VST · Analyze — Financebot

A portfolio-specific passage was removed from the public build.

What VST is

The largest US competitive/merchant independent power producer: ~44 GW fleet (nuclear + natural gas + coal + solar/storage) plus a ~5M-customer retail electricity book (TXU Energy). Three things drive the 2026 thesis: (1) irreplaceable nuclear baseload now under 20yr hyperscaler PPAs, (2) existing gas capacity worth more than replacement cost while turbine backlogs run to 2030, and (3) a buyback machine that has cut the share count ~14% since 2022.

1. Fundamentals — healthy on cash & coverage, noisy on GAAP

Metric	Value	Read
Revenue FY25	$17.74B (+3% YoY, +8.9% 3yr CAGR)	capacity-driven, not price
Adj. EBITDA TTM	~$6.6B	2026 guide $6.8–7.6B (tracking ≥midpoint); 2027 base $7.4–7.8B ex-Cogentrix/Meta
GAAP EPS TTM / FY25	$5.92 / $2.22	distorted by hedge mark-to-market — not earnings power
FY2026 consensus EPS	$8.62	the reliable current-year figure
TTM FCF (reconciled 2 ways)	$2.255B	not the $37.1M Yahoo snapshot field (M&A-quarter mis-net — pitfall confirmed, add VST to the list)
FCF trend FY23→25	$3.78B → $2.48B → $1.32B	fell 3 straight years on a capex ramp ($1.3B→$2.75B); TTM recovering to $2.26B
Adj. FCF-before-growth (2026 guide)	$3.925–4.725B	~8–9.5% FCFbG yield on market cap — the capital-return capacity
Net debt	~$19.3–19.5B	net debt/adj EBITDA ~2.9x; Fitch upgraded to Investment Grade
Adj. EBITDA/interest	~6.8x	comfortable
Shares out	335.6M	−13.9% since 2022 (real buyback tailwind); +5M pending on Cogentrix
Dividend	$0.916/yr, 0.61% yield, ~15% payout	modest, growth decelerating to ~3–4%/yr — not the thesis
Contracted	~3,809 MW nuclear 20yr PPAs (Meta 2,609 + AWS 1,200)	<9% of fleet; deliveries start Q4 2027
Hedged volume	~100% 2026 / 94% 2027 / 72% 2028	the real near-term durability number

The honest read: VST is financially healthy on the metrics that matter for a merchant generator — adjusted EBITDA/interest ~6.8x, ~100%-hedged 2026 volume, IG credit, and a genuine 14% buyback-driven share-count reduction. But two caveats are real, not cosmetic: (1) GAAP earnings and GAAP leverage swing hard on unrealized hedge marks — always use the adjusted lens and say so; (2) the capital return has been funded more by debt recycling than by free cash flow (gross issuance > repayment every year), and FCF fell three straight years into a capex ramp. TTM FCF recovering to $2.26B is the number that tells you whether 2025 was the trough — watch it.

Cogentrix correction (scout said "44GW"): Cogentrix is ~5,500 MW of gas, ~$4B (~$2.3B cash + $1.5B assumed debt + 5M shares at $185), FERC-approved Aug 2026, closing late 2026 — the "44 GW" is VST's total existing fleet. It adds PJM/ISO-NE/ERCOT breadth (diversifies away from pure ERCOT) but also steps up leverage and modestly dilutes the buyback — the central tension in the bull case. A junior-subordinated-notes offering launched 9/10, plausibly financing the close.

2. Moat — narrow (4.5/10), evergreen weak (3.5/10)

Merchant power is the hardest place in public markets to hold a moat: the product is undifferentiated electrons priced by a regional auction VST doesn't control. Running the four sources honestly, only one is structural — the existing NRC operating licenses on sited, permitted, grid-connected nuclear plants (Comanche Peak; Perry/Davis-Besse/Beaver Valley via the 2024 Energy Harbor deal), which no amount of capital replicates in under a decade and which hyperscalers now pay premium 20yr prices for. Everything else — hedged merchant gen, retail, Cogentrix gas, the Helix JV — is favorable positioning in a cyclical. The integrated retail+generation model is a volatility-reducer, not a moat (the same "long contract book is a beta-reducer, not a moat" mechanism the KB logged on Cameco — a discount-rate input, not a multiple input). ERCOT retail is the opposite of sticky: pivotal-supplier share fell from >90% (2022) to 35% (2025).

