meta deals monetize. The nuclear PTC ($15/MWh through 2032) floors the downside ``.
3. Bargaining power is rising on the sell side. Pre-AI, a merchant generator was a price-taker. Data-center demand inverts that: hyperscalers need firm carbon-free power more than Vistra needs any single buyer, and there are only a handful of owners of clean baseload at scale (Vistra, Constellation, Talen). That scarcity is the durable edge.
4. Investment-grade balance sheet (newly). S&P raised Vistra to BBB- (investment grade) from BB+ in December 2025 — lowers cost of capital for the buildout.

Where the moat is thin: generation is a commodity; absent the contracted PPAs, margins swing with gas/power prices. Retail churn is real and customer-acquisition cost can exceed margin in competitive markets. The moat is "scale + nuclear scarcity + integration," not a software-style switching-cost lock-in.

Lens 4 · Segments

Segment operating revenues, Adjusted EBITDA, and net income, FY2025 vs FY2024:

The single most important reading of this table: the −$1.868B YoY collapse in net income (to $944M) sits almost entirely in unrealized commodity mark-to-market (East −$1.1B, West −$0.46B, Texas −$0.31B), while Adjusted EBITDA actually rose +$299M to $5,838M. The GAAP earnings line is hedge-accounting noise; the operating business grew. (3-year revenue: $14,779M → $17,224M → $17,738M.)

Geography: generation 46% ERCOT / 51% PJM-East / 3% CAISO; retail is Texas-heavy (2.6M of 5M customers). The East segment is where both the nuclear growth (Meta PPA) and the GAAP volatility live.

---

Phase B — Measure performance

Lens 5 · Earnings Result

FY2025 (10-K):

• Operating revenue $17,738M (+3% YoY, +$514M)
• Operating income $1,906M (down from $4,081M — driven by the $1.8B swing in unrealized MTM losses)
• Net income $944M (−$1,868M YoY); net income to common $752M
• Diluted EPS $2.18 (vs $7.00 FY2024, $3.58 FY2023)
• Adjusted EBITDA $5,838M (+$299M / +5.4% YoY) — the metric management steers on
• Nuclear PTC revenue dropped to $220M from $545M — a real ~$325M operating headwind, not just MTM

Q1 2026 (10-Q): the mirror image —

• Revenue $5,640M (vs $3,933M Q1'25)
• Net income to Vistra $1,029M (vs −$268M loss Q1'25)
• Diluted EPS $2.87 (vs −$0.93) — a single quarter exceeded the entire FY2025 diluted EPS, confirming the GAAP line is dominated by hedge MTM timing
• Adjusted EBITDA $1,475M (+$259M / +21% YoY); segments Retail $68M, Texas $586M, East $801M, West $56M

Guidance: 2026 Ongoing Operations Adjusted EBITDA $6.8–7.6B and Adjusted FCF-before-growth $3.925–4.725B; reaffirmed at Q1. 2027 midpoint opportunity $7.4–7.8B. Critically, guidance excludes Cogentrix and the Meta PPA — both are upside to the printed range.

Balance-sheet flags: cash $785M (down from $1,188M); trade receivables $2,323M (up — watch vs revenue); margin deposits on commodity contracts $1,133M (up from $406M — the hedge book is consuming working capital; $769M of net margin posted in 2025). Operating cash flow was pressured by a $1.611B increase in net margin deposits.

Market reaction: muted-to-negative on the actual prints — the stock fell ~1.9% after the Q1 beat as "the accounting loss overshadowed the operational achievement" ``. The market rewards the contracts (Lens 8), not the quarterly GAAP.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts/ on disk → -sourced. Management's consistent through-line across FY2025 → Q1 2026: (1) "**integrated business model**" as the cash-flow stabilizer; (2) "**long-term contracts underwrite higher base profitability**" — the PPA refrain that intensified after AWS (Sept 2025) and Meta (Jan 2026); (3) capital-return discipline — buybacks + the move to investment grade. Q1 2026 added a forward cash-generation flex: **"$10B+ cumulative cash generation through 2027"** , and disclosed a hedge book ~98% covered for 2026, ~89% 2027, ~65% 2028 `` — i.e., near-term earnings are largely locked, the open upside is 2028+.

