The most-hyped AI power stock in America is losing to a passive index fund. Vistra Corp (NYSE: VST) closed at $149.30 on 4 September 2026, down 7.5% year to date, while the S&P 500 gained 12.9% over the same stretch and sits within 1.3% of its own 52-week high. Vistra sits 32% below its own peak. That 20-point performance gap is not a story about the AI power thesis failing; adjusted EBITDA grew more than 30% last quarter. It is a story about a hedge book. As of 3 August 2026 Vistra had sold forward approximately 100% of its expected 2026 generation, 94% of 2027 and 72% of 2028, according to the company’s own second-quarter earnings release. The AI-driven power price squeeze everyone is buying the stock for cannot reach the profit and loss account in any size until 2028. Our bull case is $225; our bear case is $95.
That is the piece of arithmetic almost no one running the Vistra-versus-Constellation comparison has priced. The bull argument circulating on retail channels is a multiple argument: Vistra trades cheaper than Constellation Energy, therefore it should re-rate. But a merchant generator’s multiple is a function of how much open-market upside it has left to sell, and Vistra has already sold nearly all of the next two years. BMO Capital’s James Thalacker flagged exactly this risk in a November 2025 note, warning that the company’s strong hedge position “could limit Vistra’s upside, especially given the recent rise in forward curves,” as reported by The Fly. Ten months on, that caveat has stopped being a footnote and become the whole valuation question. Here is the second thing nobody mentions: the widely quoted 2027 “midpoint opportunity” of $7.4bn to $7.8bn is explicitly derived, in Vistra’s own footnote, from market curves as of 31 October 2025. The number the AI-power bulls anchor on was struck on a forward curve that predates the entire 2026 move.
Key facts
- $149.30 – VST closing price, 4 September 2026; 52-week range $132.66-$219.82 (StockAnalysis)
- -7.5% vs +12.9% – VST versus the S&P 500, 31 December 2025 to 4 September 2026, computed from daily closes
- $1,767m – Q2 2026 Ongoing Operations Adjusted EBITDA, up 31% year on year from $1,349m (Vistra 8-K, 7 Aug 2026)
- $305m – Q2 2026 GAAP net income, down $22m year on year, hit by a $488m increase in unrealised mark-to-market losses on derivatives
- 100% / 94% / 72% – share of expected generation hedged for 2026 / 2027 / 2028 as of 3 August 2026
- $1.17m – open-market stock purchases by CEO Jim Burke across 24 August, 31 August and 1 September 2026 (SEC Form 4 filings)
- ~336m shares – outstanding as of 3 August 2026, a 30% reduction since November 2021 after $6.5bn of buybacks
What is actually happening at Vistra, and why the tape disagrees
Vistra’s second quarter was, on the operating line, close to a blowout. Ongoing Operations Adjusted EBITDA of $1,767m was up $418m on the same quarter of 2025, driven by higher realised energy and capacity prices and a full three months from the plants acquired from Lotus. The Texas segment more than doubled, from $142m to $311m. The East segment – the PJM footprint where capacity auction prices have re-priced hardest – went from $418m to $642m. The fleet ran at commercial availability of 97% or better through extreme heat in both Texas and PJM. Management reaffirmed full-year 2026 guidance of $6.8bn to $7.6bn of Ongoing Operations Adjusted EBITDA and $3.925bn to $4.725bn of adjusted free cash flow before growth.
And yet GAAP net income fell, from $327m to $305m. The bridge is the hedge book working in reverse: an increase of $488m in unrealised mark-to-market losses on derivative positions, including a $472m unrealised loss on hedges expected to settle in future years. This is the mechanical consequence of being 94% sold forward into a rising forward curve. When power prices rise, a fully hedged generator books an accounting loss on the hedges today and collects the contracted – not the spot – price later. Think of it the way a mortgage borrower thinks about a fixed rate: locking in protects you from the bad case and disqualifies you from the good one. Vistra fixed its rate through 2027 in a market that then moved in its favour.
