The New AI Infrastructure Isn't Chips. It's Power Plants.

For most of the last decade, independent power producers were valued like the rest of the commodity complex: earnings rose and fell with natural gas prices and the weather. Vistra's second-quarter results, and the forward book of contracts behind them, are the cleanest public evidence yet that the market is starting to treat these companies as something different: critical infrastructure underneath the AI economy.

Vistra, one of the largest U.S. independent power producers, reported Q2 2026 Ongoing Operations Adjusted EBITDA up more than 30% year over year to $1.77 billion (Vistra Q4 2025 earnings call transcript, The Motley Fool). The fleet ran at 97% or greater commercial availability, meaning the company's power plants were actually running and selling electricity rather than sitting in maintenance outages, across strong demand in Texas and the PJM grid, the regional electricity market stretching from the Mid-Atlantic into the Midwest. Vistra reaffirmed full-year 2026 guidance: Ongoing Operations Adjusted EBITDA of $6.8 billion to $7.6 billion, and Ongoing Operations Adjusted FCFbG, or free cash flow before growth capital, of $3.925 billion to $4.725 billion (Vistra Q3 2025 results press release).

The number that actually matters sits a layer down. As of early August, Vistra had hedged roughly 100% of its expected 2026 generation, 94% of 2027, and 72% of 2028 (Insider Monkey, citing Wolfe Research and Vistra disclosures). A "hedge" in this context is a forward contract that locks in a future sale price for electricity, effectively converting a volatile commodity business into something closer to a contracted utility. When a power producer can tell the market that the next two years of cash flow are already priced, the market starts to value the company like infrastructure: steady cash flows, low risk, durable, rather than like a coal-and-gas bet that swings with the weather.

That is the re-rating, and it is showing up in the analyst note. Wolfe Research's Steve Fleishman, one of the longest-running sell-side power analysts, kept a Buy on Vistra with a $232 price target, framing the print as evidence that the company's earnings power is structural rather than cyclical (Insider Monkey summary of Wolfe Research). Vistra has also committed $1 billion to Helix, the KKR-led $10 billion-plus digital infrastructure platform that will house hyperscale AI data centers (MLQ news on KKR's Helix launch). The capital is small relative to Vistra's market value, but the direction of travel is the point: a power company investing in the data center buildout, rather than waiting for it as a customer.

Two cautions belong next to the thesis. The first is the cyclical-versus-structural line. Independent power producers still move on natural gas prices, weather, and capacity markets. Those are the same forces that have always moved them. Data center load is layered on top, not a replacement. The second is hedging itself. A forward book that locks in 100% of 2026 looks like bulletproofing when forward power prices are high. It is a liability if forward prices collapse and the company has committed to sell below the market. The hedge book cuts both ways, and the next 18 months of wholesale power prices will be the test of whether 2026 was the start of a structural premium or a peak-cycle high.

The "AI infrastructure" label is a market framing, not a contractual one. Most of Vistra's AI-related exposure is still a forecast: the data center customers who have not yet signed long-term contracts, the load growth that ERCOT, the Texas grid operator, projects but has not yet connected (stockanalysis.com Vistra transcripts archive). The falsifier for the re-rating is concrete: a wholesale power price collapse, a data center offtake that fails to materialize, or a regulatory cap on the price a generator can charge data center customers. Until one of those breaks, the market is going to keep paying for cash flow visibility it did not pay for a year ago.