Content uploaded by Alex de Vries
Author content
All content in this area was uploaded by Alex de Vries on Mar 06, 2022
Content may be subject to copyright.
Commentary:
Bitcoin Boom: What Rising Prices Mean for the
Network's Energy Consumption
Alex de Vries,
1
,
2
,*
Joule Volume 5, Issue 3, P509-513, March 17, 2021. DOI: https://doi.org/10.1016/j.joule.2021.02.006
Abstract
As the price of Bitcoin rises, the negatively externalities associated with Bitcoin mining increase in kind.
This article shows how a simple economic model may be used to estimate the potential environmental
impact of Bitcoin mining for a given Bitcoin price. These estimates reveal that the record-breaking surge
in Bitcoin price at the start of 2021 may result in the network consuming as much energy as all data
centers globally, with an associated carbon footprint matching London’s footprint size.
1
School of Business and Economics, Vrije Universiteit Amsterdam, The Netherlands
2
Founder of Digiconomist, Almere, The Netherlands
*Correspondence: alex@digiconomist.net
1. Introduction
For the popular digital currency Bitcoin, the year 2021 started with the price of a single Bitcoin reaching
progressively higher records in quick succession. In less than one month’s time, the price of Bitcoin not
only broke the previous record (from December 2017) of almost $20,000 per coin but doubled it, surging
past $40,000 for the first time on January 8, 2021. In the same period, the demand for Bitcoin “mining”
devices also increased rapidly. Within the Bitcoin network, these devices are used to participate in the
process of creating new blocks for Bitcoin’s underlying blockchain, with successfully created blocks
providing a certain amount of bitcoins as a reward to the creator.1 In January it was reported that Bitmain,
one of the largest manufacturers of specialized Bitcoin mining devices, had sold out through August
2021 due to “overwhelming demand.”2
The increasing popularity of Bitcoin mining quickly sparked a fresh debate on the energy use—and the
resulting carbon footprint—of the Bitcoin network. Bitcoin mining devices require electrical energy to
function, and all devices in the Bitcoin network were already estimated to consume between 78 and 101
terawatt-hours (TWh) of electricity annually prior to the latest surge in the price of Bitcoin (Figure 1).
With a growing number of active machines, the network as a whole also requires more power to operate.
Figure 1: Historic Bitcoin energy consumption estimates and price development. The fluctuations in Bitcoin price (from
finance.yahoo.com/quote/BTC-USD) along with estimates of Bitcoin’s energy consumption (in annualized terawatt-hours)
from the Bitcoin Energy Consumption Index (BECI, bitcoinenergyconsumption.com) and the Cambridge Bitcoin Electricity
Consumption Index (CBECI, cbeci.org) since the start of 2017 to the end of 2020.
This article examines how to estimate the amount of electricity the Bitcoin network consumes for a
given Bitcoin price level. Additionally, the article explores the implications of this energy consumption
for the environment and broader economy and discusses how policymakers can limit the network’s
growing energy demands.
2. Predicting future Bitcoin network electricity consumption
In order to assess what a certain price level may mean for the energy hunger of the Bitcoin network, it
is crucial to first understand the relationship between these two variables. As noted, Bitcoin miners are
rewarded with bitcoins for successfully creating new blocks for Bitcoin’s blockchain. Per May 2020,
this reward amounted to 6.25 newly minted bitcoins per block, plus a variable amount depending on the
Bitcoin transactions that get processed in the block. In December 2020, transaction fees made up for
about 10% of the total overall miner income.3
These combined rewards provide a strong incentive to participate in the mining process, but the Bitcoin
protocol also purposely makes it difficult for miners to actually obtain these rewards. To successfully
create a new block for the blockchain, the block has to satisfy a set of requirements. For the miners, this
translates into a process of trial and error. As of January 11, 2021, it is estimated that all miners combined
make over 150 quintillion (blockchain.com/charts/hash-rate) attempts every second of the day to
produce a valid new block. Moreover, the Bitcoin protocol self-adjusts the difficulty of meeting these
requirements to ensure that, on average, only one block is created every 10 minutes.1 All of the mining
devices in the network are constantly competing with each other to be the first to produce a valid new
block. With the chance of success being random, a miner’s share of the total available miner income
will average to the proportional share of the total computational power in the network owned.
