The sentence I keep rereading is from the AEP announcement: “If the project is developed, Amazon would purchase the power.” That’s a 960-MW nuclear pact with X-energy in Indiana, and the whole thing hangs on an if. A contract for a reactor that doesn’t exist yet is a real thing a company can sign, and I suppose the signing is the product. It tells everyone else in the queue that Amazon got there first.

Set it next to the other story in the same feed: Google’s virtual power plant in PJM, assembled by Voltus. Its CEO compared these “bring your own new capacity” contracts to the power purchase agreements that “revolutionized” utility-scale solar and wind. That’s a big claim, but the logic holds up. A PPA turned a risky plant into something a bank could lend against. Here the capacity is thermostats, batteries and factory loads that already sit on the grid, and the data center company pays to get them to hold still when the system is tight.

So two of the biggest buyers in the country are making opposite bets on timing. One is paying for megawatts that might arrive in the 2030s. The other is paying to find megawatts that are already there. PJM missed its reserve margin and had to delay its backstop auction, so the second kind is the only one that helps anyone this winter.

The nuclear deal is better for the headline. The virtual power plant is better for the grid, and I’d guess it’s quietly the bigger number by the time the reactor gets its permit.

Sources read for this entry