What they stopped emphasizing: stand-alone merchant-price commentary. The narrative has shifted from "we're a cheap merchant cash-flow story" to "we're the contracted picks-and-shovels of AI power." Tone is confident/promotional on demand; appropriately hedged on Moss Landing and execution timing.

Lens 7 · Comps

Independent power producers / clean-baseload owners. Multiples are ``, dated; where unsourced, marked n/a.

Read: VST trades at a ~4-turn EV/EBITDA discount to CEG (~10x vs ~14.3x) — but it carries ~3.4x CEG's net debt ($19.6B vs $5.9B). The discount is partly a leverage/commodity-beta discount, not pure mispricing. Goldman prefers to express the power-demand theme via Talen, Vistra, and NRG over CEG on valuation/optionality . Trailing P/E ~31.5x is meaningless here given the MTM-depressed 2025 EPS — the forward number and EV/EBITDA are the only honest gauges.

Lens 8 · Stock-Price Catalysts (last ~5 yrs, moves the market actually reacted to)

`` throughout — pattern matters more than any single tick:

• 52-week range $132.66–$219.82; all-time high $219.82 on 2026-09-22 ``. Spot ~$163.75 (2026-06-18) — ~26% off the high.
• AI/data-center PPA announcements = the dominant up-catalyst. Meta deal (2026-01-09): stock +11.9% with target hikes . "Up ~11% in June [2026] on a $10B AI power deal" . The market pays for contracted demand, full stop.
• Earnings prints = muted/negative when GAAP is MTM-ugly — Q1 2026 beat → −1.9% ``. The tape has learned to look through the GAAP line but won't reward an in-line print.
• Macro overhang: "Vistra reassessed as AI power projects meet higher Treasury yields" `` — long-duration contracted cash flows are rate-sensitive; the de-rate from the Sept-2025 peak tracks the rates/AI-sentiment unwind.
• Regulatory: the FERC PJM co-location order (2025-12-18, Lens 10) is a structural up-catalyst — it removes the regulatory uncertainty that had hung over behind-the-meter/co-located deals.

Conclusion: this stock is a demand-contract + macro-duration trade. It re-rates on PPA signings and de-rates on rates/AI-bubble fear. Earnings quality (GAAP) is a non-event; Adjusted EBITDA + contract backlog are what move it.

---

Phase C — Judge people & books

Lens 9 · Management

CEO James (Jim) Burke (President & CEO; prior CFO/COO/Retail head — an internal operator who knows both halves of the business).

1. Track record ``: relative TSR +291% in 2024 (100th percentile vs S&P 500) and +627% over three years. Whatever one thinks of the multiple, shareholder value creation under Burke has been exceptional.
2. Skin in the game ``: owns 298,002 shares (0.31%, ~$46M). Meaningful in dollars, modest as a % — professional-manager profile, not founder.
3. Capital allocation — strong and disciplined:

• Buybacks: 167M shares / $5.9B retired since Oct 2021 ≈ ~40% of the share count; $1.8B authorization remaining; committed ≥$1B more in 2026. Diluted share count fell 370M → 345M → ~342M (2023→2025→Q1'26).
• Dividends: $306M common paid 2025; preferred $192M.
• M&A: Energy Harbor ($3.065B, nuclear), Lotus ($1.231B final, gas), pending Cogentrix ($4.7B, gas) — a deliberate roll-up of dispatchable + carbon-free capacity into the demand wave.
• Net leverage target <3x ``; achieved investment grade (BBB-, Dec 2025). Capital-allocation history reinvests and returns — value-additive, not value-destructive.

1. Red flags: comp is $16.0M for 2025 (mostly stock) `` — "about average for similar-size companies," not egregious. No related-party deals surfaced. The inherited Ohio HB6 / Energy Harbor history (Lens 10) is a governance overhang but predates Vistra ownership and Vistra opposed the original subsidy.
2. Archetype: professional manager running a capital-cycle playbook (buy dispatchable assets cheap, contract them long, lever modestly, buy back stock). Well-suited to this stage; the risk is over-paying into the AI-demand euphoria (Cogentrix at $4.7B is the test).

