Cambridge University just updated its methodology to compute the Cambridge Bitcoin Electricity Consumption Index (CBECI), an index aimed at estimating the network Power Usage.
Bitcoin electricity consumption: an improved assessmentExamining the underlying assumptions behind the Cambridge Bitcoin Electricity Consumption Index (CBECI) has led to its first significant revision since its launch in 2019 – a response to evidence indicating a periodic overestimation of electricity consumption.
They discovered that the Power estimate was constantly overrated as they underestimated the substitution rate of older miners. Newer miners have increased efficiency: either they churn more hash per J or consume less J to churn the identical hashes.
If a more significant part of the mining hardware than previously thought is new, then the average power consumption per hash is lower than previously thought.
|  |  |
| Over the last 10 years miner became almost 10x more efficient. | The increased efficiency estimate led to a sharp reduction in the CBECI index |
The comment is the following:
There are also less pronounced but still noteworthy differences in 2022 and 2023. The 2022 estimate was adjusted downward by 9.8 TWh, from 105.3 TWh to 95.5 TWh. To put this in perspective, the revised figure is comparable to the electricity consumption of countries like Belgium (83 TWh) or the Netherlands (113 TWh), [31] the energy use of tumble dryers in the US (108 TWh), [32] and renewable curtailment in China (105 TWh) [33]. In terms of global electricity consumption, it represented about 0.38%.[34] As for 2023, the year-to-date electricity consumption estimate has been revised from 75.7 TWh to 70.4 TWh.[/i]
I guess tumble-drying industry didn't show the same increase in efficiency.
