Vistra has been one of the major beneficiaries of the AI power narrative over the last few years, but after the stock’s recent correction I think the story deserves another look. The important distinction with VST is that this isn’t simply a bet on data centers continuing to consume more electricity.
Vistra already owns the thing everyone suddenly needs: large amounts of existing, grid-connected and dispatchable power generation.
The company operates roughly 44 GW of generation across natural gas, nuclear, coal, solar and battery storage, while also serving approximately 5 million retail customers. More importantly, much of that generation sits directly inside the markets seeing some of the largest projected load growth in the country—particularly ERCOT and PJM.
That combination is becoming increasingly difficult to replicate.
Building new generation sounds straightforward until you consider transmission requirements, interconnection queues, permitting, equipment availability and construction timelines. Gas turbines, transformers and other critical equipment remain constrained, while connecting a new plant to the grid can take years. That makes an existing MW considerably more valuable than a hypothetical MW five years from now. And Vistra owns a lot of them.
The Moat
What separates Vistra from many of the other independent power producers is its integrated model. The company combines wholesale generation with one of the largest competitive retail electricity businesses in the country. That retail operation creates a natural hedge against the generation portfolio. When wholesale electricity prices move higher, generation economics improve. When prices fall, the retail side can benefit from lower procurement costs.
It’s one of the reasons I view VST as somewhat different from a pure merchant generator.
The company then layers an extensive hedging program over the portfolio. As of August, Vistra had hedged approximately:
2026: 100%
2027: 94%
2028: 72%
That is an important part of the story.
We aren’t simply depending on tomorrow’s power curve staying elevated. A substantial portion of future economics has already been locked in, giving management considerably more visibility into cash generation.
AI Changes the Value of Existing Power
Data centers are creating an unusual problem for hyperscalers. They have capital. They have chips. They can build the data centers. What they increasingly don’t have is enough reliable electricity available exactly where and when they need it.
A hyperscale facility consuming hundreds of megawatts, or potentially more than a gigawatt, cannot simply wait five or six years for new generation and transmission. That shifts negotiating leverage toward owners of existing generation. Nuclear is especially interesting because it provides around-the-clock carbon-free power, while natural gas provides the dispatchability necessary to balance increasingly complicated grids.
Vistra owns both.
The Meta transaction gives us a good example of what this can look like. Vistra signed 20-year PPAs with Meta covering approximately 2.6 GW of nuclear capacity from Beaver Valley, Davis-Besse and Perry. Approximately 2.18 GW comes from existing generation while another 433 MW will come from nuclear up-rates.
This is important.
Vistra doesn’t have to build 2.6 GW of entirely new generation from scratch. Instead, it is monetizing assets that already exist and investing incremental capital to increase output from sites where much of the infrastructure is already in place.
That’s a significantly different return profile.
Vistra estimates the incremental EBITDA associated with these Meta contracts will convert to Adjusted Free Cash Flow before Growth at roughly 80%, excluding the capital required for the uprates and tax effects.
That is exactly the type of AI infrastructure exposure I prefer.
Rather than spending enormous amounts of capital hoping future demand appears, Vistra can contract existing assets and selectively deploy capital where customers are effectively underwriting the investment.
And There Is Still More Capacity to Monetize
The Meta agreement doesn’t exhaust the opportunity.
Management has identified approximately 3.2 GW of additional nuclear capacity across Beaver Valley and Comanche Peak that could potentially be contracted under long-term agreements. There may also be roughly another 300 MW of expansion opportunities within its existing PJM gas fleet.
This creates something I think the market occasionally underestimates: contracting optionality.
Every additional long-term agreement potentially converts another portion of Vistra’s merchant earnings into predictable contracted cash flow. The business therefore becomes more valuable even without generating considerably more electricity. The earnings mix itself improves. Management has already indicated that, based on contracts signed to date and the stability of the retail business, nearly half of future Adjusted EBITDA could eventually come from highly stable earnings sources. More PPAs would push that percentage higher.
Cogentrix Adds Another 5.5 GW
Then there is Cogentrix. Vistra is acquiring approximately 5.5 GW of primarily modern natural-gas generation across PJM, ISO-New England and ERCOT. Again, look at the locations. These aren’t random assets. They sit inside some of the markets where electricity supply is becoming increasingly valuable. Management expects the transaction to produce mid-single-digit per-share free-cash-flow accretion in 2027 and high-single-digit average accretion from 2027 through 2029.
Vistra also expects the investment to exceed its mid-teens levered-return hurdle. The acquisition has already received FERC approval and is expected to bring Vistra’s overall fleet toward approximately 50 GW once completed. That scale matters.