Lens 10 · Forensic Red Flags

Accounting risk map:

• Unrealized commodity MTM is the dominant distortion. $808M of unrealized hedging losses ran through 2025 EBITDA-before-adjustments; the East segment alone swung −$1.1B. GAAP net income is therefore not a usable proxy for operating performance — Adjusted EBITDA is the honest line, but it relies on management's add-backs (MTM, purchase accounting, "decommissioning-related activities," insurance income). Scrutinize the bridge, not the headline. This is structurally legitimate (hedge accounting for a generator) but it makes the P&L noisy and gives management wide non-GAAP latitude.
• Working capital / margin deposits: margin deposits jumped to $1,133M (from $406M); a $1.611B increase in net margin deposits depressed operating cash flow. Not fraud — collateral on a large hedge book — but it means reported FCF can lag EBITDA in volatile-price years.
• Receivables $2,323M vs $1,982M: up ~17% vs ~3% revenue growth — modestly outrunning revenue; worth monitoring but explainable by weather/consumption and Energy Harbor consolidation.
• Goodwill/intangibles & impairments: $228M of long-lived-asset impairments in 2025 (Moss Landing, development projects) — management is taking writedowns rather than hiding them, a positive sign.
• Insurance income add-backs: $191M of insurance income (Martin Lake / Moss Landing) excluded from Adjusted EBITDA — appropriately, but it inflates the GAAP "other income" line.
• SBC: $113M non-cash comp added back — modest relative to a $5.8B EBITDA base; non-GAAP is not materially SBC-flattered (unlike a tech name).

Regulatory findings (required sub-section):

• SEC: No Litigation Releases or AAERs naming Vistra in 2021-06-23 → 2026-06-23, verified via SEC EDGAR EFTS (LR + AAER).
• Item 3 / Note — Legal Proceedings:
• Ohio HB6 (RICO): nuclear-subsidy bribery scandal inherited via the Energy Harbor merger. Vistra opposed the subsidy; it was repealed before any funds were distributed. Ohio AG RICO case vs. FirstEnergy/Energy Harbor entities is stayed; motions to dismiss pending. Reputational overhang, not (on current facts) a material Vistra liability.
• Dorrell antitrust (filed July 2025): alleges nuclear-industry wage-fixing/comp-info exchange since ~2003; Vistra + Luminant are co-defendants with Constellation + 25 others. Motions to dismiss filed Dec 2025. Early stage; Vistra "intends to defend vigorously."
• Winter Storm Uri MDL: generator defendants won dismissal (First Court of Appeals, Dec 2023); plaintiffs petitioned the Texas Supreme Court (briefed Sept 2025). Favorable posture for Vistra.
• Moss Landing battery fire (Jan 16, 2025): multiple CA federal/state suits vs. Vistra + LG Energy Solution; EPA ASAOC remediation ~$110M (≈$70M spent through Q1'26, ~$40M accrued). $500M insurance limit fully collected by Feb 2026. The financial hit is largely insured + reserved; litigation tail is open.
• Illinois AG vs IG&E: pre-acquisition retail-marketing conduct; narrowed by statute of limitations.
• Non-SEC enforcement: no new material FTC/DOJ/FDA/CFPB action surfaced beyond the inherited HB6 matter and the routine environmental/EPA proceedings (GHG rule, Good Neighbor Plan, Regional Haze) which are industry-wide and currently trending favorably under the new EPA posture.

Net forensic read: clean SEC record; no accounting-fraud markers. The real "red flag" is interpretive — GAAP earnings are MTM-distorted and management's Adjusted EBITDA bridge requires trust. Legal exposure is mostly inherited (HB6), insured (Moss Landing), or favorably postured (Uri). Leverage (Lens 7) is the balance-sheet risk, not the accounting.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 / FY2027 / FY2028)

GAAP EPS is uninformative here (MTM). The defensible forward path is built on management's Adjusted EBITDA guidance (largely locked by the ~98%/89%/65% hedge book) and a buyback-driven share-count glide. All inputs labeled; outputs ``.

