BTC's actual electricity consumption is only 0.15% of the world's electricity generation?

By Tyler Bain

Translated & edited by Sherrie

In recent years, there have been many claims that Bitcoin and the miners who secure the network via the SHA-256 proof-of-work consume excessive amounts of energy. But what data are these claims based on? Do the sources’ calculations use flawed or reasonable methods and assumptions? How much electrical energy has the Bitcoin network used historically?

Methods and Misconceptions

Due to the large, globally distributed topology of the Bitcoin network, the amount of electricity and energy consumed by miners cannot be fully verified, but rather must be estimated . A number of reputable sources have weighed in on energy consumption over the past few years, attempting to estimate Bitcoin’s network energy consumption in a more sober, data-based manner:

University of Cambridge, Judge Business School (JBS)

The International Energy Agency (IEA)

Electric Power Research Institute (EPRI)

Coin Center

CoinShares

Marc Bevand

Hass McCook

Alex de Vries

Myself

Estimation methods seem to fall into two broad categories: economic methods based on financial assumptions and physics-based methods based on engineering principles . At BTC2019, we conducted a comprehensive comparison and contrast of these two estimation methods.

When making sense of all of these annual usage estimates, it’s important to understand that electricity consumption is typically measured in two ways: instantaneously (power, watts, kilowatts, etc.) and as the same instantaneous power measurement integrated over time (energy, joules, kilowatt-hours (kWh)), etc.

The Problem of Network Energy Estimation Based on Economics

Economics-based approaches to estimating Bitcoin network energy consumption typically assume perfectly rational market behavior, which can be easily manipulated by incorrect assumptions about some input variables.

In theory, the Bitcoin mining industry is rational, profit-maximizing, and perfectly competitive: the marginal revenue from mining should tend to equal the marginal cost (MR = MC) . This means that over a long enough period of time, the market should find an equilibrium point where the energy cost of each unit of Bitcoin produced should be roughly equal to the market value of the Bitcoin at the time of minting. This calculation can be distilled to: "How much electricity can Bitcoin network miners afford?"

Typically, these types of estimates rely too heavily on a single volatile variable: the market trading price of Bitcoin. Here is a simple example of such an estimate:

Let's try this estimate, Bitcoin produces a block approximately every 10 minutes - 6 per hour, or 144 per day. Currently, a Bitcoin block contains a coinbase block reward of 6.25 BTC; that's 37.5 Bitcoins per hour, or 900 new Bitcoins rewarded to miners every day. As of this writing, Bitcoin's current market price is trading at approximately $10,750, which equates to an amount of $9,675,000 available for Bitcoin miners to generate power every day.

This energy is equivalent to the approximately 35.3 terawatt-hours of electricity used by Bitcoin miners each year , assuming the price of Bitcoin remains constant for a year and the average cost of electricity in the United States remains constant.

While this approach is overly dependent on the Bitcoin price, it also relies heavily on the assumed electricity costs for miners. The calculations and conclusions of this type of estimate can vary significantly, and could even be manipulated, depending on the assumptions used as inputs: energy costs ($/kWh) and Bitcoin price ($/BTC).

Here we use the average U.S. electricity cost of $0.10/kWh. However, in the U.S., electricity costs are actually seasonal, state to state, city to city, and in some cases, community to community. The same inconsistency exists in electricity costs around the world. This doesn’t even include broad industrial, commercial, or residential electricity rates, adding even more sources of error to these economic-based estimation techniques. In fact, this calculation’s heavy reliance on energy prices has another flaw: some highly creative miners have near-zero fuel costs when they mine excess energy that would otherwise be wasted, unavailable, or curtailed .

In my opinion, this quick exercise highlights why this economic-based approach to estimating is an oversimplification and comes with the following problems:

Bitcoin mining, computing power, and network energy consumption are less responsive to sudden price changes than these economic-based estimation methods .

Economics-based models claim that energy usage and network miner rewards are cut in half after the Bitcoin block reward halving cycle (i.e. every 210,000 blocks or approximately 4 years), and difficulty and proof-of-work-based data support this.

