Vistra reported its second quarter on 7 August 2026, before the open. We are writing about it today.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
Why Vistra Matters at All
Vistra generates electricity and sells it. It owns power plants, including a nuclear fleet, and it also sells directly to residential and business customers.
And why is that an AI story? Because data centres consume enormous amounts of power, and they need it available twenty-four hours a day. Nuclear provides exactly that. It is why Vistra and Constellation Energy became stocks that trade on the AI thesis rather than only on the power thesis.
The Headline Against the Numbers
The company reports adjusted EBITDA up more than 30%. That is true:
| The quarter | A year ago | |
|---|---|---|
| Adjusted EBITDA, ongoing operations | $1,767 million | $1,349 million |
And now the other lines:
| The quarter | A year ago | |
|---|---|---|
| Revenue | $4,017 million | $4,250 million |
| GAAP net income | $305 million | $327 million |
| EBITDA before adjustments | $1,411 million | $1,417 million |
Revenue fell. Net income fell. And EBITDA before adjustments fell by $6 million.
So Where Does the $418 Million Increase Come From
From one line: unrealised hedging.
What is unrealised hedging? A power producer sells part of its future output forward under contracts, to lock in a price. Those contracts are marked to their market value each quarter, before the electricity has actually been delivered. If the market price of power has risen, a contract that locked in a lower price "loses" on paper. That loss is non-cash, and it is stripped out of adjusted EBITDA.
This quarter a loss of $472 million was added back. In the comparable quarter a year ago a gain of $16 million was deducted instead. A swing of $488 million in a single line - more than the entire reported increase.
This does not mean the company did anything improper. Stripping this line is standard practice across the industry, and the company presents the full reconciliation. But anyone reading "growth of more than 30%" and picturing a business a third larger comes away with the wrong picture.
What Genuinely Did Improve
The operating segments showed real improvement:
| Segment | The quarter | A year ago |
|---|---|---|
| Texas | $311 million | $142 million |
| East | $642 million | $418 million |
| West | $68 million | $49 million |
| Retail | $773 million | $756 million |
And the cash: first-half operating cash flow was $2,222 million against $1,171 million a year ago. That is close to a doubling, and it is a real cash number that cannot be adjusted away.
But note the last line of the table. Retail, the segment carrying direct customer margin, rose only 2.2%.
Guidance
Full-year guidance was reaffirmed rather than raised: adjusted EBITDA of $6,800 million to $7,600 million.
And two details worth knowing about it. First, per a footnote in the reconciliation table in the filing itself, the guidance was prepared in November 2025 on market curves from the end of October. Second, it excludes the Cogentrix acquisition and the Meta power agreements - which is to say, precisely the catalysts drawing investors to this stock.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
This is the second time this week I am writing the same sentence, and it is starting to be a pattern.
At Nova the highlight line placed "record net income" beside a percentage describing earnings per share. At Vistra the headline places "growth of more than 30%" beside a number that is almost entirely a hedging line. In neither case is there a lie, and in both cases the fast reader comes away with the opposite conclusion.
What I take from it is a simple rule: when a company reports both a GAAP number and an adjusted number, and they point in opposite directions, look at the bridge between them. Here the bridge explains the whole story.
Even so, I do not want this to sound too negative. Operating cash flow nearly doubled in the half, and that is not a line you can adjust away or paint. Operational execution in Texas and the East improved substantially, and that is real.
What troubles me is the guidance. A company trading on the AI electricity story is reaffirming a forecast set in November, one that excludes the two deals that are the story itself. I want to see an updated forecast that brings them inside. Until then, the number the company shows the market is not measuring what the market is buying.
