Hard to Lobby a Foundry

Buried in BofA’s report today is a sentence that does more work than the 100-gigawatt supply gap it’s attached to: large gas turbines are largely sold out through 2030. Not a permitting problem, not a NIMBY problem, not even really a money problem. GE Vernova, the newly rebranded Omterra (formerly Siemens Energy), Mitsubishi Power: order books full for years, and no amount of capital changes how fast a factory machines a rotor. Two years of headlines about interconnection queues and capacity auctions treated the constraint as paperwork. Turns out somewhere down the chain it’s a foundry.

Which explains the report’s other finding, basically “utilities will do whatever’s left”: extend coal plants that were supposed to retire, lean harder on batteries, let data center developers buy small on-site gas engines instead of the big turbines nobody can get. Less a strategy than a scavenger hunt.

Meanwhile Best Buy put solar panels on a Long Island City rooftop and called it a milestone, its first “brick-and-mortar community solar” store. Genuinely nice, and part of a net-zero pledge from 2020 that predates any of this AI-load talk entirely. Set next to Google’s 2.5-gigawatt Arkansas complex from last week, though, the two don’t even share a unit of measurement. One company is buying turbines it can’t get. The other put up panels the size of a parking lot and wrote a press release about it. Both get filed under “solar.” Neither is playing the same game.

The turbine backlog is the part I keep coming back to. Two years of fighting over who pays for transmission and who jumps the interconnection line, and the actual bottleneck turns out to be how many rotors a factory can machine in a year. Hard to lobby a foundry.