Vistra Corp. (NYSE: VST) — Strong Execution, Softer Texas, Capped Scarcity

Prepared as: Independent Equity Research Report date: 2026-08-14 · Update of: 2026-06-13 initiation Price (2026-08-13 close): $146.40 · 52-wk range: $132.66–$219.82 · Market cap: ~$49.1B · Enterprise value: ~$71.1B · Shares out: 335.6M Sector / sub-industry: Utilities — Independent Power Producers (merchant generation + competitive retail)

⚡ Claude’s Take

Claude’s independent subjective opinion, provided for general information and not investment advice. The institutional analysis below carries no position and no price target.

Verdict: HOLD; accumulate on a reset into roughly $125–$135, with no change from the June call. At $146.40, Vistra is neither a short nor a clear bargain. The latest quarter validates the operator: second-quarter Ongoing Operations Adjusted EBITDA rose 31% to $1.77 billion, first-half EBITDA rose 26% to $3.26 billion, fleet availability exceeded 97% during July peaks, and 2026 guidance stayed intact. It also validates the caution: current ERCOT forwards are now “meaningfully lower” than the October 2025 curve behind the 2027 opportunity range, management says the legacy business trends toward that range’s low end, retail volumes contracted, and no incremental premium power contract was signed. PJM scarcity is real—the latest auction was 6.8 GW short—but its rent is censored at the $325/MW-day administrative cap.

The price has barely moved from $148.02 at the June report, yet the evidence mix deteriorated modestly. Current enterprise value is about $71.1 billion, or 9.9× the $7.2 billion midpoint of 2026 Adjusted EBITDA guidance. The $4.325 billion midpoint of FCF before growth implies an 8.8% equity yield, but that measure excludes a rising growth-capital burden. A transparent post-Cogentrix normalized range is wide: roughly $85–100 in a lower-curve/de-rating case, $145–165 if the 2027 opportunity plus Cogentrix arrives and leverage is partly worked down, and $200–240 if contracted load, fleet utilization and buybacks compound while the multiple holds. Those are scenario outputs, not precision. My preferred entry zone remains below spot because a commodity merchant deserves a margin of safety against simultaneous EBITDA and multiple compression.

Helix is strategically interesting, not yet earnings. Vistra can commit up to $1 billion and is a preferred—but nonexclusive—power partner; there is no disclosed contracted pipeline, and each project must still meet the company’s mid-teens return threshold. Cogentrix has FERC approval but remains pending for late 2026. Moss Landing worsened: restart is uncertain, cleanup is estimated at $175 million, and $311 million of carrying value remains exposed. Those facts make the capital-allocation test harder just as repurchases have migrated from an average near $38 since 2021 to $153.34 in the latest quarter.

Conviction: medium. The call turns more constructive if ERCOT forwards stabilize, incremental PPAs disclose attractive volume/term/economics, and Cogentrix delivers the promised per-share accretion without leaving leverage elevated. It turns more negative if the 2027 low-end bias becomes a formal cut, Texas supply keeps outrunning realized load, Moss Landing requires another impairment, or the merchant multiple contracts before the buyback can offset it. Best operator in a bad industry remains the right frame; the quarter strengthened “best operator” and the curve strengthened “bad industry.”

📈 Stock Price Action — Five-Year Event Map

VST’s five-year chart is a business-model reappraisal followed by an AI-power overshoot and partial unwind. Price facts below come from the AZI daily-price history through August 13, 2026; attributed causes are interpretations reconciled to filings and dated company/regulatory events.

The stock’s trailing location is therefore neither an untouched one-way street nor a clean fundamental recovery. It is roughly flat since the prior report, below all three medium/long trend averages, and still about one-third under the peak. FactorsToday’s sparse model treats the empirical factor result as positioning evidence only; the thesis continues to rest on forward curves, contracted margins and capital allocation.

Changes Since 2026-06-13

• Confirmed: operating execution and diversification. Q2/H1 Ongoing Operations Adjusted EBITDA grew 31%/26%; East and Texas generation offset weaker first-half Retail EBITDA, and 2026 guidance was reaffirmed.
• Dented: the near-term bull case. ERCOT forwards are “meaningfully lower,” legacy 2027 earnings now bias toward the low end of $7.4–$7.8 billion, and no new premium PPA was disclosed. The baseline bull falsification test is partially triggered, not completed, because formal guidance did not fall.
• Strengthened offset: PJM’s 2028/29 auction cleared every zone at the $325 cap while capacity was 6.8 GW short of the reliability requirement. Vistra cleared 10,924.4 MW, implying about $1.30 billion of annual gross capacity revenue before performance/seasonality adjustments, but the uncapped scarcity value cannot reach shareholders under the collar.
• New optionality and obligation: Helix gives VST a preferred, nonexclusive channel into digital infrastructure and asks for up to $1 billion before project economics are disclosed. Cogentrix received FERC approval but had not closed.
• New risks: Texas ordered a comprehensive data-center audit that lowers the evidentiary value and may slow realization of the 400+ GW queue. Moss Landing’s remaining restart is “when or if,” cleanup rose to $175 million, and $311 million of carrying value remains.
• Valuation: price is nearly unchanged, current 2026 midpoint EV/EBITDA is about 9.9×, and midpoint FCFbG yield is about 8.8%. The valuation conclusion has slightly less fundamental slack than in June.

1. Executive Summary

Vistra Corp. is the largest competitive (merchant) power producer in the United States by generation volume and the largest competitive electricity retailer, serving ~5 million customers across 18 states and the District of Columbia from a fleet of ~43,641 MW (pre-Cogentrix) of natural gas, nuclear, coal, solar, and battery capacity. It was forged out of the 2016 bankruptcy reorganization of Energy Future Holdings (the old TXU), bulked up via the 2018 Dynegy merger, the 2019 Ambit retail deal, the March-2024 Energy Harbor nuclear acquisition, the 2025 Lotus gas deal, and the pending late-2026 Cogentrix gas acquisition. Cogentrix would add ~5.5 GW for an unadjusted economic package of about $4.725B before tax benefits and final adjustments. VST is run by CEO Jim Burke and CFO Kris Moldovan from Irving, Texas.

The investment tension is the same one that governs the entire merchant-power complex, sharpened by VST’s specific asset mix. Merchant generation is, historically, a structurally bad business: a megawatt-hour is a perfect commodity, price is set by the marginal unit, there are no barriers to entry, and the sector’s capital cycles have repeatedly bankrupted their participants (Calpine, NRG, Mirant, Dynegy, and VST’s own TXU predecessor). Greenwald’s framework is unforgiving here — no supply, demand, or scale advantage means ROIC mean-reverts to the cost of capital. Against that prior, two things distinguish VST: (1) a ~6.4 GW nuclear fleet that carries the same federal 45U PTC revenue floor as CEG’s — but at ~15% of capacity, not two-thirds; and (2) a demonstrated operating and capital-allocation competency — buying assets cheaply in distress, integrating them, and routing the cash into a share count that has shrunk ~28% in four years. The first is a narrow, government-granted moat on a minority of the business; the second is a real edge but a capability, not a barrier to entry.

The de-rate is still the valuation story, while the quarter is now the operating story. From the $219.82 high, VST is down 33.4% to $146.40. It is 2.1%, 4.2% and 8.0% below its 21-, 50- and 200-day exponential averages; adjusted total returns are +3.3%, –10.0% and –28.3% over three, six and twelve months. Yet Q2 produced $1.767B of Ongoing Operations Adjusted EBITDA and maintained the annual range. Current valuation is ~9.9× the 2026 EBITDA midpoint and an 8.8% midpoint FCFbG yield. AZI’s own-history composite remains elevated at 75.2 (P/S 91st percentile; P/B 84th), only modestly below the prior 77th-percentile reading. The drawdown removed euphoria, not enough cyclical risk to make the stock historically distressed.