Such models assume a single, global average cost of energy ($kWh); electricity costs vary widely by region, season, and even by energy source.

This is probably an upper limit estimate.

Benefits of Physics-Based Network Energy Estimation

On the other hand, physics-based approaches to network energy estimation tend to be of a very strict “run the numbers” type that the Bitcoin community is used to.

These methods use independently verified on-chain difficulty, proof-of-work data, and original equipment manufacturer (OEM) published heat rate standards to more accurately estimate historical energy inputs into Bitcoin mining systems. The physical assessment attempt can best be described as “Bitcoin stoichiometric ratio unit analysis calculations:”

So let's try this type of estimation using the Bitcoin Proof of Work data and the data published by OEM. The Bitcoin network difficulty adjusts itself every 2016 blocks, or about every two weeks. This difficulty adjustment is intended to compensate for differences in the speed of block production, and thus fluctuations in the network hashrate .

This relationship between difficulty and proof of work allows us to derive an estimate of the network's computing power based on the block generation rate and the associated difficulty level. From the amount of work done at different difficulty levels over the past decade, we can roughly estimate the amount of SHA-256 hashes calculated by the Bitcoin network each year, as shown in the figure below, in terahashes per year (Th/year) or trillion hashes per year. We can also do the same exercise for daily data to produce a more nuanced calculation.

As of 2020, there are approximately 3934 yota hashes computed on the Bitcoin network, or approximately 3934 septillion hashes (“yota” and “septillion” are the largest prefixes in Science International (SI) to date, (10²⁴)).

Now that we have an estimate of the hashrate per year, we next have to compile the last 11 years of mining efficiency data to understand how much energy is required to produce this much work.

Here, it is important to understand the different types of mining equipment that have provided work for the Bitcoin blockchain over the years. Each era and year has significantly different proof-of-work efficiency characteristics that change the energy consumption value of the network over time. The Bitcoin Genesis block was generated by work from the CPU (central processing unit), and blocks were eventually taken by GPUs (graphics processing units), then FPGAs (field programmable gate arrays), and finally ASICs (application-specific integrated circuits). The Bitcoin network has developed at an astonishing rate .

Important note: Efficiency is defined as the useful work done for the energy consumed to complete that work (terahash/joules- Th/J). However, ASIC OEMs often quote a type of thermal rate specification, or the inverse of efficiency, showing the energy consumed versus useful work (joules per terahash- J/Th).

As you can see in the logarithmic scale chart below, the popularity of Bitcoin mining ASICs has been steadily declining each year for the past eight years, meaning that the network mining efficiency has been increasing .

Converting this data into an average heat rate per year (below) shows a similarly sharp decline throughout the history of Bitcoin mining. CPU, GPU, and FPGA benchmarks and published OEM power usage data were used to estimate the average network hashrate from 2009 to 2012. ASIC miners announced in 2020 are visualized above and below to show the continued decline in hashrate heat rate, but they are discarded from the energy estimates as they are not yet publicly available.

So, now that we have compiled all the necessary data (annual hashrate and annual hashrate heat rate), let’s combine them with an engineer’s attempt at Bitcoin mining energy stoichiometry:

Simply multiply the work done per year (terahash/year) by the estimated annual heat rate of the miners on the system (joules/terahash) and you get an estimate of joules/year. We will convert joules/year to kWh/year (1kWh equals 3.6 million joules) and the following chart shows the annual energy estimate.

However, this physics-based estimation method also has some problems:

The number of active miners by efficiency level is unknown, and this physics-based model assumes that all miners on the market have equal participation in the launch year.

The model also uses a step function of the annual heat rate data as input. This annual data will change suddenly on the first day of each year, and the gradual decline in heat rate as old miners gradually retire and new miners start working will be more realistic.

It assumes that older miners retire after a year, which is also unlikely since the life cycle of equipment is currently two years or more.

This is probably a lower bound type estimate.

Comparing different network energy estimates

Where do these annual energy consumption estimates fall in the context of the aforementioned calculation attempts? Interestingly, our two calculations, even with completely different methodologies and all the shortcomings discussed above—the economics-based estimate (35.3 TWh) and the physics-based estimate (40.17 TWh)—are numerically very similar . They also fall within the range of various other popular estimates made by noteworthy individuals, entities, and institutions, as shown in the table below. All of these estimates are fairly similar, which lends credibility to the variety of different estimators and the various methodologies and different assumptions used.

