Content uploaded by Katie Taladay
Author content
All content in this area was uploaded by Katie Taladay on Oct 29, 2018
Content may be subject to copyright.
comment
Bitcoin emissions alone could push global
warming above 2°C
Bitcoin is a power-hungry cryptocurrency that is increasingly used as an investment and payment system. Here
we show that projected Bitcoin usage, should it follow the rate of adoption of other broadly adopted technologies,
could alone produce enough CO2 emissions to push warming above 2 °C within less than three decades.
CamiloMora, RandiL.Rollins, KatieTaladay, MichaelB.Kantar, MasonK.Chock, MioShimada
and ErikC.Franklin
Leaders from 176 countries have ratified
the Paris Agreement, reached during the
Twenty-first Conference of the Parties
to the UNFCC (COP 21), to mitigate GHG
emissions and keep anthropogenic global
warming within 2 °C to avoid the impacts
of ever-more-catastrophic climate hazards
such as drought, heatwaves, wildfire, storms,
floods and sea-level rise, among others.
From 1860 to 2014, humanity emitted
~584.4 GtC from fossil fuel combustion,
industry processes and land-use change,
which has been mirrored by ~0.9 °C of
global warming (green line in Fig. 1a).
Temperature projections from 42 Earth
system models (ESMs) developed for the
recent Coupled Model Intercomparison
Project Phase 5 (CMIP5) under four
alternative emission scenarios show that
an additional 231.4 to 744.8 GtC would
push global warming across the 2 °C
threshold (Fig. 1a; the range represents
the 5th and 95th percentiles among
model projections, see Methods). Reducing
emissions to keep warming below 2 °C is
already regarded as a very difficult challenge
given the increasing human population and
consumption1 as well as a lack of political
will2. Then came Bitcoin.
Bitcoin is a decentralized cashless
payment system introduced in early 2009,
and it is now accepted by over 100,000
merchants and vendors worldwide3.
Each transaction paid for with Bitcoin
is compiled into a ‘block’ that requires a
computationally demanding proof-of-
work to be resolved, which in turn uses
large amounts of electricity4. Based on the
assumptions that 60% of the economic
return of the Bitcoin transaction
verification process goes to electricity,
at US$5¢ per kWh and 0.7 kg of CO2-
equivalent (CO2e) emitted per kWh,
Digiconomist5 estimated that Bitcoin
usage emits 33.5 MtCO2e annually,
as of May 2018. Foteinis6 repeated this
approach using emissions adjusted by a
broader life cycle of electricity generation
and found that for 2017, the global emissions
from Bitcoin and Ethereum usage were
43.9 MtCO2e. Compiling data on the
electricity consumption of the various
computing systems used for Bitcoin
verification at present and the emissions from
electricity production in the countries of the
companies that performed such computing,
we estimated that in 2017, Bitcoin usage
emitted 69 MtCO2e (s.d. = ± 0.4; see Methods).
Globally, ~314.2 billion cashless
transactions are carried out every year7,
of which Bitcoin’s share was ~0.033% in
20175. The environmental concern
regarding Bitcoin usage arises from the
large carbon footprint for such a small
share of global cashless transactions,
and the potential for it to be more
broadly used under current technologies.
Bitcoin usage has experienced an
accelerated growth (Supplementary
Fig. 1), which is a common pattern
during the early adoption of broadly
used technologies8. Should Bitcoin follow
the median growth trend observed in
the adoption of several other technologies
(Fig. 1b), it could equal the global total of
cashless transactions in under 100 years.
Yet, the cumulative emissions of such
usage growth could fall within the range
of emissions likely to warm the planet by
2 °C within only 16 years (red line in
Fig. 1b). The cumulative emissions of
Bitcoin usage will cross the 2 °C threshold
within 22 years if the current rate is similar
to some of the slowest broadly adopted
technologies, or within 11 years if adopted at
the fastest rate at which other technologies
have been incorporated (that is, the red
area in Fig. 1b). Projections in this analysis
assume that the portfolio of fuel types used
to generate electricity remains fixed at
today’s values (see Supplementary Table 3).