What the analysis finds: (1) No durable franchise moat — a commodity merchant plus a competed, low-switching-cost retail book, overlaid by a 45U floor on a minority nuclear slug and a genuine acquire-integrate competency; “best operator in a bad industry.” (2) GAAP earnings are noise — non-cash mark-to-market on commodity hedges swung net income from +$2,659M (FY24) to +$944M (FY25) while Adjusted EBITDA rose; value VST only on Adjusted EBITDA and Free Cash Flow before Growth (FCFbG), never P/E or P/B. (3) A genuine cash machine — Adjusted EBITDA roughly tripled ($1.9B FY21 → $5.84B FY25), FCFbG grew to ~$3.6B (guided $3.9–4.7B FY26), funding the buyback and an IG upgrade — but the cash is commodity-/weather-exposed, lumpy (collateral swings turned CFO negative in 2021–22), and excludes a large, back-end-loaded growth-capex program. (4) A structurally cyclical industry at or near a cyclical high — PJM capacity at the auction cap (down only via clawback), ERCOT forwards softening as supply arrives — where VST’s geographic mix concentrates its earnings in the most-contested, least-protected corner (energy-only ERCOT). (5) Strong-but-tested capital allocation — a value-creating buyback and a well-aligned, per-share-based incentive plan, now plowing rising debt and dilutive equity into gas-asset M&A at a euphoric moment, with insiders selling (all 10b5-1) and none buying the dip.

No recommendation and no price target appear below this summary, per firm policy; the body discusses valuation only as embedded expectations and scenarios. The single deliberate exception is the labeled Claude’s Take block above.

2. Business Overview

What Vistra is. Vistra is an “integrated retail electricity and power generation company” — a merchant power producer bolted to the largest competitive retail electricity book in the United States. It owns power plants, sells the output into wholesale markets (ERCOT, PJM, ISO-NE, MISO, NYISO, CAISO) or under bilateral contracts, hedges that output forward, and separately serves ~5 million end customers through a portfolio of retail brands. It is not a regulated utility: it owns no rate base, earns no allowed return, and bears commodity-price risk. (FACT — 10-K FY2025, Item 1.)

The integrated model — the core structural argument. Management’s central claim is that generation and retail form a natural hedge: generation is a short physical position (you must sell power), retail load is a long demand position (you must buy power to serve customers), so the two offset and the combined entity carries lower earnings volatility than either standalone. Q1-2026 was a live demonstration — an exceptionally mild ERCOT winter hurt the retail segment (lower demand, compressed margins), but the generation segment more than offset it, producing a record calendar-first-quarter Adjusted EBITDA of ~$1.49B. (FACT — Q1-2026 call, 2026-05-07.) This is genuine diversification and a real volatility damper. Whether it is a moat is a separate question, addressed in (it is not).

Q2 extended the hedge evidence and exposed its limits. Ongoing Operations Adjusted EBITDA was $1,767M versus $1,349M, with Retail $773M, Texas $311M, East $642M and West $68M. For the half, Generation expanded while Retail EBITDA fell to $841M from $940M and retail volumes declined 7.0%. A portfolio that offsets weak load with higher generation margin is valuable; a customer book whose volume shrinks is not captive. The correct conclusion is smoother consolidated cash flow, not a franchise multiple. (Q2 2026 10-Q, filed 2026-08-10.)

Reporting structure. Through FY2025 VST reported five segments: Retail, Texas (ERCOT generation), East (PJM/ISO-NE/MISO/NYISO generation, now including the Energy Harbor nuclear), West (CAISO), and Asset Closure (retiring/decommissioning coal and the Moss Landing battery remediation). FY2025 segment Adjusted EBITDA: Retail $1,622M, Texas $1,834M, East $2,282M, West $244M, Asset Closure –$74M, Corp –$70M. Management has begun guiding on a simplified two-bucket “Ongoing Operations” basis — Generation (~$4,290M FY25) and Retail (~$1,622M FY25) — which strips the perennially-negative Asset Closure segment and flatters the reported total by ~$75–150M/year. (FACT — 10-K FY2025 segment tables; Q3/Q4-2025 calls.)

The fleet (12/31/2025, pre-Cogentrix): 43,641 MW. Natural gas 26,989 MW (62%) — 28 combined-cycle plants (22,167 MW) plus 12 peakers; coal 8,743 MW (20%), being retired/converted; nuclear 6,448 MW (15%) — six units at four sites; solar/battery 1,274 MW (3%). By region: Texas/ERCOT 19,858 MW (46%), East 22,254 MW (51%), West 1,529 MW (3%). Expected annual generation exceeds 230 TWh. Fleet utilization runs ~60%, leaving substantial headroom to absorb load growth without new build. (FACT — 10-K FY2025, Items 1–2.)

The nuclear fleet — six units, four sites. Comanche Peak Units 1 and 2 (2,400 MW combined, ERCOT, licensed to 2050/2053) are the legacy Texas plant; Beaver Valley Units 1 and 2 (1,872 MW, PJM), Perry (1,268 MW, PJM), and Davis-Besse (908 MW, PJM) came via the March-2024 Energy Harbor acquisition. The nuclear fleet is the only part of VST with a government-backed revenue floor (the 45U PTC; see ). (FACT — 10-K FY2025.)

The retail book — ~5 million customers. TXU Energy (a 20±year trademarked brand, ~2.6M Texas customers) anchors a portfolio that also includes Ambit, Dynegy Energy Services, Energy Harbor retail, Homefield, TriEagle, Public Power, and US Gas & Electric. Retail is capital-light and genuinely high-return, and it is the counter-cyclical hedge described above. FY2025 retail Adjusted EBITDA of $1,622M was a record, flattered by non-repeating Energy Harbor integration gains; management explicitly guides to a normalized run-rate of ~$1.4B. (FACT — 10-K; Q4-2025/Q1-2026 calls.)

How it makes money — the revenue stack. (1) Merchant energy — wholesale and retail sales of generation at market prices; the commodity-exposed core. (2) Capacity payments — only in PJM/ISO-NE/NYISO (the East segment); ERCOT is energy-only and pays no capacity. (3) Ancillary services and ERCOT scarcity pricing — paid when the grid is tight. (4) Nuclear 45U PTC — the downside floor on ~6.4 GW. (5) Retail margin — unit margins on ~5M customers’ load. The mix matters enormously: VST’s single largest geography (ERCOT) earns no capacity annuity, only volatile scarcity-energy margins, which is the heart of the bear case (, ).

Verdict (business model): A merchant generation + competitive-retail business whose distinguishing features are (a) the largest competitive retail book in the U.S., providing a real natural hedge that lowers earnings volatility, and (b) a ~6.4 GW nuclear slug with a federal PTC floor. It is not a regulated annuity; it is a commodity merchant with a partial hedge and a partial government backstop, run by an unusually good capital allocator. Cash-flow generation is real but commodity-, weather-, and forward-curve-dependent.

3. Industry Dynamics

The base rate is bad. Through both the Greenwald and Marathon lenses, merchant competitive power generation is a structurally bad industry. Output is a perfect commodity, there is no differentiation, price is set by the marginal generating unit, and the sector has run repeated boom/bust capital cycles that destroyed equity (the early-2000s merchant build-out bankrupted Calpine, NRG, Mirant, NEG; VST’s own ancestor TXU/Energy Future Holdings was the largest leveraged-buyout bankruptcy in history). In Greenwald’s terms: no barriers to entry → strategy is irrelevant, only operating efficiency matters, and ROIC mean-reverts to WACC. Any bull case must specify what, precisely, overrides that prior — and for how long.

The defining structural fact: VST is not CEG. The critical distinction is geographic and fuel mix. VST’s fleet is ~46% ERCOT (energy-only, no capacity market), ~51% PJM-led East (capacity market), ~3% CAISO — and only ~15% nuclear. CEG is ~two-thirds nuclear with a federal PTC floor under most of its output. VST therefore sits on the less-protected side of the same demand window.

ERCOT — energy-only, capped on the upside, and supplying its way back toward the mean. ERCOT is unique among U.S. markets in having no capacity payments: generators are compensated only through energy and ancillary-service prices, supplemented by scarcity-pricing adders when reserves run thin. This rewards VST’s flexible gas and batteries only when scarcity actually shows up — higher-volatility, lower-floor than CEG’s PJM capacity annuity. Three post-Uri (2021) features now truncate the upside: the Operating Reserve Demand Curve was replaced by a redesigned scarcity adder (effective Dec-2025); a “peaker net margin” circuit-breaker and the PUCT Emergency Pricing Program both chop the effective system cap toward ~$2,000–4,500/MWh (down from the old $9,000/MWh that made 2021–2023 so lucrative); and the proposed Performance Credit Mechanism — a quasi-capacity construct that would have added ~$1B/year of generator revenue — was shelved by the PUCT. ERCOT stays energy-only; no capacity floor is coming. Meanwhile the Texas Energy Fund is actively subsidizing new gas entry ($5B+ of 3%-rate state loans; first project interconnected April 2026) — VST’s own 10-K flags the TEF as a risk that “may materially change market fundamentals,” i.e., depress incumbent energy prices. (FACT — 10-K FY2025 Item 1A; PUCT; web, 2026.)

