The Climate Impact of ICT
ISMB/ECCB 2021, July 2021
Adrian Friday
School of Computing and Communications
Lancaster University
a.friday@Lancaster.ac.uk @gulliblefish

Team effort…
Dr Kelly Widdicks Dr Bran Knowles Prof Gordon Blair Prof Adrian Friday
Charlotte Freitag Prof Mike Berners-Lee

This talk has 3
parts:
1. Review of estimates of ICT’s
impacts by leading authors
2. What might the future impact
be?
3. What can you do about it?
unsplash/ levi-xu-dOhJtfXJZfw

• ICT estimated at
4000 TWh by
2030
• US electricity =
3660 TWh in
2020
• Global PV + Wind
~1.2 TWh in
2019
unsplash/ nasa-Q1p7bh3SHj8

Climate change, net 0? Much to do…
https://www.instituteforgovernment.org.uk/explainers/net-zero-target

What are ICTs energy/CO2e impacts, and
where are we heading?
Creation, recycling, supply
chain externality
Impacts we’ve not considered
Devices, networks & data
centres etc. energy use
unsplash/ olena-sergienko-W2BI02QOncg/ lars-kienle-IlxX7xnbRF8/ dogherine-z1_z4hX9WhM/ arno-senoner-yqu6tJkSQ_k/ johannes-plenio-5_9inhy4NSE/ zhang-fengsheng-gUKv_yrQ_tk/ https://www.upsbatterycenter.com/blog/ev-makers-pushing-hard-for-ethical-cobalt/#prettyPhoto
Energy and how cleanly it’s
generated
vs

Why is this not trivial?
https://www.carbontrust.com/resources/briefing-what-are-scope-3-emissions
Scope 1 covers direct
emissions
• Lack of openness, frequently
undeclared, time varying
01
Scope 2 covers indirect
emissions from the generation
of purchased electricity, steam,
heating and cooling
• Changes with time and location
(constant conversion factors typically
used) on short and long timescales
02
Scope 3 includes all other
indirect emissions that occur
in a company’s value chain
• Hard to account for completely,
challenging for ICT products due to
their complexity (c.f. simple foods)
03

Available on
arxiv
https://arxiv.org/abs/2102.026
22

Methodology
1. Literature searches for peer-reviewed
studies on ICT’s carbon footprint
2. Supplemented by interviews with key
analysts, i.e.
• Anders Andrae (Huawei Technologies)
• Jens Malmodin (Ericsson)
• Lotfi Belkhir (McMaster University)
3. Small World Consulting ongoing
research with UK & global ICT companies

When do
we use
energy?
WHEN WE COMPUTE OR
COMMUNICATE WE USE
ENERGY
ENERGY USE IS THUS
PROPORTIONAL BUT NOT
WHOLLY LINKED TO
‘COMPUTATIONAL WORK’
COMPUTERS AND
NETWORKS ALSO NEED
POWER JUST TO BE
‘ON’/IDLE
COMPUTE GENERATES HEAT,
COOLING ALSO TAKES
POWER
EFFICIENCY COMES FROM
DOING MORE WITH LESS
POWER (E.G. MOORE’S LAW)
TO MAKE THE
INFRASTRUCTURE IN THE
FIRST PLACE

1.8-2.8%
of global
greenhouse
gas emissions
Where are we?
1,087 1,107
1,306
887
1,860
3,634
-
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
Malmodin (2020) Belkhir &
Elmeligi (2018) -
2020 minimum
Belkhir &
Elmeligi (2018) -
2020 maximum
Andrae & Edler
(2015) - 2020
Best case
Andrae & Edler
(2015) - 2020
Expected case
Andrae & Edler
(2015) - 2020
Worst case
Greenhouse
gas
emissions
in
MtCO2e
User devices Data centres Networks TVs

Considerable differences
User
devices,
37%
Data
centres,
41%
Networks,
22%
User devices,
57%
Data
centres,
18%
Networks,
24%
User
devices,
32%
Data
centres,
33%
Networks,
35%
Andre & Edler, 2015 (2020 best case) – 623 MtCO2e Belkhir & Elmeligi, 2018 (2020 mean
case) - 1,207 MtCO2e
Malmodin, 2020 (2020 estimate) – 690 MtCO2e

