According to the article, this assumes mining operations are running full-power 24-7, which is hard to determine. The article does try to match increases in reg

According to the article, this assumes mining operations are running full-power 24-7, which is hard to determine. The article does try to match increases in regional power plant output with bitcoin mining center growth, but that's somewhat circumstantial with many confounding factors. The issue is: > "The best data on IT’s energy and emissions characteristics are closely held proprietary secrets among tech companies, which compounds these problems." (2021) "Does not compute: Avoiding pitfalls assessing the Internet's energy and carbon impacts", Koomey & Masanet, Joule https://www.cell.com/joule/fulltext/S2542-4351(21)00211-7 https://www.cell.com/joule/fulltext/S2542-4351(21)00211-7 Additionally, whenever you see these kind of estimates you really want the data acquisition method applied across the entire sector in a uniform fashion, to avoid someone trying to cook the data a bit in their favor. This is a very common problem in the energy sector, e.g. energy input/cost estimates for building and operating solar/wind/nuclear/gas/etc power plants tend to suffer from this agenda-driven skewing. Thus, you'd also want to see energy use estimates for all the AWS and Google etc data centers (NSA Utah?) side by side with the bitcoin miners and the high-frequency traders and the industrial manufacturing centers and households etc, and then check that the sum of estimates from different sources obtained via the same method is <= 100% of production and so on.