The ERCOT supply response is already arriving — the Marathon tell. Battery storage in ERCOT roughly doubled to ~16 GW by mid-2026, solar continues to surge, and TEF-subsidized gas is interconnecting. Management itself confirms this is softening ERCOT forward power prices and states bluntly that ~three years of battery builds have “returned virtually nothing.” This is classic late-boom mean reversion — capital flooding a hot theme, depressing the very margins that attracted it — and it is biting in VST’s single largest market right now. (FACT — Q4-2025/Q1-2026 calls; ERCOT data, web 2026.)

The August curve check moved from qualitative to explicit. Management said ERCOT forwards were “meaningfully lower” than the October 31, 2025 curve used for the $7.4–$7.8B 2027 opportunity and that legacy operations now trend toward the low end. As of August 3, the portfolio was approximately 100% hedged for 2026, 94% for 2027 and 72% for 2028, so the damage is deferred and partly buffered rather than absent. Q2 average ERCOT North/West prices fell to $30.20/$24.71 per MWh from $32.02/$31.44; PJM West rose to $51.43 from $42.35. That divergence is the portfolio hedge in market form. (FACT — Q2 earnings release, 2026-08-07; Q2 call.)

PJM (~51%) — undeniable shortage, administratively censored economics. The 2028/29 Base Residual Auction cleared every zone at the $325/MW-day cap. Cleared plus fixed-resource capacity was 149,181.6 MW, 6,831.3 MW below the reliability requirement; the installed reserve margin was 14.4% against 20.0% required. New generation plus uprates contributed only 524.7 MW even as forecast load rose 1,374.5 MW. VST cleared 10,924.4 MW at a $325 weighted average, equivalent to about $1.30B of gross annual capacity revenue before performance and seasonal adjustments. The shortage validates incumbent scarcity; the collar means the price signal cannot fully monetize it. (PJM auction report, 2026-07-14; VST 8-K, 2026-07-14.)

Demand — records are real; queues are not demand. ERCOT set a 91,089 MW all-time record on July 22. Management retains 4–6% ERCOT and 2–3% PJM annual load-growth assumptions through 2030, but estimates credible Texas data-center load at only 12–15 GW versus a gross queue above 400 GW. Governor Abbott’s August 3 audit can pause or reject connections and cites more than 474 GW of requests, roughly 90% data centers; a June directive requires large loads to bear their infrastructure costs. Those rules may improve economic discipline, but they also delay realization. Record demand supports existing assets; the queue cannot support valuation without project screening. (ERCOT record, updated 2026-07-27; Texas audit, 2026-08-03.)

Policy. The IRA Section 45U nuclear PTC (up to $15/MWh, phasing out as gross receipts rise between ~$26.00 and ~$44.75/MWh in 2025, inflation-indexed through 2032) is the only hard floor — and it covers just ~6.4 GW of ~44 GW. VST recognized $545M (FY24) then $220M (FY25) of transferable nuclear PTC; the drop as power prices rose confirms it is a downside floor, not a driver. The 2025 “One Big Beautiful Bill” (OBBBA) accelerates the solar/wind credit phase-out — a material headwind to Vistra Zero (VST has confirmed deferral/abandonment of some solar/battery capex) but also a partial offset (it slows the renewable flood depressing ERCOT prices). Federal coal/gas environmental pressure is receding (proposed EPA GHG-rule repeal). (FACT — 10-K FY2025; web 2026.)

Cyclicality. Power prices are closer to a cyclical high than a low: PJM capacity at the cap (downside only), ERCOT forwards softening, post-Uri caps on the right tail. 2026–2027 is heavily hedged (near-term locked), so the cyclical risk is concentrated in 2028+, as hedges roll into a softer, more-supplied curve. VST is implicitly long natural gas (gas sets the marginal price in ERCOT), with the integrated retail book as the counter-cyclical hedge.

Verdict (industry): structurally cyclical, no-moat commodity industry in a favorable-but-maturing demand phase — and structurally weaker for VST than for CEG. Merchant power has zero entry barriers and mean-reverting returns. VST sits on the less-favorable side of the AI window: ~46% of its fleet is in energy-only ERCOT with no capacity floor, truncated upside, and a fast-arriving subsidized + battery supply response already softening forwards; its ~51% PJM exposure rides the same politically-contested capacity windfall CEG depends on but with far less PTC-floored nuclear. Capital-cycle location: late-boom / early supply response, with ERCOT meaningfully closer to the top of the cycle than PJM. The offsets (retail hedge, firm-gas scarcity from the turbine bottleneck, a genuine ~5–6%/year demand tailwind) are real but bounded.

4. Competitive Position

The moat question, asked precisely. A moat exists only if removing it would cause returns to deteriorate toward the cost of capital. For VST, we test each candidate advantage against Greenwald’s taxonomy (supply/cost, demand/captivity, economies-of-scale + captivity) and the financial outcome it should produce.

Generation: no moat. Eighty-two percent of capacity is gas and coal — mid-merit-order thermal that is a pure price-taker on power and spark spreads. There is no supply advantage (fuel and technology are available to all), no demand captivity (the grid buys the cheapest cleared MWh), and no scale-plus-captivity (the market clears on marginal cost regardless of fleet size). This is the canonical commodity with zero barriers to entry. The genuinely new entrants — TEF-subsidized gas, merchant batteries — are arriving precisely because returns looked attractive, the Marathon signature.

Nuclear: a narrow, government-granted edge — but small. The ~6.4 GW nuclear fleet carries the same two advantages CEG’s does: (a) a near-absolute regulatory barrier to new nuclear (irreplaceable, supply-starved carbon-free baseload), and (b) the 45U PTC revenue floor. This is a genuine, financializable advantage — but at ~15% of capacity it protects a far smaller share of VST than of CEG (~two-thirds). The irreplaceable-asset moat is real and proportionally minor.

Retail: demand captivity, but weak and competed. TXU is a real brand with ~20 years of equity and ~2.6M Texas customers; the broader book serves ~5M. Greenwald would classify this as weak demand-side captivity from habit, brand, and switching inertia — in a market (ERCOT’s Power-to-Choose) deliberately engineered for easy switching. Retail is high-ROIC and a real earnings stabilizer, but it is a competed book with low switching costs, not a structural barrier. The share-stability test is suggestive but not decisive: VST’s ERCOT residential share (~32% post-Ambit) leads but is contested by NRG and is stable-to-eroding at the margin.

The newest outcome strengthens that caution. Retail volumes fell 4.4% in Q2 and 7.0% in the first half; first-half Retail Adjusted EBITDA fell 11% even as consolidated EBITDA grew 26%. Weather contributed, but customer-volume contraction is the opposite of a demonstrated captivity advantage. By contrast, generation availability above 97% during July peaks is evidence of operating competence. Greenwald’s distinction matters: competence can preserve margin and win contracts, while captivity would make the demand base hard to lose. VST demonstrates the first, not the second.

The one genuine, financializable edge is capability, not a barrier: capital allocation + acquire-integrate-at-scale. VST has repeatedly bought assets cheaply in distress (Dynegy out of the post-2008 merchant bust, Energy Harbor out of FirstEnergy’s bankruptcy reorganization, Lotus, Cogentrix), integrated them, and routed the cash into a share count that fell ~28% in four years at an average cost under $36. That is a real, demonstrated cost-discipline/operating competency that has compounded per-share FCF — but it is a management capability, not an entry barrier, and it is precisely the kind of edge that erodes if the team changes or overpays late in the cycle.

The headline ROE is a trap. VST’s ~43% trailing ROE is not a moat signal: it is flattered by (a) a thin, post-bankruptcy, buyback-shrunk equity base (~$5–5.6B equity on ~$17–19B debt; D/E ~3x) and (b) cyclically peak power prices. Through-cycle, with mean-reverting merchant margins and a normalized equity base, returns are far more pedestrian. Do not read 43% ROE as evidence of franchise quality.

Direct comparison. On the quality axis (nuclear scarcity + contracting), the order is CEG > VST ≈ Talen > NRG. On the cheapness axis, TLN < NRG < VST < PEG < CEG. VST is the mid-quality / mid-price merchant: you pay less than CEG and accept a more commodity-exposed, less-moated, more-ERCOT-concentrated book, backed by a stronger capital-return engine than most peers.

