Blocking vs. Building: a growing grid behind a permitting bottleneck

A growing grid can only build what its permitting system lets through each year. This model asks what that does to prices and to fossil fuel generation when you speed the process up, leave it slow, use it to block gas plants, or wait for a better deal.

How it works. Two stocks, clean capacity and the fossil fleet, in average gigawatts delivered. Demand compounds. Each year the grid wants to build for demand growth, replace retiring plants and keep a 15 percent reserve margin, but it can only build what the rules allow (a pace that scales with the grid), and about 80 percent of what gets built is clean. Fossil plants supply whatever clean power can't, so the fossil power needed (our emissions proxy) is demand minus clean output. When the grid runs short, prices rise and old plants that were due to close stay open.

What it shows.

1. Blocking a gas plant only cuts fossil power if something clean gets built in its place. Otherwise it changes nothing except the reserve margin: higher prices, older plants.
2. Build pace matters until it covers demand growth plus retirements, about 19 GW a year here. Past that, extra pace buys nothing.
3. A gas ban is a price policy. Across every setting we tried it raises prices by 2 to 20 percent. For emissions it does nothing while the bottleneck binds and gains a few percent once the bottleneck is cleared, because a slightly short grid keeps building clean. What costs emissions is delay.
4. Holding the bill hostage to get the ban is a bad bet. If you win, emissions end 1 to 3 percent lower than the deal and prices 4 percent higher. If you lose, you get no reform: 11 percent more fossil power and 9 percent higher prices. On emissions alone you need to be 80 to 90 percent sure of winning to break even, and prices lose in every branch. The ban's value comes from a cleared bottleneck, so if you want it, pass the bill first and fight for the ban after.

Premises and limits. The model assumes the bottleneck binds today, which five-year queues, tightening margins and deferred coal retirements suggest. Capacity is in effective gigawatts, so the reserve margin is a stylized slack index, not NERC's number. Demand doesn't respond to price. Coal and gas are lumped together, which understates the climate cost of keeping old coal open. Numbers are stylized, roughly the US in 2026. Rhodium's full energy-system model of the bill finds prices 3 to 5 percent lower and power emissions up to 18 percent lower by 2040; this toy lands at 7 and about 20 percent by 2046, for a bill that lifts the pace to 18 to 20.

Built by Second Order Lab around Datta and Amarnath's "Energy isn't just about climate change" (The Argument, Nov 6, 2026).