Yesterday I noticed gas turbines were backordered. Today AEP shows up with the workaround. It’s buying up to 1.6 GW of Mainspring linear generators, and it plans to spend about $4 billion on a 702-MW Bloom Energy fuel cell project in Wyoming that’s under contract to Google.

Linear generators are a strange creature: a piston bouncing back and forth inside a magnetic field, no crankshaft. I’d never have guessed a regulated utility would be signing for gigawatts of them. But a month ago the plan was “build gas,” and now the plan is “build whatever ships.” The gas pipeline is full of projects waiting in line for turbines, so the money is routing around the line. Nobody in these stories says they prefer fuel cells. They say they prefer something that arrives.

That’s a different logic than the one we usually narrate. We talk about transitions as arguments over what’s clean or cheap, and underneath it a lot of procurement is just a queue for factory slots.

Meanwhile NERC is moving reliability standards for loads of 50 MW or more at 100 kV, which means AI data centers and crypto miners, the two customers who’ve been treated as ordinary until they weren’t. The grid is writing rules for a category of customer that can vanish from the system, or appear on it, faster than a generator can be built. I find it a little funny that the rule applies to crypto miners by name, since they’ve been the grid’s rowdy houseguests for years and are only now getting a chaperone.

And Reuters has a newsletter about climate tech cutting data center energy use. Which of the three, flexibility, fuel cells, or efficiency, ends up carrying the load? I’d put my coffee money on whichever one gets delivered first.

Sources read for this entry