Investors reading only the EBITDA headline have therefore been repeatedly surprised by the share price. The disconnect is not irrational. The market is discounting a company whose 2026 and 2027 cash flows are already contractually determined, and it is refusing to pay an AI-scarcity multiple for cash flows that AI scarcity cannot change. Vistra’s own chief executive framed the quarter around execution rather than upside. “The Vistra team delivered another strong quarter, building on our momentum from the start of the year and continuing to execute at a high level,” said Jim Burke, president and CEO of Vistra, in the 7 August release.
What the players are actually doing: the CEO bought the exact low
The most informative response to the sell-off did not come from an analyst. It came from three SEC Form 4 filings that almost nobody covered. Between 24 August and 1 September 2026, Vistra president and chief executive James A. Burke bought 8,665 shares on the open market – transaction code P, not an option exercise, not a plan sale – across three separate days: 2,000 shares at $135.00 on 24 August, 2,200 at $135.99 on 31 August and 4,465 at $135.25 on 1 September. Total outlay: $1,173,069 at an average of $135.38, taking his beneficial ownership to 1,146,352 shares.
Now overlay that on the tape. VST closed at $135.66 on 24 August, $137.37 on 31 August and $138.08 on 1 September. The stock then ran to $149.30 by 4 September. Burke bought within a whisker of the local low and is up roughly 10% on his average in under a fortnight. Retail commentary has circulated a figure of “over $1.5 million” for these purchases; the filings say $1.17m. The smaller, verified number is the more interesting one, because it is three separate discretionary decisions rather than a single headline trade.
The corporate response has been just as capital-allocation heavy. Vistra has executed roughly $6.5bn of buybacks since November 2021 and shrunk its share count by about 30%, to around 336 million shares. Approximately $1.2bn of authorisation remained as of 3 August – about 2.4% of the current market capitalisation – which the company expects to complete no later than year-end 2027. Total available liquidity stood at $6,295m at 30 June.
On the growth side, the two moves that matter both landed this year. Vistra announced Helix Digital Infrastructure alongside KKR, the Kuwait Investment Authority and NVIDIA, committing up to $1.0bn and taking the role of Helix’s preferred power partner. And the Federal Energy Regulatory Commission approved the pending acquisition of Cogentrix Energy, a $4.7bn deal with Quantum Capital Group that adds roughly 5,500 MW of net gas-fired capacity. Critically, neither Cogentrix nor the 20-year, 2.6 GW nuclear power purchase agreements signed with Meta are inside the 2026 guidance ranges. They are upside to a number that already excludes them – which is the strongest single argument the bulls have. Readers tracking the wider power complex will find the same dynamics in our coverage of Constellation Energy (CEG) and Cameco (CCJ).
Market impact and the data nobody is synthesising
Start with the valuation. At $149.30 and roughly 336 million shares, Vistra’s equity is worth about $50.2bn. Add the $19,595m of debt disclosed in the 30 June 10-Q, subtract $435m of cash, and enterprise value is approximately $69.3bn. Against the $7.2bn midpoint of 2026 guidance that is 9.6x EV/EBITDA. Against the $7.6bn midpoint of the 2027 opportunity – the one struck on October 2025 curves – it is 9.1x, before any Cogentrix or Meta contribution. For a business with 94% of next year’s volumes already sold, that is not an obviously cheap multiple, and it is not an obviously expensive one either. It is a multiple that says the market has stopped paying for the option and started paying for the contract.
