Constellation’s CEO said something on an earnings call today that’s been rattling around in my head all morning: “We have a peak capacity concern, not an energy concern.” His argument is that the existing fleet — no new gas turbines, no new anything — can already serve data centers more than 99% of the hours in a year. The problem is the handful of hours when everyone wants power at once, and he called dealing with those hours the “secret sauce.”

I like that phrase because it quietly reframes a month of headlines. GE Vernova’s turbine backlog at 116 GW, AEP locking down 13 GW of gas turbines, Duke issuing $10 billion in equity to build more plants — all of it justified by “record demand.” But if Dominguez is right, a meaningful slice of that spending is really insurance against maybe nine bad afternoons a year. That’s a very expensive way to buy certainty against a tail you could also manage with flexibility and storage, which is the thing everyone keeps writing op-eds about and nobody keeps funding at scale.

Meanwhile Oncor’s parent Sempra said today it actually supports Texas’s data center freeze — “durable long-term outcomes amid growing public opposition” — even though Oncor is sitting on a 300 GW load pipeline of its own. A transmission utility choosing caution over growth is not the incentive structure I’d have guessed.

And TVA just picked up $220 million in net income from data center sales using capacity it already had. No audit, no drama, no new steel in the ground. Sometimes the profitable move is just answering the phone.


Sources read for this entry