Anchor: FY2026 Ongoing Ops Adj EBITDA $6.8–7.6B (mid ~$7.2B); FY2027 opportunity $7.4–7.8B (mid ~$7.6B) — both exclude Cogentrix + Meta PPA.

Bull/base/bear (FY2027 anchor):

• Bull: load growth + tight power markets push EBITDA above the $7.8B top, Cogentrix + Meta accretive, 2028 open position prints high → EPS power $10+, stock re-rates toward CEG's multiple.
• Base: guidance delivered (~$7.6B), buyback continues, modest accretion → adj. EPS ~$8.5, stock tracks ~10–12x EBITDA.
• Bear: ERCOT/PJM load disappoints, power prices soften into a now-larger fleet, Cogentrix dilutes near-term, rates stay high → EBITDA flat ~$7B, multiple compresses, equity de-rates given leverage.

Brier forecast: not logged — per --watchlist rules, skip forecast.ts create in the unattended sweep. Candidate base call for a later human-gated pass: "VST FY2026 Ongoing Ops Adjusted EBITDA ≥ $7.2B (midpoint), resolves 2026-12-31, p≈0.70" given the ~98% 2026 hedge.

Lens 12 · Bull vs Bear

Bull case. Vistra owns ~6.4 GW of irreplaceable carbon-free nuclear plus ~38 GW of dispatchable capacity into the largest power-demand shock in a generation. The AWS (1,200 MW) and Meta (2,609 MW) 20-year PPAs convert merchant exposure into contracted, investment-grade cash flow — and aren't even in guidance yet. The ~98%/89% hedge book locks 2026–27; Cogentrix adds 5.5 GW of gas for peakers exactly when grids need firming. Management has retired ~40% of shares and reached investment grade. At ~10x forward EBITDA — a 4-turn discount to Constellation — this is the cheapest large-cap expression of the data-center power trade, with $10B+ of cash generation through 2027 to fund both growth and buybacks. The contrarian point: the market is paying for the contracts it can see and ignoring the 2028 open position + Cogentrix + uprate optionality.

Bear case (permanent-impairment risks).

1. Leverage in a commodity business. ~$19.6B net debt (vs CEG $5.9B) against a P&L that can swing $1.8B on hedge MTM. If power prices fall into a now-larger fleet (post-Cogentrix), the equity is the shock absorber. Investment grade is new and thin (BBB-).
2. The demand may not show up — or may bypass the grid. The entire re-rating rests on ERCOT/PJM load growth materializing. If AI capex slows, or hyperscalers go behind-the-meter / off-grid (gas turbines, SMRs on-site), centralized generators lose the very demand they're being valued on ``.
3. Rate/duration sensitivity. 20-year contracted cash flows de-rate when Treasury yields rise — the stock already fell ~26% from its Sept-2025 high partly on this ``.

Pre-mortem (18 months out, thesis broke): AI data-center order growth decelerated; a couple of marquee hyperscaler projects went behind-the-meter; ERCOT added so much gas/solar/storage that scarcity pricing collapsed; Cogentrix closed near the top and looked expensive; and with $20B+ of debt, the equity de-rated to ~7x EBITDA. The GAAP optics (another MTM-ugly year) gave bears the headline.

Are multiples too high? No — at ~10x forward EBITDA / ~18x forward P/E, VST is the cheap one in the cohort. The risk is earnings/EBITDA disappointing, not multiple excess.

