Vistra (VST): what it does, how it makes money, and why a Meta nuclear deal has not re-rated it

The standard $VST story writes itself: a power company signs a giant nuclear deal with $META, and AI demand for electricity does the rest. Then the tape refuses to cooperate. Over the year to 2026-09-11 the stock is down about 27%, even though its most recent quarter grew adjusted operating earnings by roughly 30%.

The more accurate frame: the Meta deal is real, but it is small relative to the whole company and it arrives on a schedule, not on the day it was announced. Vistra is mostly a merchant power producer whose results still move with electricity prices, and the contracted AI story is a slice of it that pays out over years. This piece walks through what Vistra actually owns, how it makes money, what the Meta contract really contains, where it sits in the datacenter-power stack, and why the operations and the multiple have moved in opposite directions. Figures are as of 2026-09-11 unless noted.

Why it matters now

The gap is the story. In the second quarter of 2026, adjusted EBITDA came in at about $1.77B, up roughly 30% year over year, and full-year 2026 guidance was reaffirmed at $6.8B to $7.6B. Over the same twelve months the stock fell about 27%. When a business improves and its price falls, the market is repricing something other than this quarter's results: here, how much of Vistra deserves a contracted, AI-grade multiple and how much still trades like merchant power. That is the question worth answering carefully.

What does Vistra do?

Start with what Vistra is, because the phrase "power company" hides the structure. Vistra is an independent power producer: it owns the plants and sells the electricity, without a regulated monopoly setting its returns. That is the key difference from a utility. A regulated utility earns an allowed return on its asset base; Vistra earns whatever the power it generates can fetch, plus whatever it has locked in by contract.

The fleet is about 50 GW of generation across 18 states. The crown jewel is nuclear: about 6,448 MW, the second-largest competitive nuclear fleet in the US. Around it sits a large gas fleet, expanded by the roughly $4B Cogentrix acquisition (10 gas plants, about 5.5 GW). And Vistra runs retail brands, TXU Energy and Dynegy, serving about 5 million customers. That retail book matters: it lets Vistra match its wholesale generation against its own customers instead of selling all of it naked into the spot market.

The Meta deal: read the split

In January 2026, Vistra agreed to supply Meta with 2,609 MW of carbon-free power on 20-year contracts. The headline number hides the part that sets the value: not all of it exists yet.

• 2,176 MW comes from plants operating today: 1,268 MW from Perry and 908 MW from Davis-Besse, both in Ohio.
• 433 MW comes from uprates, engineering work to lift output at existing plants: 213 MW at Perry, 80 MW at Davis-Besse, and 140 MW at Beaver Valley in Pennsylvania. That engineering is not yet done.

The timing is the part most coverage skips. The operating power starts flowing in late 2026 and is fully delivered by the end of 2027. The uprates do not start until 2031 and run to 2034, and they depend on NRC approval. So the contract pays out over years, and a meaningful piece of it is generation that has to be built and approved before it can be sold.

How Vistra makes money

Vistra makes money by generating electricity and selling it, through three channels that behave very differently:

• Merchant power, the large majority. Output sold at market prices and into capacity auctions (the PJM and ERCOT markets). This rises and falls with electricity prices, which is why the print moves with power markets more than with any AI headline.
• Contracted power, the growing slice. Long-dated deals like the Meta PPA lock in price and volume for decades, which is the kind of revenue the market is willing to pay a higher multiple for.
• Retail, through TXU Energy and Dynegy, which pairs generation with a customer book and smooths some of the merchant exposure.

There is also a second AI channel worth holding carefully. In June 2026, KKR launched Helix Digital Infrastructure alongside the Kuwait Investment Authority and $NVDA, with more than $10B of long-duration capital committed, and Vistra was named its preferred power provider. That is a real relationship and a real door. It is also, today, zero signed megawatts. Treat it as optionality, not revenue.