The AI/data-center thesis — genuinely contracted, but read it carefully. VST has signed ~3.8 GW of nuclear under 20-year PPAs — more than any U.S. power company: a 1,200 MW AWS deal at Comanche Peak (Amazon sites a facility on VST land, brings 1-for-1 backup gen; energization Q4-2027, full ramp Q4-2032; options on uprates/SMRs), and January-2026 Meta PPAs covering 2,176 MW of operating capacity (Perry + Davis-Besse) plus 433 MW of uprates. The Meta deals are front-of-the-meter (not colocation-dependent), insulating them from PJM colocation-rule risk. Management projects ~25% adjusted-FCFbG accretion at full nuclear-PPA ramp. The skeptic’s distinction: the operating-capacity PPAs largely re-price existing output — the plant keeps flowing to the grid, but the price is now contracted rather than merchant. That is valuable de-risking (converting commodity exposure to a 20-year annuity), but it is not incremental MW; the genuinely new capacity is only the ~633 MW of long-dated uprates (2031–2034). VST is also steering toward gas data-center structures (colocation, bridge power, new-build gas with the customer taking gas risk), which are broader but lower-premium than CEG/Talen’s pure carbon-free-baseload scarcity play.

Helix is access plus capital exposure, not an exclusive distribution moat. The June partnership with KKR, Kuwait Investment Authority and NVIDIA makes VST a founding investor and preferred power provider. The 10-Q specifies an initial $500M commitment plus another $500M once power-supply milestones are reached—or earlier at VST’s election—and accounts for it as an unconsolidated equity-method investment. Management clarified that preferred is nonexclusive, each power project is separately negotiated, and VST has no supply obligation. The relationship can originate attractive projects, but there is no disclosed contracted pipeline or network effect. Up to $1B must therefore be underwritten as exposed capital until contracts disclose volume, term, counterparty and returns.

Verdict (competitive position): no durable franchise moat — a crowded commodity market with weak differentiation, overlaid by a 45U floor on a minority nuclear slug, a weak/competed retail captivity, and a strong capital-allocation capability. “Best operator in a bad industry” is the honest characterization. The only advantages that survive the financial-outcome test are the narrow nuclear scarcity/PTC floor (small) and the demonstrated capital-allocation competency (real but a capability, not a barrier). The integrated gen+retail model is diversification, not a moat. The ~3.8 GW of signed nuclear PPAs is the single best new development — it converts merchant exposure to contracted annuity on the highest-quality part of the fleet.

5. Growth History and Forward Opportunities

Reported revenue is the wrong lens. Revenue grew from $12,077M (FY21) to $17,738M (FY25) — but this figure is gross-settled, riddled with commodity pass-through and large intersegment eliminations (FY25 elimination/corporate –$8,528M), and rose only ~47% while Adjusted EBITDA tripled. EV/Revenue is meaningless for VST. (FACT — 10-K FY2025.)

Adjusted EBITDA is the real growth story — and it is largely acquired and cyclical, not organic. Total segment Adjusted EBITDA: $1,908M (FY21, depressed by Winter Storm Uri’s –$236M Texas hit) → $2,994M (FY22) → $4,101M (FY23) → $5,539M (FY24) → $5,838M (FY25). That ~tripling decomposes into three drivers: (1) the post-Uri cyclical recovery of ERCOT/PJM power prices off a 2021 trough; (2) the Energy Harbor acquisition (closed March-2024), which added ~$2B+ of East-segment EBITDA on a full-year basis; and (3) genuine operating execution. Very little is organic volume growth — generation volume is roughly flat (the fleet is ~60% utilized, not capacity-constrained). This is high-quality in the sense that it converts to cash and funded buybacks, but it is not a secular volume compounder; it is a cyclical-price + acquisition story. (FACT — 10-K segment tables.)

FY2025 was flattered and partly won’t repeat. Retail Adjusted EBITDA of $1,622M (a record) leaned on non-repeating Energy Harbor integration/supply gains; management guides retail down to a ~$1.4B run-rate. Generation benefited from still-elevated power prices. The honest read: FY2025 is near a cyclical-and-integration high for the legacy business, with the next leg of growth dependent on Cogentrix, the PPA ramp, and demand absorbing the ERCOT supply flood.

First-half 2026 is a better bridge than FY2025 alone. Ongoing Operations Adjusted EBITDA increased to $3.261B from $2.589B: Texas rose to $897M from $632M and East to $1.443B from $932M, while Retail declined to $841M from $940M. East benefited from higher PJM capacity pricing and the acquired Lotus assets; Texas benefited from gas optimization and Martin Lake’s return. This is strong delivery, but much of the growth is acquired, capacity-price driven or operational recovery rather than customer-volume compounding. The portfolio was 100%/94%/72% hedged for 2026/2027/2028 as of August 3, so reported growth is insulated from—not evidence against—the current weak Texas curve.

Forward opportunities, ranked by quality:

1. Contracted nuclear PPAs (highest quality) — the ~3.8 GW of AWS/Meta deals ramping 2027–2032, projected at ~25% FCFbG accretion at full ramp. This is the genuine de-risking growth.
2. Cogentrix integration (~5.5 GW gas, closes 2H-2026) — adds EBITDA (excluded from current guidance), bought at ~$730/kW (cheap vs new-build), but at a cyclically elevated moment for gas-asset values.
3. Nuclear uprates (~633 MW, 2031–2034) and Permian gas — incremental firm capacity into a tight firm-power market.
4. ERCOT/PJM demand absorption — if the ~5–6%/year load growth materializes and supply does not overwhelm it, merchant margins hold; this is the swing factor and the lowest-certainty lever.
5. Vistra Zero solar/battery — now impaired by OBBBA credit phase-outs and the fact that ERCOT batteries have “returned virtually nothing”; management is deferring/abandoning capex here. A diminished opportunity.
6. Helix digital infrastructure — a preferred, nonexclusive origination channel paired with up to $1B of equity commitments. Potentially high quality if it creates contracted power at mid-teens returns; currently unquantified and capital-consuming.

The 2027 opportunity now needs a two-step bridge. Management retained $7.4–$7.8B for legacy operations but says lower ERCOT forwards bias that business toward the low end. Cogentrix and the Meta contract are excluded and together could contribute roughly $700M around the midpoint, before financing, integration and timing effects. Thus the bull cannot simply add transaction EBITDA to the old midpoint: it must first subtract the curve deterioration that management says PJM, hedging and the PTC do not fully offset.

Verdict (growth): mixed quality — cyclical-price + acquisition-driven historically, with the highest-quality forward growth (contracted nuclear PPAs) genuine but modest, and the lowest-quality (merchant margin in oversupplied ERCOT) the largest swing. Do not extrapolate the FY21→FY25 EBITDA tripling; much of it was a price recovery off a trough plus a one-time acquisition step-up. The per-share growth engine has been the denominator (buybacks), not organic EBITDA volume.

6. Financial Quality

Rule one: discard GAAP earnings. VST’s GAAP net income is dominated by non-cash, unrealized mark-to-market on its commodity hedge book and is genuinely uninformative for valuation. GAAP net income: –$1,274M (FY21), –$1,227M (FY22), +$1,493M (FY23), +$2,659M (FY24), +$944M (FY25); GAAP diluted EPS gyrated $3.58 → $7.00 → $2.18 (FY23–25). FY25 net income fell $1,868M year-over-year while Adjusted EBITDA rose $299M, driven by a –$1,963M unfavorable swing in unrealized hedge MTM with zero cash impact. The trailing P/E of ~24.7x is meaningless; even the “13.5x forward P/E” is a Street MTM-stripped construct, usable only directionally. (FACT — 10-K FY2025 MD&A.)

Adjusted EBITDA — the metric that matters. Roughly tripled FY21→FY25 . Q2 2026 Ongoing Operations Adjusted EBITDA was $1.767B versus $1.349B, and the first half was $3.261B versus $2.589B. Management reaffirmed 2026 Adjusted EBITDA of $6.8–$7.6B (midpoint $7.2B) and FCFbG of $3.925–$4.725B (midpoint $4.325B), both excluding Cogentrix. The $7.4–$7.8B 2027 legacy opportunity stayed on paper, but the current curve biases it toward the low end. The “Ongoing Operations” framing excludes Asset Closure; this is economically defensible for retiring assets but makes reconciliation to consolidated GAAP and simple cash flow essential.