Bitcoin’s price directly impacts the value of the mined coins and therefore the amount of resources
miners can afford to spend on mining. Since Bitcoin mining is a competitive market, economic theory
suggests that it should cost a bitcoin to mine a bitcoin.4 If it costs any less than one bitcoin to mine a
bitcoin, miners can profit by adding more units of computational power to the network. The opposite is
also true as miners would be operating at a loss and start removing units of computational power from
the network if mining costs exceed one bitcoin. As previously noted by De Vries, “these market forces
drive the industry towards an equilibrium where firms will earn zero economic profit.”5
This dynamic makes it possible to estimate the amount of energy the network is consuming given a
certain amount of total miner income. While we should not expect a market to be in perfect equilibrium
because circumstances are constantly changing, we do expect total miner expenses to gravitate towards
the total amount of miner income as well. Knowing how miner expenses will develop does not
immediately give us an energy consumption estimate, but fortunately, the cost structure of mining is
extremely simplistic. A miner only requires a machine capable of mining (which can be a central
processing unit, graphic processing unit, or a specialized application-specific integrated circuit) and
electricity to run the device. For this reason, the primary cost components of participating in the mining
process are only hardware and energy. De Vries previously estimated that, in the long run, the share of
electricity costs in the total costs of mining is around 60%.5
If, on top of the previous assumption, we assume miners pay around USD 5 cents per kilowatt-hour
(kWh) of electrical energy on average, we can estimate the network’s energy requirement at a given
amount of miner income. With a Bitcoin price of $42,000 (Bitcoin’s all-time high per January 10, 2021)
and transaction fees comprising 10% of total miner income, miners will earn around $15.3 billion
annually (6.25 coins per block * 52,560 blocks per year / 0.9 * $42,000). With 60% of this income going
to pay for electricity at a price of USD 5 cents per kWh, the total network could consume up to 184
TWh per year (sensitivities to different assumptions are shown in Table 1); this is not far from the
amount of energy consumed by all data centers globally (200 TWh per year).6
Table 1: Bitcoin Annual Energy Consumption (TWh) Model Sensitivity Table. The table shows how various assumptions on
the share of electricity costs in the total costs of mining, as well as the average price of electricity (in USD per kWh), influence
the expected future energy consumption of the Bitcoin network at four different price levels. For every scenario, it is assumed
that fees make up 10% of the total miner income next to a fixed block reward of 6.25 bitcoins.
BTC (USD) BTC (USD)
32,000$ 0.03 0.04 0.05 0.06 0.07 42,000$ 0.03 0.04 0.05 0.06 0.07
50% 195 146 117 97 83 50% 256 192 153 128 110
55% 214 161 128 107 92 55% 281 211 169 141 120
60% 234 175 140 117 100 60% 307 230 184 153 131
65% 253 190 152 127 108 65% 332 249 199 166 142
70% 273 204 164 136 117 70% 358 268 215 179 153
BTC (USD) BTC (USD)
37,000$ 0.03 0.04 0.05 0.06 0.07 47,000$ 0.03 0.04 0.05 0.06 0.07
50% 225 169 135 113 96 50% 286 214 172 143 123
55% 248 186 149 124 106 55% 315 236 189 157 135
60% 270 203 162 135 116 60% 343 257 206 172 147
65% 293 219 176 146 125 65% 372 279 223 186 159
70% 315 236 189 158 135 70% 400 300 240 200 172
Price per kWh (USD)
Price per kWh (USD)
Price per kWh (USD)
Electricity cost
Electricity cost
Electricity cost
Electricity cost
Price per kWh (USD)
3. Adoption speed of miners and hardware capacity restrictions
Economic models do not indicate exactly when the network will reach an annual energy consumption
of 184 TWh, but miners have a strong incentive to add new mining devices as fast as possible. As the
total amount of computational power in the network grows, the proportional share of each individual
device declines over time. The amount of bitcoins a device is expected to mine therefore declines as
well, meaning that the first person to use a new device type will always reap significantly more rewards
than the last. In fact, it is mostly due to limits in the availability of hardware that the network does not
reach these levels of energy consumption overnight. Miners have some flexibility in the way their
devices are set up, so there is likely to be a swift impact from miners boosting (overclocking) their
device’s performance. However, this effect is equally likely to be limited as devices become unstable
and/or require additional cooling if pushed too far. The speed at which the network’s energy
consumption grows then depends on the availability of previously obsolete device models (that can
operate profitably again) and the rate at which newer device models can be produced. With Bitmain
having sold out up to the third quarter of 2021 by the start of the year, it may take several months, if not
longer, for the network’s energy consumption to reach the predicted level. Assuming that total miner
income stabilizes at $15.3 billion annually, the network will ultimately consume 184 TWh per year.
