Texas’s AI “scarcity rent” is being earned inside merchant P&Ls — not paid for by customers that think they’re buying power

Texas’s merchant generators don’t simply “sell kWh at spot.” In ERCOT’s energy-only market, the scarcity premium is earned when price clears high relative to marginal cost—then that cash flow is economically determined by how much future generation is hedged when the AI-driven demand hits.

For investors, the key shift is that the AI load story can look like a straightforward revenue tailwind while the reported P&L can lag, lead, or even flip sign depending on hedge roll timing and settlement mechanics.

The brief’s core set—Vistra, NRG, Talen, and Calpine (via legacy Calpine/asset footprint)—sits on the center of ERCOT scarcity economics, but their earnings can reflect different hedge coverage shapes and risk balances across years.

Below is the hard, investor-relevant linkage: companies explicitly state the percentage of expected generation volumes hedged across the ERCOT-relevant horizon window. That percentage is the bridge between “AI causes higher scarcity prices” and “whose P&L actually captures it.”

Vistra’s mix matches how a hedge-driven merchant behaves: results can stay strong even when forward spot exposure is not maximal, because hedges smooth the path between scarcity price regimes and reported settlement timing. Importantly, Vistra also telegraphs the gradient—hedged coverage stays very high in 2026, then drops in 2027.

That combination is why this can be an “underrated” trade: the AI load narrative may already be priced, but hedge-roll timing into 2027 can dominate near-term earnings revisions.

This is also where “merchant power” becomes an actual supply-chain play. Data-center power procurement doesn’t just increase demand; it forces grid adequacy planning, which changes the probability-weighted path of scarcity events. Merchant generators are positioned to capitalize on that path, but the distribution of realized rent across years is governed by hedge structure.

To underwrite the Texas AI scarcity trade, you want one simple dashboard per merchant: (1) stated hedged share of expected Texas volumes for the next 12 months and the subsequent year, (2) how that share changed since the prior quarter, and (3) whether realized results are consistent with that hedge shape.

In the sources accessible here, Vistra provides a clean, investor-usable example of this gradient, while Talen provides another datapoint. The remaining names in the brief require direct, verifiable hedging disclosure to avoid guesswork.