Free Cash Flow before Growth (FCFbG) — management’s headline cash metric, useful only with a capex bridge. FCFbG grew from $2,491M (FY23) to ~$2,888M (FY24) to ~$3.6B (FY25), and 2026 is guided to $3.925–$4.725B. At the $4.325B midpoint on 335.6M shares, that is ~$12.89/share and an 8.8% yield on the August 13 equity value. But simple GAAP CFO less total capex was –$1.239B, –$816M, +$3.777B, +$2.485B and +$1.318B in FY2021–25; first-half 2026 CFO of $2.222B less $1.572B capex was $650M. The gap is not an accounting error: FCFbG deliberately excludes growth capex and normalizes specified working-capital/other items. For valuation, it is distributable capacity before the growth program, not cash automatically available to common holders.

Cash flow is real but lumpy — the collateral swing. Underlying GAAP CFO: –$206M (FY21), +$485M (FY22), $5,453M (FY23), $4,563M (FY24), $4,070M (FY25). The 2021–2022 near-zero/negative CFO reflects massive collateral posting during the energy-price spike (margin deposits on hedges) — a structural feature of a large hedge book that can consume cash violently when prices move against posted positions. FY25 CFO fell year-over-year partly on a +$727M working-capital use from margin deposits. Conversion of Adjusted EBITDA to FCFbG runs ~60%. (FACT — 10-K cash-flow statements.)

Balance sheet and leverage — investment grade, but prefunded transaction debt is already present. June 30 debt including current maturities was $19.595B versus $17.043B at year-end; cash fell to $435M, leaving $19.160B of funded net debt before the $300M accounts-receivable financing. Preferred stock carried about $2.476B of additional senior claims. January’s $2.25B notes directly prefunded Cogentrix, while April’s $4B issuance largely refinanced 2027 maturities and Term Loan B-3. Revolver commitments increased from $3.44B to $5.50B in June and available credit was substantial, reducing close/refinancing risk without changing the economics of added leverage. The company has two investment-grade ratings and targets all three in the mid-investment-grade area; that is a genuine funding advantage, not permission to ignore transaction leverage.

Capex — modest maintenance, rising growth. Total capex rose from $1,033M (FY21) to $2,752M (FY25). Maintenance is a modest ~$1.0–1.3B; the ramp is growth (solar/battery, Permian gas, the 433 MW of nuclear uprates tied to the Meta PPAs). Because FCFbG excludes this, the gap between FCFbG and true free cash flow widens as the growth program builds.

The share-count engine — historically exceptional, marginally ordinary. Shares outstanding were 335.961M on June 30 and 335.635M on August 3, versus 469.1M at FY2021. Since November 2021, management says it retired 171M shares at an average near $38, about 30% of the count. In 2026 through August 3, however, it repurchased 4.966M shares for $778M at $156.57; Q2’s average was $153.34. Authorization remaining was $1.222B. The pending Cogentrix issuance of 5M shares is larger than 2026 gross retirement through August 3. The early program was plainly accretive; today’s decision must clear the same opportunity-cost test as debt reduction, Cogentrix and Helix.

Quality-of-earnings flags. (1) Q2 GAAP net income was $305M despite $1.767B of Ongoing Operations Adjusted EBITDA, including a $472M unrealized hedge loss expected to settle in future years. (2) GAAP–CFO–FCFbG divergence and collateral/margin swings can be large. (3) Asset Closure is excluded from Ongoing Operations; decommissioning trusts and ARO estimates add noise. (4) Moss Landing remains live rather than finished: the cleanup estimate increased $65M in Q2 to $175M, the remaining ~350 MW restart is uncertain “when or if,” and carrying value was $311M after collecting the full $500M of insurance. Further impairment is possible even though the cash insurance recovery is complete.

ROIC/ROE — do not headline. ROE is volatile and inflated by the thin, levered equity base . GAAP operating income (FY25 $1,906M) is itself depressed by the hedge MTM swing, impairments, and Moss Landing write-offs. Use EV/Adjusted EBITDA and FCFbG/EV, not ROE or ROIC, for this business.

Verdict (financial quality): a genuine cash machine, not a capital sink — economics do improve with scale (the integrated hedge, repeatable acquisition integration, fixed-cost operating leverage) — with three honest caveats. (1) GAAP earnings are pure MTM noise; value only on Adjusted EBITDA/FCFbG. (2) FCFbG excludes a large, back-end-loaded growth-capex program, so post-growth FCF is lower than the headline. (3) FY2025 leans on non-repeating Energy Harbor/retail gains and still-elevated power prices, and the cash is commodity-/weather-/collateral-exposed and lumpy. The IG upgrade and the sub-$36 buyback are the two best financial facts in the file.

7. Capital Allocation

The per-share engine is real and well-executed; its marginal return is lower. VST runs a clear waterfall: protect investment grade, grow a modest dividend, repurchase stock, then fund growth and M&A. Management reports 171M shares retired since November 2021 at an average near $38. Through August 3, 2026 it spent another $778M at $156.57 per share and had $1.222B authorized. The historic program is the single most value-creative act in VST’s history. The latest dollars are being deployed at about 9.9× guided midpoint EBITDA, not at a distressed post-bankruptcy valuation, so the proper comparison is the after-risk return on repurchases versus deleveraging, Helix and incremental generation—not the historical average cost.

M&A — coherent history, but correct the Cogentrix price before judging it. Dynegy, Ambit, Energy Harbor and Lotus built today’s portfolio. Cogentrix remains pending for late 2026 after receiving FERC approval. Filed consideration is approximately $2.3B cash net of estimated assumed debt, 5M VST shares stipulated at $185 ($925M), and approximately $1.5B of assumed debt—about $4.725B of unadjusted economic consideration before tax benefits and final closing adjustments. The June baseline’s ~$1.9B shorthand understated the package. At the disclosed 7.25× 2027 expected Adjusted EBITDA multiple net of tax benefits, it may still be accretive; because standalone EBITDA is not disclosed, a conservative $0.4–$0.6B contribution implies roughly 11.8×–7.9× on the unadjusted package. The transaction is not proven cheap until closing leverage, segment EBITDA and realized FCFbG/share are visible.

Helix adds another explicit claim on the waterfall. The initial $500M and conditional/elective additional $500M commitment compete with the remaining buyback, Cogentrix deleveraging and organic projects. Management says projects must clear its mid-teens threshold. That hurdle is sensible, but compensation still lacks an explicit ROIC/ROCE gate, and Helix has not disclosed a contracted pipeline. The burden of proof rises when capital is committed before exclusivity or unit economics exist.

Growth capex. Vistra Zero (solar + battery, now impaired by OBBBA and poor ERCOT battery economics), Permian gas, and ~433 MW of nuclear uprates tied to the Comanche Peak/Meta PPAs, plus 1,200 MW of supply to an IG customer (AWS) from Q4-2027. Management cites a mid-teens levered return hurdle. A discipline flag: the 2025 annual-incentive “Total Cost” metric ran unfavorable to threshold, “primarily driven by higher spend for opportunistic growth projects” — i.e., the company is spending into the theme.

Incentive alignment — unusually good on a per-share basis, with one real gap. The DEF 14A (2026-03-18) shows: Annual Incentive weighted Adjusted EBITDA 25%, Adjusted FCFbG 25%, Total Cost 10%, Generation Operations 20%, Retail Operations 10%, Stakeholder 10% (funded 130% for 2025, with an executive limiter capping at 50% if Adjusted EBITDA misses threshold). Long-Term Incentive is 65% PSU / 35% RSU, with PSUs tied 100% to Adjusted FCFbG per share plus a relative-TSR modifier (±25% vs the S&P 500) and an absolute-TSR floor (capped at 100% if absolute TSR is negative), 200% cap. FCFbG-per-share is genuinely owner-aligned — it directly rewards the buyback and penalizes dilution. The 2023 PSUs paid the 200% maximum (FCFbG/share $24.85 vs a $17.17 target; relative TSR ~629%, ~100th percentile) — calibration is arguably soft (both the annual plan and the PSUs maxed), but the 629% TSR makes it defensible. Say-on-pay support was >97% (2025) and >98% (2024). CEO Jim Burke; independent Chairman Scott B. Helm; CFO Kris Moldovan; ~91% of CEO pay is at-risk. The real gap: there is no ROIC/ROCE gate anywhere in the incentive structure — the acquire-and-build flywheel is not explicitly return-tested in pay, and the FCFbG/share targets are undisclosed (we cannot audit their rigor). Insider ownership is thin (the ~15-person Section-16 group holds ~3.1M of 338.5M shares, <1%; Burke ~1.51M). Top holders: Vanguard 12.6%, BlackRock 8.4%, Qatar Investment Authority 5.5%, FMR 5.1%. (FACT — DEF 14A 2026-03-18.)