The headline risk is scale. 2,609 MW is contracted out of a roughly 50,000 MW fleet. The overwhelming majority of Vistra's cash flow is still merchant. That is the number to hold onto through the rest of this piece.

Where it sits in the datacenter-power bubble

Vistra is a core name in QA's Datacenter Power bubble: the cluster built on one idea, that AI eats electricity and the companies that generate it are the unlock. For the full map of that bubble, see the datacenter power bubble explainer. Vistra also maps onto the Nuclear, AI Utility, and AI Power themes.

The useful way to place Vistra is by layer. It generates electricity at utility scale, from an existing fleet. That separates it from equipment makers like GE Vernova, which sell the turbines and grid gear; from on-site power like Bloom Energy, which puts generation next to the load; and from next-generation nuclear developers like Oklo, whose reactors are not yet producing power. Downstream, the data-center plant that consumes the power, the cooling and power-conditioning gear of names like Vertiv, is a different layer again. Vistra's edge is that its nuclear capacity already runs.

Its peers make the bubble visible. VST's tightest correlations in the QA universe (252 trading days, market beta stripped out) run to Talen Energy ($TLN, 0.79), Constellation ($CEG, 0.78), and NRG ($NRG, 0.76), all Datacenter Power names. After those three the correlation drops sharply, to about 0.5. So VST trades as part of a tight independent-power cluster: owning it next to TLN, CEG, or NRG is less diversified than the four separate tickers suggest.

The numbers

One row deserves a caveat. The Street mean target is an estimate, an average of analyst opinions, not a promise or a QA view. A target well above the current price on a stock that fell while its operations improved mostly tells you the Street expects the re-rating the tape has not delivered; it does not tell you when, or whether, it arrives.

The bull case

• The operations are not the problem: Q2 2026 adjusted EBITDA up about 30% year over year, and full-year guidance reaffirmed.
• The nuclear capacity already runs. Vistra is selling electrons from operating plants, not promising future reactors.
• Contracted revenue is growing. The Meta PPA converts part of the fleet into 20-year, fixed-volume demand, the kind of revenue that earns a higher multiple.
• Optionality through Helix: a named preferred-provider role with a KKR, Kuwait Investment Authority and NVDA vehicle carrying more than $10B of committed capital.

The bear case

• Scale. 2,609 MW contracted out of about 50,000 MW means the large majority of cash flow is still merchant, and power prices plus PJM and ERCOT capacity auctions move the print more than the AI headline does.
• Not all of the deal exists. The 433 MW of uprates depend on NRC approval and execution running to 2034.
• Nuclear concentration cuts both ways. An unplanned outage at Perry, Davis-Besse or Comanche Peak would hit the contracted delivery and the merchant book at the same time.
• Helix is a door, not revenue: zero signed megawatts today.

How to access

Vistra trades on the NYSE as VST, a clean US listing, so direct ownership needs nothing exotic. To trade it from a US-retail account alongside the rest of the AI-power names, see /stack/ibkr. The live ETF holdings breakdown, showing which funds carry VST and at what weight, is on /stocks/vst.

Bubble-correlation shifts and rule-based alerts on $VST, the kind that fire when a name crosses a curated level or its bubble correlation breaks, are part of /pro.

What to watch

• Delivered megawatts on the Meta contract: whether the operating 2,176 MW starts flowing on the late-2026 schedule.
• The PJM and ERCOT capacity auctions and power prices, the variables that still drive most of the cash flow.
• NRC progress on the 433 MW of uprates, the part of the deal that does not exist yet.
• Any signed megawatts through Helix, the step that would turn a relationship into revenue.
• The curated $140 level (marked 2026-04-25, just below the recent ~$148 tape) is best read as the nearest observable structural support, not a trade level or a target.

Live data on this ticker: /stocks/vst. Price, ETF holdings, bubble correlation, curated levels, bot positions.

Bubble context: /bubbles/datacenter-power. The cluster this name belongs to and how it's moving.

QuantAbundance is educational research. Nothing here is investment advice. See /disclosures.