Hyperscalers aren’t searching for a 50 MW solution anymore. They’re increasingly looking for hundreds or thousands of megawatts. Very few independent generators have enough assets, geographic diversity and balance-sheet capacity to negotiate those types of contracts.
Follow the Cash Flow
This is probably the most important part of the thesis. Vistra expects more than $10 billion of cash generation across 2026 and 2027.
The current capital allocation framework roughly looks like this:
- About $4 billion toward growth investments including Cogentrix, Permian gas generation and the nuclear uprates.
- About $3 billion toward shareholder returns through buybacks and dividends.
- And approximately $3 billion potentially remaining for additional allocation through year-end 2027.
Meanwhile, management expects net leverage around 2.3x Adjusted EBITDA by the end of 2027 under that framework. That’s significant because Vistra isn’t being forced to choose between growth, deleveraging and shareholder returns. At least under the current earnings outlook, it can potentially do all three.
The buyback program has already been meaningful.
Since November 2021, Vistra has repurchased approximately $6.5 billion of stock and reduced the outstanding share count by roughly 30%. That becomes increasingly powerful when the share price falls while underlying cash generation remains intact.
The Numbers Remain Strong
Q2 reinforced the fundamental story. Ongoing Operations Adjusted EBITDA increased more than 30% YoY to $1.77 billion. Management maintained 2026 guidance of:
- Adjusted EBITDA: $6.8B-$7.6B
- Adjusted FCF before Growth: $3.925B-$4.725B
Vistra also continues to see a $7.4B-$7.8B midpoint opportunity for 2027 Adjusted EBITDA. Importantly, that 2027 range currently excludes the pending Cogentrix acquisition and the Meta PPAs. So there remains a legitimate pathway for earnings estimates to move higher as those contributions become incorporated into the model.
What Can Go Wrong?
There are certainly risks here. The biggest is probably the same one affecting the entire AI infrastructure trade.
If hyperscaler CAPEX slows materially, data-center development gets pushed out or grid operators dramatically reduce their load forecasts, forward electricity prices could fall and some of the scarcity premium attached to generation assets would disappear.
Regulation also remains a major variable.
PJM continues working through data-center interconnection, colocation and capacity-market reforms. Texas is also becoming more aggressive about evaluating enormous new load requests.
I don’t necessarily view those developments as bearish. In some cases they’re simply the grid acknowledging that demand requests are becoming too large to process using the old framework. But they can delay projects.
Vistra also remains exposed to operational outages, weather volatility, commodity prices and execution risk surrounding acquisitions and new generation projects.
And finally, investors shouldn’t assume every projected gigawatt of data-center demand actually materializes. Some won’t. The important question is whether enough demand arrives to keep power markets tight. At this point, I believe it will.
Where I Stand
For long-term positioning, I continue to like the setup. The fundamental thesis isn’t that AI causes electricity demand to rise forever at today’s projected rate. The thesis is simpler: Reliable, existing power generation located in constrained markets is becoming more valuable.
Vistra has one of the largest collections of those assets in the country.
It also has the retail business to stabilize the portfolio, a heavily hedged near-term earnings profile, substantial internal cash generation and several billion dollars of additional nuclear capacity that could still be moved into long-term contracted structures.
That’s a strong combination.
For swing trading, the setup is becoming more interesting after the significant reset from the highs. VST recently traded around the upper-$130s compared with a 52-week high near $220. At the same time, the underlying earnings outlook hasn’t experienced anything approaching that magnitude of deterioration.
Wall Street targets have come down in several cases, but the current consensus remains around $217, with many major firms still maintaining Buy or equivalent ratings. I don’t use analyst targets as a reason to buy a stock, but the divergence is worth noticing.
The stock has experienced substantial multiple compression while the business continues producing higher Adjusted EBITDA, contracting more generation and adding additional capacity. That creates the type of setup I generally prefer: fundamentals improving while price resets expectations.
I would still treat VST as a higher-beta position rather than a traditional defensive utility. It trades much more like an infrastructure/growth stock and can move violently as power prices, AI sentiment and interest rates change.
So position sizing matters.
For a swing, I’m more interested in buying confirmation following weakness rather than trying to catch every downside move.
For a longer-term position, I think pullbacks are increasingly attractive provided the fundamental thesis remains unchanged: strong cash generation, continued power-market tightness, successful Cogentrix integration and additional conversion of merchant nuclear capacity into long-term contracts.
The next major rerating catalyst may not necessarily be another earnings beat. It may simply be another hyperscaler signing a long-term agreement for some of that remaining 3.2 GW of nuclear capacity. If that happens, the market will have to assign a higher value not only to that contract, but potentially to everything Vistra still has available to monetize.
That’s the optionality I’m watching.