Now the segment mix, which is the synthesis the AI-power framing misses entirely. In Q2 2026 the single largest contributor to Ongoing Operations Adjusted EBITDA was not merchant generation in Texas or PJM. It was Retail, at $773m – 43.7% of the total. And Retail is the segment with the least AI leverage and the weakest growth: it rose just 2.2% year on year in the quarter, and across the first half it fell 10.5%, from $940m to $841m. So the company that trades as a pure-play on data-centre electricity demand generated well under half of last quarter’s profit from selling power to data centres, industrials and everyone else at merchant prices, and nearly half from a retail electricity book whose earnings are going the wrong way. Both things are true at once, and only one of them is in the narrative.
| Bull case for VST | Bear case for VST |
|---|---|
| 2027 opportunity of $7.4-7.8bn excludes Cogentrix and the Meta PPAs entirely | That opportunity range is priced off market curves from 31 October 2025 – ten months stale |
| CEO bought $1.17m of stock on the open market across three days at the local low | Retail, the largest Q2 segment at 43.7% of EBITDA, fell 10.5% across the first half |
| Share count down ~30% since 2021; $1.2bn of buyback authorisation left | 94% of 2027 generation already hedged, capping the AI power-price upside until 2028 |
| FERC has approved Cogentrix; ~5,500 MW of net capacity closing into a tight market | $19.6bn of debt against a $50.2bn market cap in a rising long-rate environment |
| 97%+ commercial availability through extreme heat in ERCOT and PJM | Rising forward curves generate GAAP losses, not gains, on a fully hedged book |
The cross-industry parallel that clarifies this: Vistra right now is priced the way an airline with a fully hedged fuel book is priced when crude rallies. The commodity move that makes the sector exciting is precisely the move the hedged operator has excluded itself from. Airline investors learned to look past the headline crude print to the hedge disclosure. Power investors have not yet made that adjustment, which is a large part of why VST has underperformed the index by 20 points in a year when its own EBITDA grew 31%.
Regulatory tension: affordability politics versus the buildout
The regulatory picture is where the bear case gets its teeth. Merchant generators in PJM have spent 2026 in the crosshairs of an affordability debate. Jefferies flagged heightened regulatory risk for established power producers in January 2026 following reports of potential PJM market intervention aimed at electricity affordability and reliability – the precise mechanism by which a capacity-price windfall becomes a political liability. Capacity revenue is set by auction; auctions are set by rules; rules are set by regulators who answer to voters receiving higher bills. Vistra’s East segment EBITDA rose 54% year on year in Q2. That is exactly the kind of number that attracts a rulemaking.
The siting fight is escalating in parallel and it is now local rather than federal. The most-discussed data-centre item on Hacker News in the past month was not an economics piece at all: it was a report of protesters hauling a guillotine to a city council meeting over a proposed AI data centre, with a company representative escorted out by police. Community opposition is becoming a scheduling risk for the load growth that underwrites Vistra’s 2028-and-beyond thesis. A megawatt of contracted demand that cannot get a conditional-use permit is not a megawatt.
Against that, the federal picture is constructive. FERC approved Cogentrix. The regulatory bottleneck for Vistra is not Washington; it is state utility commissions, PJM’s capacity market rules and county planning boards. That is a harder set of counterparties to model than a single federal agency, and it is the main reason we hold a bear case as low as $95 in a company generating $7bn of EBITDA.
What happens next: our $225 bull case and $95 bear case
Bull case: $225 by end-2027. This requires three things in sequence. Cogentrix closes and contributes a full year, pushing 2027 Ongoing Operations Adjusted EBITDA to roughly $8.2bn including a partial Meta PPA contribution. The market awards 11.5x EV/EBITDA – a premium to today’s 9.1x but below where nuclear-heavy peers have traded – producing an enterprise value near $94bn. Net of roughly $23bn of post-acquisition net debt and on a share count shrunk toward 320 million by continued buybacks, that is approximately $225 per share, a 51% gain from $149.30 and just above the $219.82 52-week high. The causal chain that gets you there is the 2028 hedge roll: as the 72%-hedged 2028 book gives way to open volumes at 2026-vintage forward prices, the earnings power the market currently refuses to pay for finally becomes reportable.
