3 ms·

> It really ought to be possible to structure the utility contracts such that a new data center lowers every one else’s rates instead of raising them. That wou

by RhysU 6mo ago

> It really ought to be possible to structure the utility contracts such that a new data center lowers every one else’s rates instead of raising them. That would imply that increased demand strictly decreases prices, no? Given fixed supply, no dice. To expand supply, one would use incrementally more expensive mechanisms to generate the incremental supply. (Because, why wouldn't you already be generating via the cheapest supply?) Either all existing customers would pay the same and the new customer would pay the higher rate OR smear the incremental costs across everyone. Prices might hold steady under the former choice but they would not decrease. Is the idea that the new customer would unlock some better generation capabilities through capital investment? Something not already incentivized by the distributed grid? Or is the idea that one should soak the new customer to subsidize the existing ones? Maybe rejigger some pricing tiers to push more of the existing customers into lower tiers while charging the new customer more. My guess is you're proposing this last option. I otherwise can't see how to square your suggestion with supply vs demand.

amluto 6mo ago
I can think of a couple of utility models that could work and one that definitely doesn’t. First, the bad. Build a facility that consumes 1GW at existing rates in a market with slow growth like the US. The supply sources are roughly fixed, so the grid will need to run more expensive sources. Prices go up. Now the good. Choice A, in a growing market like China and like the US arguably should have. Lots of demand is coming online all the time (not just datacenters), and everyone plans for this. Power plants of various sorts get build, and there is so much construction that costs can be quite low. Oh well, we can wish. Choice B: suppose there’s a market with roughly constant demand and enough cheap supply to go around (maybe a good hydro resource or some solar and wind and/or cheap natural gas). Residents have cheap power, which is a good thing. But the hydro doesn’t magically get bigger just because someone builds a 1GW data center. Some careful market design is needed, but that datacenter’s grid connection could be contingent on the operator actually sourcing 1GW of new generation and paying the marginal cost of its demand, with appropriate corrections if the time that the generation produces doesn’t line up with the demand. As possible pretend numbers, suppose that existing prices, all-in for the customer, are 12 cents/kWh. 5c of that is distribution and we’ll ignore it. So the data center operator sources 1GW of average supply at 10c/kWh and tries to connect itself fairly directly, so their transmission is cheap. They are allowed to buy from everyone else and sell to everyone else, but they are paid 2c/kWh selling to the grid (which the grid and residents like!) and they pay 15c/kWh when they plus whatever capacity they supplied have a shortfall and they need to buy. And, if the numbers were picked right, the grid makes a small profit selling peak power to the datacenter while still selling at peak times to residents at 12c/kWh. Would this work? I don’t know, but I think it could be done in a way that makes residential and ordinary commercial rates go down as a result of someone building a giant new load and also paying for the new capacity to supply it.

by RhysU 6mo ago

amluto 6mo ago

I can think of a couple of utility models that could work and one that definitely doesn’t. First, the bad. Build a facility that consumes 1GW at existing rates in a market with slow growth like the US. The supply sources are roughly fixed, so the grid will need to run more expensive sources. Prices go up. Now the good. Choice A, in a growing market like China and like the US arguably should have. Lots of demand is coming online all the time (not just datacenters), and everyone plans for this. Power plants of various sorts get build, and there is so much construction that costs can be quite low. Oh well, we can wish. Choice B: suppose there’s a market with roughly constant demand and enough cheap supply to go around (maybe a good hydro resource or some solar and wind and/or cheap natural gas). Residents have cheap power, which is a good thing. But the hydro doesn’t magically get bigger just because someone builds a 1GW data center. Some careful market design is needed, but that datacenter’s grid connection could be contingent on the operator actually sourcing 1GW of new generation and paying the marginal cost of its demand, with appropriate corrections if the time that the generation produces doesn’t line up with the demand. As possible pretend numbers, suppose that existing prices, all-in for the customer, are 12 cents/kWh. 5c of that is distribution and we’ll ignore it. So the data center operator sources 1GW of average supply at 10c/kWh and tries to connect itself fairly directly, so their transmission is cheap. They are allowed to buy from everyone else and sell to everyone else, but they are paid 2c/kWh selling to the grid (which the grid and residents like!) and they pay 15c/kWh when they plus whatever capacity they supplied have a shortfall and they need to buy. And, if the numbers were picked right, the grid makes a small profit selling peak power to the datacenter while still selling at peak times to residents at 12c/kWh. Would this work? I don’t know, but I think it could be done in a way that makes residential and ordinary commercial rates go down as a result of someone building a giant new load and also paying for the new capacity to supply it.