The Nuclear Stocks That Actually Sell Power Are Up on Price, Not Volume
π Vistra reported a 30.9% year-over-year increase in Ongoing Operations Adjusted EBITDA to $1.767 billion for Q2 2026.
π Total consolidated revenue fell 5.5% to $4.017 billion due to a $488 million unrealized hedging mark-to-market loss.
β‘ Nuclear fleet electricity production decreased 2.2% year-over-year to 44,160 gigawatt-hours with capacity factor at 93.0%.
π° Vistra trades at a 10.3x EV/EBITDA multiple, significantly lower than Constellation Energy's 14.2x and Talen's 49.5x.
π GAAP net income of $305 million understates operational performance compared to the adjusted EBITDA figure.
π The earnings growth is driven by capacity market price increases rather than increased power generation volume.
π€ Vistra's commercial terms for its Comanche Peak agreement with AWS were not disclosed in the filings.
π Capacity market prices surged nearly nine-fold from 2024/25 to 2026/27 delivery years due to data-center load forecasts.
β οΈ The article warns that treating AI power purchase agreements as the primary earnings engine is a false narrative for these operators.
π Constellation Energy adjusted operating EPS rose 33.5% while total operating revenues grew 22.9% in the same quarter.
- Vistra achieved a 30.9% year-over-year increase in Ongoing Operations Adjusted EBITDA to $1.767 billion, demonstrating strong underlying operational performance.
- The company trades at an attractive 10.3x EV/EBITDA multiple compared to peer Constellation Energy's 14.2x and Talen's 49.5x.
- Vistra is identified as one of the three primary operators that actually own and run nuclear reactors, providing direct grid exposure.
- Total consolidated revenue declined 5.5% to $4.017 billion due to a significant $488 million unrealized hedging mark-to-market loss.
- Nuclear fleet electricity production decreased 2.2% year-over-year, indicating growth is driven by price rather than volume expansion.
- The earnings boost relies on capacity market prices that have surged nearly nine-fold based on forecasts which could unwind if data-center loads do not materialize.