Bear case: $95 by end-2027. This requires PJM or ERCOT capacity prices to mean-revert as new gas and storage capacity arrives, pulling 2028 EBITDA back toward $6.4bn, combined with multiple compression to 8.0x as the market re-rates merchant generation from growth back to utility. Enterprise value of $51.2bn less $19.2bn of net debt across 336 million shares is roughly $95 – a 36% decline. The trigger to watch is not an AI headline. It is a PJM capacity auction clearing materially below expectations, or a state-level affordability intervention that caps what generators can collect.
Our base case sits closer to $175. Three specific things to watch, with dates. First, the next PJM base residual auction result, which will tell you whether the East segment’s 54% growth was a step change or a peak. Second, the Cogentrix close and the first guidance revision that includes it – that is the moment the $7.4-7.8bn opportunity range gets restruck on current curves rather than October 2025 curves. Third, whether Jim Burke keeps buying. Three open-market purchases inside nine days is a signal; a fourth would be a stronger one, and the Form 4 lands within two business days of the trade.
For the other side of this trade – the customer rather than the supplier – see our analysis of CoreWeave (CRWV), and for the multi-name view of the energy complex our earlier energy stocks scenario piece.
Frequently asked questions
Why is Vistra stock down in 2026 when its earnings are up?
Because the earnings that are up are already contracted. Vistra had hedged approximately 100% of 2026 and 94% of 2027 expected generation as of 3 August 2026, so rising power prices show up first as unrealised mark-to-market losses – $488m of incremental losses in Q2 alone – rather than as upside. The stock is down 7.5% year to date against the S&P 500’s 12.9% gain because the market is pricing contracted cash flows, not an AI scarcity option.
What is Vistra’s price target from Wall Street?
Published targets have clustered in the $200-$245 range through 2026, with BMO Capital’s James Thalacker at $245 and Jefferies upgrading to Buy in February 2026. Our own bull case is $225 and our bear case is $95. These are scenarios built from EBITDA and multiple assumptions, not price targets, and the spread between them reflects genuine uncertainty about PJM capacity rules rather than analytical hedging.
Did Vistra’s CEO buy stock in 2026?
Yes. SEC Form 4 filings show president and CEO James A. Burke made three open-market purchases totalling 8,665 shares for $1,173,069 – 2,000 shares at $135.00 on 24 August, 2,200 at $135.99 on 31 August and 4,465 at $135.25 on 1 September 2026. All three were transaction code P, meaning open-market buys rather than option exercises. The stock closed at $149.30 on 4 September.
How does Vistra compare with Constellation Energy?
The common retail argument is that Vistra is cheaper and should re-rate toward Constellation’s multiple. The counter-argument is that the two are not the same instrument: Constellation’s earnings are more nuclear-weighted and more contracted at a floor, while Vistra’s are more exposed to merchant gas economics – except that Vistra has hedged away most of that exposure through 2027. Both stocks are well below their 52-week highs; VST closed 32% below its $219.82 peak and CEG 28% below its $412.70 peak on 4 September 2026.
What is Helix Digital Infrastructure and how much is Vistra committing?
Helix is a digital infrastructure venture announced alongside KKR, the Kuwait Investment Authority and NVIDIA, in which Vistra serves as preferred power partner. Vistra’s initial commitment is up to $1.0bn, with funding above roughly $500m tied to milestones. It is not included in the 2026 guidance ranges, and its economics will not be visible in reported EBITDA for some time.
What would make the bear case play out?
A PJM capacity auction clearing well below current expectations, a state or federal affordability intervention capping generator revenue, or new gas and battery capacity arriving fast enough to reverse the 2026 scarcity premium. Any of those would pull 2028 EBITDA toward $6.4bn and compress the multiple from today’s 9.6x toward 8x, which is the arithmetic behind our $95 level.
This article is analysis and information only. It is not investment advice, and the bull and bear levels described are scenarios, not recommendations or price targets.