Contrarian view of what the market refuses to see: the bears fixate on the GAAP loss-quarters and the leverage; the market underprices that ~89% of 2027 is already hedged and the two hyperscaler PPAs are additive to guidance, not in it. Conversely, bulls underprice how much of the thesis is a single macro bet (AI load growth) on a levered balance sheet.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case:

• Where the money really comes from: strip out the contracted nuclear PPAs and Vistra is still, today, predominantly a merchant generator + retailer whose earnings are set by power prices it doesn't control. The "integrated hedge" smooths but does not eliminate commodity beta — and the $1.8B MTM swing proves the P&L is violent.
• Concentration risk, inverted: the bull thesis concentrates on two counterparties (AWS, Meta) and a handful of nuclear units for the entire growth narrative. A delay in the 433 MW of unbuilt uprates, a regulatory snag on co-location, or one hyperscaler renegotiating would gut the story. And 433 MW of uprate capex runs 2026–2034 — this is a long, execution-heavy build, not a signed-and-done cash flow.
• The most dangerous competitor bulls underestimate: not Constellation — it's behind-the-meter self-generation and SMRs. If hyperscalers decide to own their power (on-site gas, fuel cells, modular nuclear), the grid-scale IPP gets disintermediated from the highest-growth demand. The 10-K itself flags behind-the-meter crypto/AI load as a risk to the centralized model.
• Worst capital-allocation tail: buying 5.5 GW of gas (Cogentrix, $4.7B) at the top of a demand cycle, into a decarbonizing East — if gas economics or carbon rules turn, that's a stranded-asset risk on top of the debt.
• Assumptions that must hold for $163: sustained ERCOT/PJM load growth, power prices firm into a bigger fleet, Cogentrix accretive, uprates on time/on budget, rates not spiking. Knock 20–30% off the load-growth assumption and the EBITDA glide flattens — on ~$20B of net debt, the equity is geared straight into that miss.
• Single permanent-impairment scenario: AI-capex air-pocket + behind-the-meter shift collapses merchant scarcity pricing while $20B+ of debt sits on the books. Plausibility: moderate — not a base case, but not tail-risk-remote given how much demand is speculative and how fast hyperscalers can change procurement strategy.

Lens 14 · Management Questions (ordered by information value)

1. Of the 2027 EBITDA "opportunity" ($7.4–7.8B), how much is already contracted vs. dependent on open merchant prices — and what's the 2028 open-position sensitivity to a $5/MWh move in ERCOT/PJM power?
2. The AWS + Meta PPAs aren't in guidance — quantify their steady-state annual EBITDA contribution and the year each reaches full run-rate.
3. Post-Cogentrix, what is pro-forma net leverage, and what gas-price / capacity-price assumptions underwrite the $4.7B purchase price?
4. If a major hyperscaler customer chose behind-the-meter or on-site SMR generation instead of a grid PPA, how does that change your 5-year demand thesis?
5. The 433 MW of nuclear uprates span 2026–2034 — what are the gating regulatory/engineering milestones, and what's the cost-overrun exposure?
6. How much of FY2026 Adjusted EBITDA is "ongoing operations" vs. one-time insurance/MTM add-backs — walk through the bridge from GAAP net income.
7. What's the long-run capital-return split between buybacks and debt paydown now that you're investment grade — does the leverage target tighten below 3x?
8. On the FERC PJM co-location order (Dec 2025), which specific co-location/behind-the-meter deals does it unlock for Vistra, and on what timeline?
9. What is the replacement/return-to-service decision tree for Moss Landing 350 MW, and the worst-case incremental cost beyond the $110M ASAOC?
10. How dependent is the nuclear PTC ($15/MWh through 2032) on the gross-receipts phase-out interpretation, and what's the revenue-at-risk if Treasury guidance tightens?
11. What's the retail customer-count and churn trend ex-acquisitions, and is retail margin sustainable or were 2025's "one-time supply-cost gains" non-recurring?
12. How do you think about M&A discipline from here — is Cogentrix the last big gas deal, or is more dispatchable roll-up coming?
13. What's your hedging philosophy for 2028+ as the open position grows — do the hyperscaler PPAs change how much merchant length you carry?
14. How exposed is the East segment's GAAP volatility to a structural fix (more NPNS designations) vs. living with the MTM noise?
15. What capacity-market or transmission-policy change would most improve — or most threaten — the economics of the fleet over the next five years?

---