> compared to the energy a single shipping tanker, long-haul truck, car, or even microwave compute energy is microscopic. I don’t think that ever is a good arg

> compared to the energy a single shipping tanker, long-haul truck, car, or even microwave compute energy is microscopic. I don’t think that ever is a good argument. Efficiency improvements likely will come from thousands of small steps, and it’s not a given (but still fairly likely) that the cheapest (in kWh decrease/$ spent) improvements can be made by cutting down on the largest spenders. The good news is that data centers have become much more efficient over time. The bad news is that demand has about kept up for that. Data centers used about 1% of global electricity in 2018 (https://www.science.org/doi/10.1126/science.aba3758 https://www.science.org/doi/10.1126/science.aba3758, https://datacenters.lbl.gov/sites/default/files/Masanet_et_al_Science_2020.full_.pdf https://datacenters.lbl.gov/sites/default/files/Masanet_et_a...)