The future usage of Bitcoin is a topic of
considerable discussion. There is currently
considerable economic motivation for
companies to compute the proof-of-work
for each Bitcoin block (for example, the
latest block on 8 May 2018 (block 521819)
gave a reward of 12.5 bitcoins plus 0.1
bitcoins for transaction fees, with a total
monetary value of US$116,041 on that
date’s exchange rate; https://blockchain.
info) — the expected time needed to resolve
that proof-of-work is around 10 minutes.
However, Bitcoin is set up in such a way
that rewards should halve every 210,000
blocks, or approximately every 4 years (for
example, 50 bitcoins in 2008, 25 in 2012,
and so on). Over time, this could reduce
the motivation for companies to resolve the
computationally demanding proof-of-work
for each block, potentially overwhelming the
system and reducing general interest in the
use of Bitcoin. Alternatively, Bitcoin usage
may generate sufficient transaction fees to
support the system, which is how Bitcoin
was originally conceived.
Although we are unable to predict the fate
of Bitcoin, our analysis suggests that if its rate
of adoption follows broadly used technologies,
it could create an electricity demand capable
of producing enough emissions to exceed 2
°C of global warming in just a few decades.
Given the decentralized nature of Bitcoin
and the need to maximize economic profits,
its computing verification process is likely to
migrate to places where electricity is cheaper,
suggesting that electricity decarbonization
could help to mitigate Bitcoin’s carbon
footprint — but only where the cost of
electricity from renewable sources is cheaper
than fossil fuels. Certainly, high electricity cost
will push the development of more efficient
hardware. However, reducing Bitcoin’s carbon
footprint should not rest solely on some
yet-to-be-developed hardware but include
simple modifications to the overall system,
such as adding more transactions per block
or reducing the difficulty or time required
to resolve the proof-of-work — both of
which could result in immediate electricity
reductions for Bitcoin usage. Our analysis is
NATURE CLIMATE CHANGE | www.nature.com/natureclimatechange
comment
based on Bitcoin alone; however, the problem
of high electricity consumption is common
to several cryptocurrencies, suggesting that
any further development of cryptocurrencies
should critically aim to reduce electricity
demand, if the potentially devastating
consequences of 2 °C of global warming
are to be avoided. ❐
Online content
Any methods, additional references, Nature
Research reporting summaries, source data,
statements of data availability and associated
accession codes are available at https://doi.
org/10.1038/s41558-018-0321-8.
CamiloMora1*, RandiL.Rollins2,3,
KatieTaladay1, MichaelB.Kantar4,
MasonK.Chock5, MioShimada5
and ErikC.Franklin1,6
1Department of Geography and Environment,
University of Hawai‘i at Mānoa, Honolulu,
HI, USA. 2Department of Biology, University
of Hawai‘i at Mānoa, Honolulu, HI, USA.
3Pacic Biosciences Research Center, School of Ocean
and Earth Science and Technology, University of
Hawai‘i at Mānoa, Honolulu, HI, USA. 4Department
of Tropical Plant and Soil Science, University of
Hawai‘i at Mānoa, Honolulu, HI, USA. 5Department
of Botany, University of Hawai‘i at Mānoa, Honolulu,
HI, USA. 6Hawai‘i Institute of Marine Biology,
School of Ocean and Earth Science and Technology,
University of Hawai‘i at Mānoa, Kāne‘ohe, HI, USA.
*e-mail: cmora@hawaii.edu
Published: xx xx xxxx
https://doi.org/10.1038/s41558-018-0321-8
References
1. Raupach, M. R. et al. Proc. Natl Acad. Sci. USA 104,
10288–10293 (2007).
