AI and Bitcoin miners can be a part of the solution rather than the problem.

There are disincentives to overbuilding solar, wind, tidal, wave, and other passive energy collectors. If we overbuild, the lower output from suboptimal production is still enough to meet demand. But, under normal conditions we will have far more power than we can use.

We already have periods of time where power prices go negative: generators are forced to pay to dump excess power. This melts the return on their investment, and stifles further rollout.

We can justify overbuilding such sources if we can adjust our demand to meet whatever we can supply. That means turning on additional loads when the sun shines, and turning off loads when the wind stops blowing.

Data centers can be put on highly variable rate plans that are at or even below costs during ideal generation conditions, and wildly expensive during suboptimal generation conditions. Data centers on such plans will halt processing when power is overly expensive, and only draw on the grid when it is profitable to do so.

Data centers aren’t the only industry where this can be done, and this isn’t a novel concept. Steel mills operate overnight to increase the load on baseload generation like nuclear. Baseload generators need the daily demand “trough” as high as possible, and the “peak” as low as possible. They need the curve as flat as possible, so they offer incentives to heavy industrial consumers to shift their demand. As we continue to shift to passive collectors instead of traditional generation, we need to reverse these old demand shaping practices to match the capabilities of new generation methods.