Adversarial: hard to attack directly through 2028–29 — only three OEMs make large gas turbines and GE Vernova is sold out to 2030, so no rival can undercut VST's existing fleet fast; NRC licensing lead time protects the nuclear franchise. But that protection is a closing supply-chain-and-permitting window (~2028–2032), not a permanent barrier.

Disruption vectors: renewables+storage cannibalizing gas margins (durable, decade-timeline); ERCOT 2027 price softening + the Aug-2026 Texas pause on the 474 GW data-center interconnect queue (durable, live now, being hedged via Cogentrix diversification + contracted PPAs); PJM/ERCOT capacity-market redesign (flattening — pattern-regulation-standardises-or-restricts); hyperscaler behind-the-meter/SMR bypass (distant — hyperscalers are choosing existing-nuclear PPAs over SMRs today, which is the market's own verdict on SMR readiness).

Vs. CEG: same commodity-generator DNA, but CEG converts a far larger share of its fleet into 20yr contracted cash flow — which is why CEG's ~51% forward-P/E premium has stayed sticky through the pullback. The market is discriminating correctly, not irrationally. VST's upside rests on closing that gap via Cogentrix + PPA ramp, not on having a moat CEG lacks.

3. Valuation — cheap vs its own base case, models applied by type

Graham IV: N/A — tangible book value is negative (−$7.76/sh). The formula assumes positive tangible book; it does not apply to a levered merchant generator. Set aside (as with LEU).
Bogle expected return: 0.6% yield + ~12–15% forward EPS growth (EBITDA growth + Cogentrix + buyback + deleveraging; consensus $8.62→$10.37 = +20% '26→'27) ± modest P/E change → ~13–16%/yr expected return if the growth delivers — attractive, but cycle-dependent.
P/E: true FY2026 P/E ~17.1x ($8.62); FY2027 ~14.2x ($10.37). (The scout's "14.2x, near pre-mania avg" was the FY2027 figure — pitfall-vendor-forward-eps-is-the-wrong-fiscal-year fired again; the honest current-year multiple is ~17x.) Still a deep discount to CEG (~21–24x).
EV/EBITDA (primary for an IPP): TTM ~11.1x; on 2026 guide-mid ~10.2x — toward the high end of the merchant-IPP historical ~7–11x band, justified only if the AI-power contracted growth is real.
FCF-before-growth yield: ~8–9.5% on market cap — cheap for a leveraged-but-IG generator with a growth pipeline.

Scenario	Driver	Fair value
Bear	ERCOT softens hard, Cogentrix leverage/integration drags, capacity-market redesign flattens revenue, multiple compresses to cyclical 8–9x / 12–13x P/E	$120–150
Base	Cogentrix closes & integrates, Meta/AWS ramp on schedule, 2027 guide rises, ~10x EV/EBITDA / 15–16x FY27 P/E	$175–200
Bull	AI-power demand re-accelerates, PPA book expands, VST re-rates toward CEG (11x+ / ~18–20x)	$220–260

Fair value: $155–215 (base-weighted ~$185). Sell-side mean $217 / median $221 sits at the top of this range — aggressive, because it underwrites Cogentrix + 2027 guide raises + Meta all delivering. Current $147 is below the base case — genuinely undervalued on the base, with real cyclical/execution risk pulling the low end toward $135–150.

Entry: $130–155 (in-zone now at $147; this is the Aug-low region where the CEO bought). Strong buy <$130.

Trim: ~$235, dollar-form deliberately. VST's trailing P/E (24.8x) sits on hedge-distorted GAAP EPS, so a site-computed "NNx fwd" trim would inherit the broken field and render below spot (pitfall-multiple-trim-inherits-the-broken-vendor-field). $235 ≈ 11–12x EV/EBITDA on ~$8.4B 2027E EBITDA / ~22x FY27 EPS — the level at which VST's discount to CEG has closed and the thesis (buy the cheaper, less-contracted name and ride the re-rate) is spent.