2. Kemp, L. Palgrave Commun. 3, 9 (2017).
3. Cuthbertson, A. Bitcoin now accepted by 100,000 merchants
worldwide. International Business Times (4 February 2015);
https://go.nature.com/2CcKFXs
4. De Vries, A. Joule 2, 801–805 (2018).
5. Bitcoin Energy Consumption Index (Digiconomist, 2017);
https://digiconomist.net/bitcoin-energy-consumption
6. Foteinis, S. Nature 554, 169 (2018).
7. Global Payment Systems Survey 2016 (World Bank, 2016).
8. Bass, F. M. Manag. Sci. 50, 1833–1840 (2004).
Acknowledgements
The authors wish to thank the numerous data providers
named in the supplements of this paper for making their
data freely available. We also thank SeaGrant Hawaii
for providing funds to acquire the computers used in
these analyses. This paper was developed as part of the
graduate course on ‘Methods for Large-Scale Analyses’
in the Department of Geography and Environment at the
University of Hawai’i at Mānoa.
Additional information
Supplementary information is available for this paper at
https://doi.org/10.1038/s41558-018-0321-8.
a
Years since introductio
nY
ear
Adoption (%)
Credit card
dit c
ca
a
d
d
e
rd
rd
rd
ar
c
c
a
ed
e
ar
a
a
a
a
a
a
a
Electricity
Elect
Electric
lectricity
ectri
Ele
El
e
c
c
ectricity
ctricity
c
y
Dishwasher
s
bc
Temperature change relative to 1860 (°C)
0
1
2
3
4
5
6
0
20
40
60
80
100
0
1,000
500
1,500
2,000
2,500
3,000
020406080 2020 2040 2060 2080
3000
1000 500 1,000 1,500 2,000 2,500 3,000
Cumulative anthropogenic CO2e emmisions
since 1860 (GtC)
COP21 target of 2°C
Bitcoin cumulative CO2e emissions (GtC)
Fig. 1 | Carbon emissions from projected Bitcoin usage. a, Current and projected trends in global average temperature as a function of cumulative man-made
carbon emissions. Narrow lines depict the projections of individual ESMs, while the thick lines indicate the multimodel median. The dashed line represents the
COP 21 target of 2 °C global warming, and the grey shaded area represents the CO2e emissions among ESMs at which such a threshold is crossed (values are for
the 5th and 95th percentiles of all model projections). b, Trends in the adoption of broadly used technologies. Data are available for the United States, and used
here as a reference. The red shaded area indicates the margins of the upper and lower quantiles, and the red line is the median tendency among technologies (see
Methods). Grey lines indicate trends for each of the technologies (see Methods). c, Cumulative emissions from Bitcoin usage under the average growth rate of
technologies that have been broadly adopted as shown in b. The grey shaded area indicates the carbon emissions above which warming exceeds 2 °C.
NATURE CLIMATE CHANGE | www.nature.com/natureclimatechange
comment
Methods
Amount of CO2e needed to surpass the 2 °C COP
21 target. e 2015 Paris Agreement set the goal to
limit global warming to 2 °C. To quantify the amount
of CO2e emissions required to warm the planet by
2 °C (that is, CO2 emissions plus the CO2e emissions
of other anthropogenic GHGs), we used temperature
projections from ESMs and the driving CO2e emissions
of such models (see below). For each ESM, we
estimated the CO2e emissions at which 2 °C warming
was reached (the x axis value at which each narrow
line in Fig. 1a intercepted the 2 °C warming threshold)
and grouped those results to estimate the 5th and 95th
percentiles (grey box in Fig. 1a). We also collected data
on ‘observed’ temperature change and CO2e emissions
from 1860 to 2014 as validation for model projections,
to quantify current warming and cumulative emissions,
and estimate the CO2e emissions needed to surpass the
COP 21 target. e observed temperature change and
CO2e emissions since 1860 are shown as a green line
in Fig. 1a (temperature data from NOAA-CIRES 20th
Century Reanalysis V2c9, CO2e emissions data from
the Carbon Dioxide Information Analysis Center10).
e observed and projected cumulative CO2e emissions
are very similar over the time period for which they
overlap (see blue and green lines in Fig. 1a), however,
they used dierent methods and emission sources.
Fossil fuel emissions, industrial processes and land-use
change are the main anthropogenic GHG contributing
to current warming (Supplementary Fig. 2), and are all
in common to both databases used in this analysis.