Insider transactions — still sell-skewed, with no post-baseline buying. The refreshed five-year SEC mirror passed 375/375 integrity. Since the prior report, insiders sold 46,988 shares for $7.691M, all disclosed under 10b5-1 arrangements, and recorded no open-market purchase. Over the last 24 months, sales totaled about 1.997M shares/$352.1M; at least 1.009M/$195.0M were explicitly planned. No officer has filed code P since March 2023. Planned sales do not prove a negative view, but the corpus offers no insider-conviction offset to the drawdown.

Verdict (capital allocation): competent-to-strong, entering the phase where discipline gets tested. A genuinely excellent historical allocator — the sub-$36 buyback alone made this stock for early holders — with an unusually well-aligned, per-share-based incentive plan. The asterisks: no return gate in pay, soft (maxed-out) compensation calibration, a pivot to plowing rising debt + dilutive equity into gas-asset M&A at a euphoric moment, and insiders selling (10b5-1) while none buy the dip. Strong track record; the next 18 months (Cogentrix integration, capital-cycle position) are the test.

8. Changes and Headwinds — Last Two Years

Portfolio transformation. Energy Harbor closed in March 2024 and created the nuclear platform now under Meta contracts; Lotus added ~2.6 GW of gas/battery assets in 2025; Cogentrix would add ~5.5 GW but remains pending. The ~3.8 GW of 20-year AWS/Meta nuclear contracts converted part of the best asset base from merchant to contracted margin. These changes improve earnings visibility while increasing transaction, financing and integration complexity.

Post-baseline operating evidence. Q2/H1 Ongoing Operations Adjusted EBITDA grew 31%/26%, fleet commercial availability exceeded 97% during July peaks, and the $6.8–$7.6B 2026 range held. Retail volumes fell 4.4%/7.0% and first-half Retail EBITDA fell 11%. The result confirmed execution and the portfolio hedge without establishing customer captivity.

Curve and capacity evidence. ERCOT North and West average Q2 prices weakened, and management explicitly described current forwards as meaningfully lower; legacy 2027 trends toward the low end. PJM’s 2028/29 auction, conversely, produced a 6.8 GW capacity shortfall and a $325 cap across all zones. The portfolio now spans two capital-cycle states: Texas is absorbing an aggressive supply response, while PJM cannot bring enough incremental firm capacity despite high administered prices.

Helix and large-load regulation. Helix changes an undefined data-center relationship into up to $1B of actual equity exposure and a preferred, nonexclusive power channel. FERC’s June 18 action is directionally constructive for large-load integration but leaves tariff and cost-allocation work. Texas’ August 3 audit and June cost-shield directive improve screening but can delay projects. Management’s 12–15 GW credible 2030 Texas data-center estimate is the relevant underwriting anchor, not the 400+ GW gross queue.

Moss Landing. The battery loss is not finished. Cleanup is estimated at $175M, the remaining ~350 MW unit’s restart is uncertain, and $311M of carrying value remains after the full $500M insurance collection. A final retirement would remove future optionality and could require impairment.

Financing and capital allocation. June commitments expanded the revolver to $5.50B, and a July amendment raised TXU’s receivables securitization to $1.25B and extended it to July 2027. Those changes strengthen liquidity. They also sit beside $19.595B of funded debt, preferred claims, transaction prefunding and a pending 5M-share issuance. Repurchases remain active but occur near $150 rather than the program’s roughly $38 historic average.

Price and positioning. The stock is down only 1.1% from the prior report but traveled from a $171.35 post-baseline high to $134.75 low. FactorsToday shows positive Momentum alongside negative Quality and Value loadings, strong Utilities/Market/Beta exposure and 32.46% specific volatility. The signal is thematic/systematic sensitivity, not independent evidence of business quality.

Verdict (changes): execution improved; structural slack narrowed. Strong reported earnings, the held annual range and PJM shortage are genuine positives. Persistent Texas curve weakness, the low-end 2027 bias, increased capital commitments and Moss Landing uncertainty are genuine negatives. The baseline thesis is not falsified on either side: the bull lacks curve stabilization/new premium contracts, and the bear lacks a formal guidance cut or breakdown in cash generation.

9. Risk Analysis (Risk Matrix)

Catastrophic-loss / total-loss risk: low. VST is investment grade, cash-generative, diversified across fuels/geographies and supported by a retail hedge. A total loss would require a simultaneous structural collapse of power prices, major operating losses and a refinancing crisis. The realistic downside mechanism is EBITDA softening plus multiple compression, not disappearance of the enterprise.

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation. Embedded-expectations and scenario analysis only.

Use the right lens. GAAP P/E is distorted by hedge mark-to-market and P/B by a post-bankruptcy, repurchase-shrunken equity base. EV/Adjusted EBITDA captures the funded debt, accounts-receivable financing, preferred stock and minority claims that common equity must support. Price/FCFbG measures the cash-return engine, but FCFbG excludes growth capex. Both lenses are needed; neither is a complete answer alone.

Current bridge (August 13 close, ex-Cogentrix). At $146.40 and 335.635M shares, common equity is about $49.1B. Add $19.595B of funded debt, $300M of accounts-receivable financing, $2.476B of preferred stock and $12M of noncontrolling interest; subtract $435M cash. Inclusive enterprise value is approximately $71.1B. Excluding the receivables financing would reduce EV by only $0.3B and does not change the conclusion.

The midpoint is fair for a diversified, investment-grade merchant with contracted nuclear exposure; it is not a trough valuation for a no-moat cyclical. The FCFbG yield looks more attractive than the EBITDA multiple, but first-half simple CFO less total capex was only $650M because growth spending sits outside management’s preferred cash measure.

Cogentrix bridge—do not treat the deal as free EBITDA. The unadjusted economic package is approximately $4.725B: $2.3B cash, 5M shares valued at the stipulated $185, and about $1.5B assumed debt. Using the current share price for the new equity and adding cash/assumed-debt funding produces a conservative day-one pro-forma EV of about $75.6B and roughly 340.6M shares. Because target EBITDA is not disclosed, $0.4–$0.6B is a sensitivity, not a forecast. Against legacy 2027 EBITDA of $7.4–$7.8B, combined EV/EBITDA spans roughly 9.2×–9.7×. On the full unadjusted transaction package, the implied acquisition multiple is 11.8× at $0.4B, 9.5× at $0.5B and 7.9× at $0.6B. Tax benefits, synergies and FCF conversion may improve the result, but realized accretion is not yet observable.

Own-history context. AZI’s August 13 index places VST at a 75.2 composite percentile, with P/E at the 50th, P/B at the 84th and P/S at the 91st percentile. That is down only modestly from the June composite near 77. The history array was unavailable, and the inputs are imperfect: P/E carries derivative noise, P/B is elevated by buybacks, and commodity settlement distorts sales. The index therefore answers only one narrow question: even after the drawdown, the equity does not screen as abandoned relative to its own accounting history.

Peer multiples deliberately omitted. Current aggregator outputs mix trailing GAAP-derived EBITDA, different price dates and transaction-distorted balance sheets for CEG, TLN and NRG. A precise-looking table would be less comparable than the company-guidance bridge above. The qualitative order remains useful: CEG deserves more for a much larger nuclear/contracted mix; NRG has less nuclear scarcity; TLN is more concentrated. That is industry structure, not a substitute for synchronized pro-forma modeling.

Scenario analysis (FY2027–28 normalized; illustrative arithmetic, not a target). Equity value equals normalized EBITDA times EV/EBITDA, less funded net debt and preferred/minority claims, divided by diluted shares. The balance-sheet and share assumptions are explicit because Cogentrix adds debt and 5M shares while subsequent FCF and buybacks may reverse part of that increase.

The current price lies inside the base range rather than below it. The bear range is lower than the June analysis because the corrected Cogentrix package and explicit close-date debt/share bridge add claims. The bull range is also lower because the latest ERCOT curve and capped PJM economics make an 11× ceiling more defensible than an automatic CEG-like re-rating.

