Long- and Short-term Prediction of Bitcoin Energy Consumption

Abstract

Bitcoin”s (BTC) mining process utilizes the proof of work (PoW) concept as the consensus algorithm., which consumes electricity. As a result., there are concerns regarding its sustainability and carbon emission. If one wants to estimate these, they first should have an estimation of the network's energy consumption which is hard to calculate due to its decentralized nature. To do so., two methods are conceivable. First, considering electricity price and miner's revenue (electricity costs are considered as a part of total revenue., then divided by electricity price to obtain electricity consumption), second, using hash rate along with the efficiency of mining devices. The latter is utilized in this study for short- and long-term predictions. In the short-term, recurrent neural network (RNN) is used, whereas in the long-term logistic functions are applied. It has been concluded that the ultimate energy consumption of the Bitcoin network, which will happen around 2025, is in the range of 40.4 to 73.41 TWh annually and its estimated value is 58.56 TWh.