Temperature projections. We analysed global annual
average surface air temperature from 42 ESMs developed
for CMIP5. We used the historical experiment, which
for all models includes the period from 1860 to 2005 and
Representative Concentration Pathways (RCPs) 2.6, 4.5
and 8.5, which include the period from 2006 to 2100.
The historical experiment was designed to model recent
climate (reflecting changes due to both anthropogenic
and natural causes) whereas the RCP scenarios represent
contrasting mitigation efforts between rapid GHG
reductions (RCP 2.6) and a business-as-usual scenario
(RCP 8.5). For each model, under each experiment,
we calculated the difference in the global average
temperature between every year in the time series and
1860. For any given experiment, global annual averages
from all models at any given year were used to estimate
the multimodel median temperature change for that
year (thick lines in Fig. 1a). Temperature change for each
model and the multimodel median are shown in Fig. 1a.
CO2e projections. Although CO2 is the primary GHG
contributing to the total anthropogenic radiative
forcing (changes in the Earth’s energy balance due
to human activities), other anthropogenic agents
also contribute to the warming trends projected
by ESMs (such as methane, aerosols and so on).
During the timeframe of this study, volcanic and
solar radiative forcings have remained reasonably
constant and proportionally very small in relation
to the anthropogenic forcing (Supplementary Fig. 2),
indicating that they contribute minimally to the
warming trends from ESMs, and thus were not
considered in this analysis. For the purpose of
standardization, we calculated the CO2e emissions for
the radiative forcing from all anthropogenic activities
used by the historical and RCP experiments. For this
purpose, we obtained CO2 emissions, their radiative
forcing and the total anthropogenic radiative forcing
under each experiment (data from Meinshausen
et al.11). We estimate the CO2e emissions for the total
anthropogenic radiative forcing as the amount of CO2
required to achieve the total anthropogenic radiative
forcing given the ratio of actual CO2 emissions and the
actual CO2 radiative forcing. As an example, from 1860
until 2005 the historical experiment shows that the
cumulative CO2 emissions from fossil fuels, cement, gas
flaring, bunker fuels and land-use amounted to 453.247
GtC and a resulting radiative forcing of 1.675 W m−2,
whereas the total anthropogenic radiative forcing
was 1.840 W m−2. Thus, the CO2e emissions for that
total anthropogenic radiative forcing were estimated
at 497.984 GtC (1.840 × (453.247/1.675)). Projected
anthropogenic CO2e emissions under different
experiments are plotted against temperature change
from the different models in Fig. 1a. Note that CO2e
emmisions are given in weight units of carbon, which
can be converted to units of carbon dioxide (CO2),
simply multiply these estimates by 3.667.
Amount of CO2e produced by Bitcoin usage. Any
given transaction using Bitcoin is compiled into a
block requiring a proof-of-work to be resolved, with
the winning company/pool being awarded a certain
amount of new bitcoins plus any extra transactions
fees. The CO2e emissions from this procedure emerge
primarily from the electricity demands of the hardware
used and the location where the electricity is generated.
To assess the carbon footprint of the global Bitcoin
Network, using as reference the year 2017, we used
the following approach. We started by compiling a
list of current hardware suitable for Bitcoin and their
energy efficiencies (hashes per electricity consumed,
Supplementary Table 1). To each block mined in
2017 (data from https://blocktrail.com), we assigned
a random hardware from Supplementary Table 1 and
multiplied the number of hashes required to solve the
block by the energy efficiency of the random hardware;
this returned the amount of electricity consumed to
solve the given block. Note that the available data for
mined blocks include their difficulty, which can be
used to estimate the number of hashes as (hashes =
difficulty × 232; equation from O’Dwyer and Malone12).