Embedded expectations. At $71.1B of EV, a 9.5× multiple capitalizes about $7.5B of EBITDA and a 10× multiple about $7.1B—essentially the existing 2026 guide. The market grants limited value to the 2027 contract/deal ramp only if one believes a roughly 9–10× merchant multiple persists; it simultaneously assumes the current guided cash engine survives the Texas curve. This is why “growth not priced” is too simple: deal consideration and leverage are already visible even though target earnings are not.

What may be mispriced. A constructive variant is that the market applies a merchant multiple to an expanding contracted/floored nuclear and retail slug, under-crediting future deleveraging and per-share compounding. A cautious variant is that the 8.8% FCFbG yield looks distributable when a $4.5–$5B growth/acquisition program, Helix commitments and Cogentrix leverage absorb much of it. Each 1.0× move on $8B of EBITDA is about $8B of EV, or roughly $24 per share at a 333M count. Multiple durability remains the largest valuation lever; ERCOT forwards determine whether the earnings base can support it.

11. Variant Perception

Consensus framing. VST is treated as an AI-power/data-center beneficiary with a best-in-class buyback, investment-grade funding and ~3.8 GW of marquee nuclear PPAs. The tape complicates that narrative: constructive company news produced only a –1.1% net move since June after a $171.35-to-$134.75 round trip. FactorsToday loads VST positively on Momentum and Utilities but negatively on Quality and Value, with about 65% of variance explained in the broad model and 32.46% specific volatility. This is a crowded thematic exposure with meaningful company risk, not a stable regulated-utility proxy.

The strongest bull case. AI/electrification drives durable tightening; VST’s ~60%-utilized fleet captures load at high incremental margins; the ~3.8 GW of nuclear PPAs convert merchant exposure to annuities; Cogentrix and Lotus add scarce firm capacity; Helix originates contracted projects; the buyback resumes after deleveraging; and the market separates contracted/floored earnings from the merchant slug.

The strongest bear case. Merchant power remains a no-moat commodity; battery/solar/TEF-gas supply structurally resets Texas forwards lower; PJM shortages remain capped; retail volumes erode; Cogentrix and Helix absorb capital before returns; Moss Landing loses more carrying value; and the thematic merchant multiple mean-reverts. EBITDA and the multiple then compress together.

The 3–5 assumptions that matter most. (1) The durability of the merchant multiple — permanent re-rating vs. mean reversion (the single biggest swing). (2) ERCOT forward power prices — does demand absorb the supply flood, or does oversupply re-base margins lower? (3) The PJM capacity windfall — held vs. clawed back. (4) Cogentrix/Lotus M&A returns — accretive vs. cyclical-top overpayment. (5) The nuclear-PPA ramp — on-time contracted annuity conversion vs. delay.

What would falsify each side. The bull is dented, not broken: ERCOT softness now biases 2027 low, but 2026 guidance held. It fails if that softness causes formal cuts, contracted projects stall, or Cogentrix fails the per-share accretion test. The bear is not falsified: Texas has not firmed and no new premium PPA appeared. It fails if screened demand absorbs supply, additional contracts disclose attractive economics, and pro-forma leverage falls while FCFbG/share grows.

12. Fact vs. Interpretation Table

13. Open Questions

1. Pro-forma leverage and accretion after Cogentrix closes — what EBITDA, FCFbG/share and closing net debt emerge from the corrected ~$4.725B unadjusted package?
2. The true total growth-capex magnitude and returns — maintenance vs. growth is not cleanly disclosed; what is the all-in post-growth FCF, and do the mid-teens levered returns hold?
3. Durable retail run-rate — is ~$1.4B the right normalized number, or does easy-switch ERCOT competition erode it further?
4. ERCOT forward-curve trajectory 2028+ — does demand absorb the battery/TEF-gas/solar supply, or does oversupply structurally re-base margins? The single most important fundamental unknown.
5. PJM policy after the capped shortfall — how do the collar, future auctions and FERC/transmission-owner compliance change East EBITDA and large-load contracts?
6. 45U “gross receipts” definition — pending Treasury guidance could alter the nuclear floor’s effective level.
7. Additional nuclear/gas PPAs — will the ~3.2 GW of uncontracted nuclear (Beaver Valley, Comanche Peak) sign at disclosed premiums, validating the annuity-conversion thesis?
8. Capital-allocation discipline late-cycle — with no ROIC gate in pay and a gas-M&A pivot, does the historically excellent allocator avoid overpaying into the theme?
9. Helix unit economics — which projects are actually contracted, on what terms, and what must VST fund before receiving power-supply economics?
10. Moss Landing outcome — will the remaining plant restart, and how much of the $311M carrying value survives?

14. What Must Be True (Bull and Bear, with Falsification Tests)

For the BULL case (re-rating durable):

• ERCOT and PJM tighten durably as AI/electrification demand outpaces the arriving supply, holding/firming forward power prices through 2028+.
• VST signs additional nuclear/gas data-center PPAs at disclosed premiums, converting more merchant output to 20-year annuity.
• Cogentrix and Lotus prove clearly accretive; the merchant multiple holds or re-rates toward CEG-lite.
• The buyback continues compounding per-share value on a shrinking count.
• Falsification test: two or more consecutive quarters of ERCOT forward-curve softening flowing into Adjusted-EBITDA guidance cuts, OR a PJM capacity clawback, OR the multiple compressing below ~8.5x EV/EBITDA on flat EBITDA — any of these breaks the bull.

For the BEAR case (de-rate + EBITDA softening):

• The ERCOT supply flood (batteries doubling, TEF gas, solar) structurally re-bases Texas forwards lower in an energy-only, capped market, compressing the ~46% Texas slug.
• PJM capacity is clawed back; the AI-era multiple mean-reverts to pre-2023 high-single digits.
• FY25’s flattering retail/Energy-Harbor gains don’t repeat; the gas-M&A pivot disappoints on returns.
• Falsification test: ERCOT forwards stabilizing or firming as demand outpaces supply, AND additional nuclear/gas PPAs signed at premiums, AND Cogentrix proving accretive — any sustained combination of these breaks the bear and validates the held multiple.

Section 15 (Source Appendix) is maintained as a separate deliverable (Appendix B in the combined report).

APPENDIX A — Standard Diligence Questionnaire

Vistra Corp. (NYSE: VST) — as of 2026-08-14

Supplemental to the research memo; Fact/Interpretation/Assumption labels are applied where material.

General

What thoughtful questions have other investors asked about this company?

• Is the AI/data-center power-demand re-rating permanent or a cyclical/thematic bubble that mean-reverts? (The single most-debated question.)
• How much of VST’s earnings power is genuinely contracted/floored (nuclear PPAs + 45U + retail) versus merchant/commodity? (Roughly 35–47% vs 53–65% of EV.)
• Is ERCOT’s energy-only market — with no capacity floor and a fast-arriving battery/gas supply response — a structural problem for VST’s largest (~46%) geography?
• Is VST overpaying for gas assets (Cogentrix, Lotus) late in the capital cycle?
• Why is GAAP EPS so volatile, and which metric should one actually use? (Adj EBITDA / FCFbG, not GAAP.)
• How durable is the retail book in a market designed for easy switching?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Closer to a cyclical high, with regional divergence (Interpretation). PJM capacity is scarce but capped; ERCOT forwards are “meaningfully lower.” The own-history composite is 75.2 (P/S 91st percentile). Hedging is approximately 100%/94%/72% for 2026/27/28, deferring rather than eliminating curve risk.

Driven by the external environment or internal actions? Both. External: post-Uri power-price recovery, the AI-demand window, PJM capacity prices. Internal: the Energy Harbor acquisition (added ~$2B+ EBITDA), operating execution, and the ~28% share-count reduction that drove per-share value.

How stable are revenues? Reported revenue is volatile, gross-settled and pass-through-laden. Adjusted EBITDA is smoother through the integrated hedge: Q2/H1 2026 grew 31%/26% even as H1 Retail EBITDA fell 11%. Underlying merchant margins remain commodity-, weather- and curve-dependent.

Outlook for products/services? Management underwrites 4–6% ERCOT and 2–3% PJM load growth through 2030. ERCOT set a 91,089 MW record, but Texas is auditing a 400+ GW large-load queue; management recognizes only 12–15 GW of credible 2030 data-center load. Supply, timing and price—not gross demand headlines—drive profitability.