4. An absence of refunds ensures mining devices will be produced regardless of
price developments
As a vast number of new devices have now been ordered, a large part of the expected increase in energy
consumption may already be “locked-in”; this means that even if Bitcoin price falls by 25% or more (on
January 11, the Bitcoin price fell to $31,000 per coin before recovering to $35,000), the estimations of
the network’s future energy consumption do not necessarily have to revised by the same amount. This
“lock-in” effect is the result of Bitmain’s policy (and similar ones of other manufacturers) stating that
“cancellation or refund requests will not be entertained.”7 As such, upon submitting an order, the
equipment becomes a “sunk cost.” A sunk cost been incurred and can no longer be recovered; thus, it
should play no further role in the decision-making process of a rational economic agent. Just like any
other miner who already has devices running, those that have ordered and paid for their new devices
should base their decision to (continue to) mine on prospective costs, which primarily only include
electricity costs. For miners who only care about electricity costs, a Bitcoin price of $25,200 (assuming
USD 5 cents per kWh) is sufficient to sustain an annual electricity consumption of 184 TWh.
5. Limits to the predictability of Bitcoin’s future electricity consumption
There is no way of knowing the precise amount of future energy consumption that has been “locked-in”
in due to non-refundable orders, but with production lines guaranteed to run at maximum capacity for a
majority of 2021, it is unlikely to be an insignificant amount. In any case, we should be mindful of this
effect when considering a drop in Bitcoin price, though a scenario similar to that which followed the
Bitcoin price peak of 2017 may still occur. After the Bitcoin price got close to $20,000 for the first time
in 2017, the market experienced a rapid and steep decline. By the end of 2018, the value of Bitcoin had
dropped by more than 80% since the price peak. A Bitcoin price crash of a similar magnitude in 2021
would reduce the network’s energy consumption from the current estimates (Figure 1). If total annual
miner income falls to $3 billion (corresponding to a Bitcoin price of around $8,000 depending on the
transaction fee percentage at this point), this could only sustain an energy consumption of at most 60
TWh per year (assuming the entire amount is used to pay for electricity).
We also have to guard against overestimating the network’s future energy consumption. In the past, the
results of studies like these have been incorrectly extrapolated to make statements on the network’s
future energy consumption. The relationship between the network’s energy consumption and miner
income is mutual, meaning that miners require a certain amount of income to support a given level of
energy consumption. This article is limited to examining what the potential future implications are in
terms of energy consumption assuming stability in current miner income levels. Long-term forecasts
require more detailed modelling, and even the simple model presented in this article should account for
relevant changes in Bitcoin mining rewards. Bitcoin’s fixed block reward has been set to halve every
210,000 blocks (roughly every four years), which will happen again in 2024. While typical useful
lifetimes of mining devices are too short8 for this reward reduction to be relevant in this article, it is
something to be considered in future research along with the growing importance of variable transaction
fees.
Furthermore, future research may study the effect of network size on the average price of electricity. In
the short run, one may expect the average price of electricity to go up as the network grows as there is
a finite source of the cheapest electricity. However, the recent growth of cryptocurrency mining in
(sanctioned) countries like Iran9 (where miners can obtain oil-fueled electricity for less than USD 1 cent
per kWh9) suggests the possibility that new mining locations might drive the average price of electricity
down instead. In any case, the chosen rate of USD 5 cents per kWh (commonly used in research on the
topic in recent years) is likely still a conservative value for the network size considered in this article as
a recent survey found miners pay a global average of USD 4.6 cents per kWh.10
Lastly, it should be noted that this article generally assumes a market with rational agents; however, in
reality, miners may make decisions like (temporarily) operating at a loss if they speculate on (further)
price increases. This speculation may also increase the aforementioned “lock-in” effect if miners order
an excessive amount of mining devices in anticipation of a higher Bitcoin price, though this will also
depend on the limits of the production capacity of mining device manufacturers and their suppliers (see
next section).