For each block mined in 2017, we also collected data on
the company claiming the given block, and searched for
their country/countries of operation (Supplementary
Table 2). For the resulting list of countries, we collected
data on the types of fuels used for electricity generation
(Supplementary Table 3), and using average standards
of CO2e emissions for the generation of electricity with
those types of fuels (under a life-cycle carbon approach,
Supplementary Table 4), we estimated the total carbon
emission equivalents to produce electricity in those
countries (Supplementary Table 2). By multiplying the
electricity consumption of every block in 2017 by the
electricity emissions in the country where the proof-of-
work was likely to be resolved, we were able to estimate
the total CO2e emissions for computing every block
in 2017. Summing the CO2e emissions from all blocks
in 2017 yielded the Bitcoin emissions in that year.
This approach was repeated 1,000 times to capture the
variability in the random selection of hardware, and
this is described as the s.d.
Projected usage and carbon emissions from Bitcoin.
The likely future of Bitcoin has been broadly discussed
in online forums with opinions ranging from it being
a case of boom and bust, or alternatively, an early stage
in a ‘new industrial revolution’. We studied the carbon
emissions of Bitcoin if it follows the adoption trends
of other broadly used technologies. For this, we used
the incorporation rate of 40 different technologies
for which data are readily available: the automatic
transmission, automobile, cable TV, cellular phone,
central heating, colour TV, computer, credit card,
dishwasher, disk brakes, dryer, e-book reader, electric
power, electric range/burners, electronic ignition,
flush toilet, freezer, home air conditioning, household
refrigerator, Internet, landline phone, microcomputer,
microwave, nitrogen oxides pollution controls
(boilers), podcasting, power steering, radial tires, radio,
refrigerator, Real Time Gross Settlement adoption,
running water, shipping container port infrastructure,
smartphone, social media, stove, tablet, vacuum,
washer dryer, washing machine and water heater (data
for the USA from ref. 13, credit card data from ref. 14).
Data include the year and percentage of population
using the given technology. The first year of usage for
each technology was set to one, to allow comparison
of trends among technologies (narrow grey lines in
Fig. 1b). For each year, we calculated the average and
lower and upper quantiles of per cent population using
all technologies and applied a logistic model to such
trends (the red line and red shading in Fig. 1b).
The projected trends of technology usage adoption
were used to estimate likely Bitcoin usage assuming a
global total of ~314.2 billion cashless transactions.
We used only cashless transactions that are likely
to occur in places where infrastructure is already
in place for the usage of Bitcoin as a reference
(for example, we do not assume that Bitcoin will
replace transactions using fiat currency). The CO2e
emissions of projected Bitcoin usage were estimated
using the CO2e emissions for Bitcoin transactions
in 2017 as a reference. We randomly sampled blocks
mined in 2017 until their total number of transactions
were equal to the projected number of transactions,
then we added the CO2e emissions from computing
such randomly selected blocks. The approach was
repeated 1,000 times.
Code availability. Raw code used for this study
are publicly available online at https://github.com/
moracamilo/Bitcoin/.
Data availability
The authors declare that all data supporting the
findings of this study are available within the article, its
Supplementary Information files and at https://github.
com/moracamilo/Bitcoin/.
References
9. NOAA-CIRES 20th Century Reanalysis V2c (NOAA, accessed 28
February 2018); https://go.nature.com/2CDGaXg
10. Global Carbon Budget (Global Carbon Project, accessed
28 February 2018); http://www.globalcarbonproject.org/
carbonbudget/17/data.htm
11. Meinshausen, M. et al. Climatic Change 109, 213–241 (2011).
12. O'Dwyer, K. J. & Malone, D. Bitcoin mining and its energy
footprint. In 25th IET Irish Signals & Systems Conference 2014 and
2014 China-Ireland International Conference on Information and
Communities Technologies http://doi.org/cvqm (IEEEE, 2014).
13. Technology Adoption by Households in the United States (Our
Word in Data, accessed 28 February 2018); https://go.nature.
com/2NCnUyj
14. Consumer Credit and Payment Statistics (Federal Reserve Bank
of Philadelphia, accessed 28 February 2018); https://www.
philadelphiafed.org/consumer-nance-institute/statistics
NATURE CLIMATE CHANGE | www.nature.com/natureclimatechange