4. Synthesis & verdict

Rating: [6.8/10] — ACCUMULATE, in-zone at $147.

The setup is unusually aligned: VST is the only name in the nuclear complex reading cheap-to-fair vs its own history, sell-side is raising targets into a pullback (mean $217, +48%), Fitch just upgraded it to Investment Grade (de-risking the leverage bear case exactly as Cogentrix adds debt), BlackRock took a fresh $4.6B stake, and the CEO bought open-market stock at ~$135 into the August low — his first discretionary purchase in the dataset after a year of routine 10b5-1 selling. It fills the one gap the theme note flagged: the portfolio owns the higher-risk enrichment/SMR layers (LEU, SMR) but not the contracted-cash-flow generation layer.

Why 6.8 and not the scout's 7.5: the moat work is sobering. This is a narrow-moat (4.5) cyclical (evergreen 3.5) having a genuinely good cycle, not a compounder — the AI-power growth is real but concentrated in <9% of the fleet by contracted MW, FCF fell three straight years, capital return leans on debt, and ERCOT 2027 softening plus the Texas interconnect pause are live headwinds, not hypotheticals. The valuation and setup justify a real position; the business quality caps the conviction. It slots correctly between CEG [6.0] (VST is cheaper for a reason worth paying less for) and LEU [7.6] (which has a genuine monopoly moat VST lacks).

Position guidance: unlike the speculative nuclear names, VST is a $49B IG-rated profitable company — it belongs in a core-adjacent / AI-infrastructure sleeve, sized as a real 2–4% position, not a lottery ticket. Reasonable to initiate here; add on dips toward $130.

Thesis break triggers: 1. Cogentrix close slips materially or the financing (junior sub notes) prices at a punitive rate / forces a larger equity component. 2. 2027 adj. EBITDA guide comes in below the $7.4–7.8B base at the Q3/Q4 print (ERCOT softening winning). 3. Texas data-center interconnect pause becomes a durable curtailment of ERCOT demand growth, or PJM/ERCOT capacity-market redesign flattens the capacity revenue stream. 4. Adj. net debt/EBITDA rises above ~3.5x post-Cogentrix without a deleveraging path (loss of the IG rating). 5. Buyback pace collapses because FCF-before-growth doesn't recover — i.e., 2025 was not the FCF trough.

Upgrade conditions: 2027 guide raised on Cogentrix + Meta/AWS ramp; FCF-before-growth confirms 2025 as the trough; a further hyperscaler PPA that lifts the contracted share of the fleet meaningfully above ~9%; the Helix (NVIDIA/KKR) JV converts to a real revenue stream.

What this pass did NOT test

5–8yr FCF/revenue CAGRs — vendor history capped at ~2–4 annual periods; the multi-year trend is directional (capex-ramp-driven FCF decline, recovering TTM), not a clean CAGR.
The exact contracted-vs-merchant split as a % of EBITDA (only hedged-volume % and PPA MW obtained) — recommend an investor-day slide / 10-K risk-factor pull next pass.
The FY2025 $1.14B "Acquisitions" cash line is not Cogentrix (deal signed 12/31/25) — likely a Vistra Vision minority buy-in tranche; needs a 10-K footnote to confirm.
No bond-by-bond maturity ladder; Helix JV economics unquantified.

Sources: python .mcp/fin.py VST --news, Yahoo Finance MCP (stock info, quarterly + annual statements, recommendations, holders, insider transactions), roic.ai; Vistra Q2 2026 release + call transcript, Cogentrix 8-K (investor.vistracorp.com, Utility Dive, POWER Magazine), World Nuclear News (Meta/AWS PPAs), Reuters/Seeking Alpha/Motley Fool/Trefis coverage, analyst actions (Morgan Stanley/Scotiabank/Mizuho/JPM/BMO/UBS/TD Cowen), Fitch IG upgrade, ERCOT/PJM capacity-market coverage (Utility Dive, BNEF/POWER, Latitude Media, Ascend Analytics), Texas interconnect-pause coverage (Utility Dive, Texas Tribune), GE Vernova turbine-backlog coverage. Companion: Knowledge/Themes/nuclear-uranium.md, screen-nuclear-uranium-pipeline-2026-09-10.md.