6. Environmental impact and broader consequences
Having an estimate of Bitcoin’s future energy consumption also permits a ballpark estimate for the
network’s future carbon footprint. To this end, the work of Stoll et al.11demonstrated that Bitcoin mining
had an implied carbon intensity of 480-500 grams of CO2 per kWh (gCO2/kWh) consumed. Assuming
this number remains constant at 490 gCO2/kWh as the network’s energy demand increases, a total energy
consumption of 184 TWh would result in a carbon footprint of 90.2 million metric tons of CO2 (Mt
CO2), which is roughly comparable to the carbon emissions produced by the metropolitan area of
London (98.9 Mt CO2, according to citycarbonfootprints.info). This number may be higher or lower
depending on the locations chosen for Bitcoin mining. While fossil-fuel dependent countries like Iran
have recently gained popularity as mining sites,9 market miners may also try to leverage “greener”
sources of power.
In any case, the remainder of the cryptocurrency ecosystem would still have to be added to the total
environmental impact of the sector. Recent research found that other understudied cryptocurrencies such
as Ethereum and Litecoin added “nearly 50% on top of Bitcoin’s energy hunger.”12 Moreover,
specialized Bitcoin mining devices cannot be repurposed, potentially resulting in a substantial amount
of electronic waste once they become obsolete in several years’ time.8
On top of the environmental impact of cryptocurrency mining, the effects of the sector’s energy-hunger
may also spill over to other parts of the economy. Prior to the latest surge in Bitcoin price, it was already
reported that there was a global shortage of chips for an array of electronic devices.13 The economic
recovery following the COVID-19 crisis has led to increased consumer demand, resulting in chip
shortages and delays in manufacturing. These shortages are also affecting the production of (self-
driving) electric vehicles which will play an important part in meeting global goals for climate change,
as well as personal electronics required to work from home. Since the manufacturers of Bitcoin mining
devices need a substantial number of chips to produce these machines, this will only exacerbate the
shortage. To produce just one million units of Bitmain’s most powerful mining device (the Antminer
S19 Pro), which can consume 28.5 TWh of electrical energy annually, Bitmain would have to book a
full month of 7-nanometer (nm) capacity at its supplier, Taiwan Semiconductor Manufacturing
Company (together with Samsung currently the only companies capable of mass-producing 7-nm chips,
Figure 2).
Figure 2: Comparison of chip production capacity versus mining device chip production requirements. The figure shows the
monthly 7-namometer (nm) wafer capacity at the foundries of Taiwan Semiconductor Manufacturing Company (TSMC) and
Samsung (wccftech.com/amd-7nm-wafer-production-set-to-double-in-2h-2020-7nm-capacity-at-tsmc-currently-fully-
booked/). These companies are the only ones capable of mass-producing 7 nm nodes
(am.miraeasset.com.hk/insight/intel_7nm_delay/). The output required to produce one million Antminer S19 Pro devices is
calculated on the assumption that the required die size is 5 x 5 mm (in line with the Antminer S17 series
bitcointalk.org/index.php?topic=5307087.0) and that each 300 mm wafer yields 2,288 viable dies (caly-technologies.com/die-
yield-calculator/) on a defect density of 0.09 (anandtech.com/show/16028/better-yield-on-5nm-than-7nm-tsmc-update-on-
defect-rates-for-n5). Each Antminer S19 Pro requires 342 chips (chainnews.com/zh-hant/articles/648268505750.htm) and runs
on 3,250 watts.
It may also be a concern that a country like Iran has adopted cryptocurrency mining as a way to boost
revenues while its oil exports suffer from international sanctions. Cheap energy has lured in many
cryptocurrency miners, and the mining activity in Iran now represents 8% of the total computational
power in Bitcoin’s network.9 If Bitcoin is enabling Iran to circumvent economic sanctions, this may
pose a threat to international safety, as these sanctions were imposed to prevent the nation from
developing military nuclear capability.
