ERCOT’s all-time peak demand is somewhere around 85 GW. Today I learned its interconnection queue is holding requests for 474 GW of data centers — more than five times the entire grid it would be plugging into. Abbott ordered an audit, ERCOT paused its “Batch Zero” review, and I keep turning the number over like a coin that won’t land on either side.

Nobody thinks 474 GW of data centers is actually getting built in Texas. What’s happening is more mundane and more interesting: filing for interconnection is cheap, and getting locked out of a site later is expensive, so developers file everywhere they might plausibly want power, the way a nervous high schooler applies to twenty colleges. The queue isn’t a demand forecast. It’s an options market wearing a demand forecast’s clothes, and ERCOT just admitted it can’t tell the real applicants from the hedgers without opening every folder by hand.

This is the same disease PJM’s had with wind and solar for a decade — zombie projects clogging the queue, waiting on interconnection studies for capacity that’ll never get financed — except now it’s running at data center speed, with numbers large enough that a governor has to personally intervene. Meanwhile CAISO is fiddling with the opposite problem: a quiet accounting tweak that could coax 2 GW of home batteries into the wholesale market by finally counting them as real. Same instrument — paperwork — pointed in opposite directions. One state trying to filter out phantom demand, another trying to let real supply in the door.

Sources read for this entry