It is worth noting that the Bitcoin network hashrate (EH/s) appears to be starting to decouple from the general annual energy (TWh/year) estimate trend. This could be due to a drop in the heat rate of SHA-256 ASIC mining rigs if the estimate is based on physics, or due to the halving and price stagnation if the estimate is based on economics.

The charts above show a snapshot of annual energy estimates at the time of publication (TWh/year), but some of these sources (Cambridge University [C-BECI] and Alex de Vries [D-BECI]) actually published these annual estimates on daily charts several years ago. This gets back to the earlier discussion about energy vs. energy: logic should prevent plotting annual energy estimates on a daily axis.

Regardless, I think it’s worth comparing these published estimates to our own using a more continuous time series of data going back to the end of 2017 (previous market all-time high). The economic and physical calculations, Cambridge’s estimates, and Digiconomist’s results are all fairly similar in time, again adding some peer scrutiny and validity to these different estimation techniques.

Our estimates above appear to agree well with various other daily-interval annual energy estimates, so they have been averaged together to create a composite Bitcoin Energy Index (CBEI), shown below in TWh/year. Each estimate has different assumptions, levels and sources of inaccuracy, so a combination of them may be more accurate. This composite estimate (CBEI) has just recently retested the 60 TWh threshold for Bitcoin's total annual network energy consumption.

How does this composite energy index compare to the hashrate of the Bitcoin network over time? Around the beginning of 2019, the CBEI showed a similar decoupling phenomenon, with hashrate and energy continuing to rise, energy consumption remaining relatively stable, and ASIC heat rates and Bitcoin mining incentives having contracted.

Interestingly, snapshot Bitcoin consumption estimates are often extrapolated for the full year, expressed as TWh/year of energy, with no supporting temporal data or evidence. Daily network power estimates are much more palatable than all of these annual energy consumption estimates plotted on a daily chart. What is wrong with the chart is a startling chart error that causes a lot of misinterpretation of the data: annual energy estimates plotted on a daily axis. I have therefore taken the liberty of converting these daily interval estimates into a daily power estimate graph to correct for the aforementioned chart error that causes misinterpretation of the data.

I present the Composite Bitcoin Power Index (CBPI) compiled from the D-BECI and minimums, C-BECI maximums, minimums and estimates, and the economics and physics based estimates we presented above .

This CBPI composite estimates Bitcoin’s instantaneous power usage, expressed in watts, a unit of electricity. The CBPI recently peaked at nearly 7.58 GW, or about 1.21 GW, the equivalent of about six DeLorean time machines.

CBPI in the Environment

Such large energy values ​​are difficult to comprehend, especially in a yearly context, so let’s put these estimates into perspective with some quick comparisons:

The banking system consumes 650TWh/year

Gold mining 200 TWh/year

PC and console games 75 TWh/year

Bitcoin mining (CBEI) 60 TWh/year

Banknotes and coins 11 TWh/year

US Christmas lights 7TWh/year

Based on our estimates above, the Bitcoin network consumes approximately 40 to 60 TWh/year, or about 0.15% of the world’s annual electricity production (26,700 TWh) and only about 0.024% of the world’s total energy production (14,421,151 ktoe) . (A ktoe is also a unit of energy: 1,000 tons of oil equivalent, or 11.36 megawatt hours.)

Therefore, Bitcoin’s energy consumption today is only what many people consider to be a major problem: just a small fraction of the ever-growing human energy consumption. An interesting solution to this problem was proposed by Nikola Tesla a century ago. As recently as September 2020, a study claimed that nearly 76% of the Bitcoin network is powered by clean energy. Moreover, remember that once Einstein discovered the equivalence of mass and energy and humans harnessed the energy stored in atoms, the energy that has powered human progress has become abundant.

The content is for reference only and is not intended as investment advice. Please assume all risks

All rights reserved. Reproduction without permission is strictly prohibited.