7. Considerations for policymakers
Given the growing implications of the cryptocurrency mining industry, policymakers may feel
increasingly pressured to intervene. At a local level, this has already occurred in places such as Québec
(Canada) and Iran. In Québec in 2018, the Canadian power company Hydro-Québec and the independent
Québec Energy Board decided to imposed a moratorium on new cryptocurrency mining operations, after
a significant number of applications threated to destabilize the local grid.14 More recently, in January
2021, Iran decided to confiscate mining equipment as the country suffered from outages blamed on
cryptocurrency mining activities.9
Even though, in both examples, policymakers did not decide to take action because of environmental
concerns, the examples illustrate how policymakers may have multiple options in putting a halt to
cryptocurrency mining. While Bitcoin may be a decentralized currency, many aspects of the ecosystem
surrounding it are not. The competitive Bitcoin market drives miners to take advantage of economies of
scale in lowering costs, which also makes it harder for them to operate under the radar. Large-scale
miners can easily be targeted with higher electricity rates, moratoria, or, in the most extreme case,
confiscation of the equipment used. Moreover, the supply chain of specialized Bitcoin mining devices
is concentrated among only a handful of companies. Manufacturers like Bitmain can be burdened with
additional taxes like tobacco companies or be limited in their access to chip production. Policymakers
can be even be more restrictive to certain cryptocurrencies by barring them from centralized digital asset
marketplaces. While the latter has no direct impact on mining, it can influence the value of a digital
currency (and thus the associated mining rewards).
Policymakers should, however, be aware that there are also some boundaries to the policy options.
Ultimately, any laptop or computer is theoretically capable of participating in cryptocurrency mining,
and any location that has access to Internet and electricity may be used to host these devices. Miners
may simply move elsewhere under adverse policy decisions, or mining may become more decentralized
(and harder to control) when large-scale mining facilities or manufacturers of specialized devices are
severely restricted.
8. Conclusion
As the price of Bitcoin rises, the negatively externalities associated with Bitcoin mining increase in kind.
This article has shown how a simple economic model may be used to estimate the potential
environmental impact of Bitcoin mining for a given Bitcoin price. These estimates reveal that the record-
breaking surge in Bitcoin price at the start of 2021 may result in the network consuming as much energy
as all data centers globally, with an associated carbon footprint matching London’s footprint size.
Beyond these environmental impacts, the production of specialized mining devices may exacerbate the
global shortage of chips, which could impact the ability to work from home, the economic recovery
following the COVID-19 crisis, and the production of electric vehicles. The increasing popularity of
mining in countries like Iran may even threaten international safety. Policymakers are not completely
powerless to stop this from materializing, but drastic joint and coordinated actions may be required to
be effective.
9. References
1. Nakamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system.
2. Voell, Z. (2021). Bitcoin Mining Machine Shortage Worsens as Bitmain Sells Out Through August.
Coindesk.
3. Voell, Z. (2021). Bitcoin Miners Saw 33% Revenue Increase in December. Coindesk.
4. Debreu, G. (1987). Theory of value: an axiomatic analysis of economic equilibrium 19. Dr. (Yale
Univ. Press).
5. de Vries, A. (2018). Bitcoin’s Growing Energy Problem. Joule 2, 801–805.
6. Kamiya, G. (2020). Data Centres and Data Transmission Networks. Int. Energy Agency.
https://www.iea.org/reports/data-centres-and-data-transmission-networks.
7. Bitmain (2017). Can I edit or cancel my orders? https://support.bitmain.com/hc/en-
us/articles/115000382293-Can-I-edit-or-cancel-my-orders-
#:~:text=After%20an%20order%20is%20submitted,not%20pay%20for%20that%20order.
8. de Vries, A. (2019). Renewable Energy Will Not Solve Bitcoin’s Sustainability Problem. Joule 3,
893–898.
9. Crypto-miners take down Iran electric grids, prompting crackdown (2021). Arab News.
https://www.arabnews.com/node/1794836/middle-east.
10. Blandin, A., Pieters, G.C., Wu, Y., Eisermann, T., Dek, A., Taylor, S., and Njoki, D. (2020).
3RD GLOBAL CRYPTOASSET BENCHMARKING STUDY.
11. Stoll, C., Klaaßen, L., and Gallersdörfer, U. (2019). The Carbon Footprint of Bitcoin. Joule 3,
1647–1661.
12. Gallersdörfer, U., Klaaßen, L., and Stoll, C. (2020). Energy Consumption of Cryptocurrencies
Beyond Bitcoin. Joule 4, 1843–1846.
13. Jin, H., Busvine, D., and Kirton, D. (2020). Analysis: Global chip shortage threatens
production of laptops, smartphones and more. Reuters.
14. de Vries, A. (2020). Bitcoin’s energy consumption is underestimated: A market dynamics
approach. Energy Res. Soc. Sci. 70, 101721.