How big will this market be — growing, shrinking, domestic or international? Domestic (US: ERCOT, PJM, ISO-NE, NYISO, MISO, CAISO). The end market (US power demand) is growing for the first time in ~two decades, driven by data centers, electrification, and reshoring. But it is a commodity market where price, not volume, determines profitability.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More, on the supply side — TEF-subsidized new gas, merchant batteries doubling to ~16 GW, and surging solar are all entering ERCOT, the Marathon capital-cycle signature. Merchant power has zero entry barriers.

How profitable is the business (ROIC, ROE)? Headline ROE is not reliable because buybacks shrink equity and derivatives move income. ROIC.ai’s Q2 ROIC/return-on-capital were 11.64%/12.36%, also cycle/accounting affected. Current EV/2026 midpoint EBITDA is ~9.9× and midpoint FCFbG yield 8.8%; these are the useful lenses.

How profitable is the industry — competitors, barriers to entry? Structurally a bad industry (Greenwald): a commodity with no barriers and mean-reverting returns, currently enjoying a demand windfall. Competitors: CEG, Talen, NRG, Calpine (now CEG), regulated utilities. No durable industry profit pool absent the current demand cycle.

Can the business be easily understood? Moderately. The asset base and retail book are simple; the earnings are obscured by commodity-hedge MTM (GAAP is noise) and require translating to Adj EBITDA/FCFbG.

Can it be undermined by foreign low-cost labor? No — power is generated and consumed locally; this is not a labor-arbitrage-exposed business.

Do brands matter? Modestly. TXU is a ~20-year brand with ~2.6M Texas customers, providing weak demand captivity (habit/inertia) — but in a market engineered for easy switching, it is not a strong moat.

What is the nature of competition? On generation: price competition on a commodity (marginal-cost dispatch). On retail: price + brand + service in a low-switching-cost market.

Customers’ switching costs? Low (generation: none — the grid buys the cheapest MWh; retail: deliberately low under ERCOT Power-to-Choose).

Moat verdict: No durable franchise moat. The only financializable edges are the narrow nuclear scarcity/45U floor (~15% of capacity) and a demonstrated capital-allocation/acquire-integrate capability (not a barrier). “Best operator in a bad industry.”

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The ~3.8 GW of 20-year nuclear PPAs (AWS/Meta) are contracted annuity value not capitalized as such; the irreplaceable nuclear licenses; the TXU brand. Conversely, the equity base understates economic value due to the bankruptcy-fresh-start accounting and ~$6.9B treasury stock.

Off-balance-sheet liabilities? Operating leases, fuel/transport contracts, and — material to this business — large commodity-hedge collateral obligations that can consume cash in price spikes (CFO turned negative in 2021–22). Nuclear decommissioning and asset-retirement obligations are on-balance-sheet but rest on critical estimates.

How conservative is the accounting? Mixed. Management’s non-GAAP presentation (Adj EBITDA, FCFbG, “Ongoing Operations”) is reasonable and the right lens, but “Ongoing Operations” flatters the total by ~$75–150M/year by stripping Asset Closure. GAAP is dominated by MTM noise. No evidence of aggressive revenue recognition.

How CapEx-hungry is the business? Maintenance capex is modest (~$1.0–1.3B), but the growth program (solar/battery, Permian gas, nuclear uprates) is large and back-end-loaded (majority post-2028). FCFbG excludes growth capex, so post-growth free cash flow is materially lower than the headline.

Capital Allocation & Management

How much FCF does the business generate, and how is it used? FCFbG was ~$3.6B in FY2025 and is guided to $3.925–$4.725B in FY2026. It funds leverage, dividend, buybacks, acquisitions and growth. First-half CFO less total capex was $650M, illustrating that FCFbG excludes material growth spending. Vistra has retired 171M shares at an average near $38, but 2026 repurchases averaged $156.57 through August 3.

Significant acquisitions recently? Energy Harbor (nuclear, March 2024), Lotus (gas, 2025), and pending Cogentrix (~5.5 GW, expected late 2026). Cogentrix’s unadjusted package is about $4.725B including cash, stipulated shares and assumed debt—not the prior $1.9B shorthand. Accretion remains to be proven after close.

Buying back shares? Yes—about 30% of the count retired at an average near $38. Marginal economics are less compelling: Q2 repurchases averaged $153.34, and the pending 5M Cogentrix issuance exceeds 2026 gross retirement through August 3.

Issuing large amounts of new shares to insiders? No large dilutive insider issuance; the ~5M Cogentrix-deal shares modestly offset the buyback. SBC is modest.

Compensation policy? Well-aligned on a per-share basis: annual incentive weights Adj EBITDA 25% / Adj FCFbG 25% / operations / stakeholder; LTI is 65% PSU tied 100% to Adjusted FCFbG per share + relative-TSR modifier + absolute-TSR floor. Say-on-pay >97%. Gap: no ROIC gate; targets undisclosed; 2023 PSUs paid the 200% max (soft calibration, though a 629% TSR makes it defensible).

Motivations of management? Per-share value creation is supported by the buyback and FCFbG/share PSU. But insider ownership is thin; post-baseline sales were 46,988 shares/$7.691M, all 10b5-1, with no purchase. Planned selling is not a negative thesis by itself, but no conviction-buy signal exists.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No — a standard US C-corp common share (NYSE: VST); 1099, not K-1.

Dividend policy? Modest and growing: ~$0.91/year (~$0.228/quarter), ~15% payout of FCFbG, ~0.62% yield. The buyback, not the dividend, is the return lever. An 8.0% Series A perpetual preferred sits atop common.

How profitable is the business? On a cash basis, genuinely profitable: ~$5.84B Adj EBITDA (FY25), ~$3.6B FCFbG, ~60% conversion. On a GAAP basis, volatile and uninformative (MTM noise).

Is net income diverging from cash from operations? Yes, structurally — GAAP net income is MTM-driven and diverges sharply from CFO/FCFbG. This is the defining QoE feature: value on cash, not GAAP earnings.

Risks & Downside

What factors would cause the stock to decline? ERCOT forward-price re-base lower (supply flood); multiple mean-reversion (the biggest swing); PJM capacity clawback; natural-gas collapse (2028+); cyclical-high M&A disappointing; retail normalization; a data-center-demand shortfall; an operational event (another Moss-Landing-type fire).

Risk of a catastrophic loss? Low. Investment-grade access, diversified fuels/geographies and retail dampen risk. The realistic mechanism is EBITDA softening plus multiple compression, not enterprise impairment.

Chance of a total loss? Very low — would require simultaneous structural power-price collapse and a refinancing crisis, remote given the IG balance sheet and cash generation.

Recent News & Events

Has the business environment changed recently? Yes: (1) Q2/H1 execution was strong and 2026 guidance held; (2) ERCOT forwards became meaningfully lower while PJM’s latest auction was 6.8 GW short at the $325 cap; (3) Texas began auditing large-load requests; (4) Helix created up to $1B of equity exposure; and (5) Moss Landing restart/impairment risk increased.

Significant acquisitions? Cogentrix (~5.5 GW, FERC-approved but pending late 2026), Lotus (2025), and Energy Harbor (March 2024). Approximately 3.8 GW of AWS/Meta nuclear contracts improve the quality of the acquired/legacy nuclear book.

Change in accounting policies? No material change; management began emphasizing a simplified two-bucket “Ongoing Operations” (Generation + Retail) presentation.

Recent changes — new markets, facilities, management? Helix adds a digital-infrastructure investment channel and a preferred, nonexclusive power-provider role. Moss Landing’s remaining ~350 MW restart is uncertain, cleanup is estimated at $175M and carrying value is $311M. CEO Jim Burke and CFO Kris Moldovan remain in place.

APPENDIX B — Source Appendix

Vistra Corp. (NYSE: VST) — as of 2026-08-14

Primary sources are listed first. Aggregators are convenience layers and are not the source of record for material financial claims.

A. SEC Filings — Primary

B. Management Calls and Company Releases — Primary

ROIC.ai’s recent-news query returned no items for the post-baseline window. The event timeline therefore uses SEC, company, market-operator and government sources rather than interpreting an empty aggregator feed as absence of events.

C. Market Operators, Regulators and Government — Primary

D. Quantitative and Market Helpers — Secondary

The report omits a synchronized peer-multiple table because available current peer feeds mixed GAAP-derived trailing EBITDA, different price dates and transaction-distorted balance sheets. Qualitative peer structure is used; false precision is not.

E. Analytical Frameworks

Evidence rule: filings, calls and regulator/operator publications control. Aggregators are labeled, reconciled and used for context rather than unsupported facts.