Title: Machine learning the carbon footprint of Bitcoin mining
Authors: Hector Calvo-Pardo - University of Southampton (United Kingdom) [presenting]
Tullio Mancini - University of Southampton (United Kingdom)
Jose Olmo - University of Southampton (United Kingdom)
Abstract: Building on an economic model of rational Bitcoin mining, we measure the carbon footprint of Bitcoin mining power consumption using feedforward neural networks. We find associated carbon footprints of 2.77, 16.08, and 14.99 MtCO2e for 2017, 2018, and 2019 based on a novel bottom-up approach, which (i) conform with recent estimates, (ii) lie within the economic model bounds while (iii) delivering much narrower prediction intervals, and yet (iv) raise alarming concerns, given recent evidence (e.g., from climate-weather integrated models). By 2024, conservative point forecasts based on an exponential trend found for the network hash rate suggest a carbon footprint of 132.01 MtCO2e, similar to the combined annualized 2019 greenhouse gas emissions of Belgium (100 MtCO2e) and Denmark (32 MtCO2e). We demonstrate how machine learning methods can contribute to non-for-profit pressing societal issues, like global warming, where data complexity and availability can be overcome.