Why do these
estimates vary?
•1.8-2.8%
•2.1-3.9%
of global greenhouse gas
emissions
• Data sources and decisions
• Truncation error
• Assumed future activities
and behaviours in society
unsplash/ sarah-kilian-52jRtc2S_VE

We ask:
Will energy
efficiency
improvements in
ICT continue?
Will efficiency
gains reduce ICT’s
overall carbon
footprint?
Will ICT’s
emissions stabilise
due to saturation
in demand?
Is data traffic now
independent of ICT
emissions?
Is ICT enabling
carbon savings in
other industries?
Will renewable
energy
decarbonise ICT?
unsplash/ science-in-hd-joGs6hU_EKE

1. Will energy efficiency improvements in ICT
continue?
Yes, there is scope
for energy efficiency
improvements in ICT
to continue
No, energy efficiency
improvements are
slowing down and
are not fully utilised
• Moore’s
law
• PoE in
data
centres
• Shift to
mobile
devices

Yes, there is scope
for energy efficiency
improvements in ICT
to continue
No, energy efficiency
improvements are
slowing down and
are not fully utilised
1. Will energy efficiency improvements in ICT
continue?
• Quantum
limits
• Hotter
CPUs =
more
cooling
• Does
new
replace
old?

2. Will efficiency gains reduce ICT’s overall
carbon footprint?
Yes, energy efficiencies
in ICT can reduce ICT’s
carbon footprint
No, energy efficiencies
drive growth in ICT’s
carbon footprint due
to rebound effects
• PoE has
improved
• Data
centres
grown yet
remained
at ~1%
global
demand
[Masanet,
2020]

Yes, energy efficiencies
in ICT can reduce ICT’s
carbon footprint
No, energy efficiencies
drive growth in ICT’s
carbon footprint due
to rebound effects
2. Will efficiency gains reduce ICT’s overall
carbon footprint?
• Demand
for more
devices
• Carbon
footprint
undeniably
growing
• Jevons
paradox!

3. Will ICT’s emissions stabilise due to
saturation in demand?
Yes, the world will
become saturated with
ICT and this will
stabilise ICT’s emissions
No, innovation will
prohibit saturation in
ICT
• 5.7bn
mobile
subscribers
by 2023
[Cisco,
2020]
• Limited
available
attention
[Malmodin]

Yes, the world will
become saturated with
ICT and this will
stabilise ICT’s emissions
No, innovation will
prohibit saturation in
ICT
3. Will ICT’s emissions stabilise due to
saturation in demand?
• Strong
commercial
incentives
• IoT, AI
• The global
south…

4. Is data traffic now independent of ICT
emissions?
Yes, ICT emissions
are largely
independent of
data traffic
No, data traffic
drives ICT growth
and the associated
emissions
• Always on
infrastructure
• Provisioned
for peak
demand
• More
demand =
more
infrastructure

Yes, ICT emissions
are largely
independent of
data traffic
No, data traffic
drives ICT growth
and the associated
emissions
4. Is data traffic now independent of ICT
emissions?
• A&E, B&E
agree data
drives ICT
growth
• Innovative
services
drive
growth
(e.g.
video)

5. Is ICT enabling carbon savings in other
industries?
Yes, ICT is enabling
carbon savings in
other industries
No, ICT drives
carbon emissions in
other industries due
to rebound effects
• ICT could
save 9.1
GtCO2e in
2020
through
efficiency
[GeSI 2015]
• ICT enabling
decarboniza
tion of
energy

Yes, ICT is enabling
carbon savings in
other industries
No, ICT drives
carbon emissions in
other industries due
to rebound effects
5. Is ICT enabling carbon savings in other
industries?
• Global
emissions
still
climbing
• ICT
enabled
rebound
effects

Yes, renewable
energy will
decarbonise ICT
No, renewable
energy is not a
silver bullet
6. Will renewable energy decarbonise ICT?
• 0.1
kgCO2e/kWh
vs. 0.63kg
CO2e/kWh
(global
energy mix)
• 86%
reduction in
emissions

Yes, renewable
energy will
decarbonise ICT
No, renewable
energy is not a
silver bullet
6. Will renewable energy decarbonise ICT?
• Availability
& share of
renewables
• Also
embodied
footprints
• Resources
(e.g. silver)

ICT emissions, assuming the 2020 level (adjusted for truncation error) remains stable until 2050, and
global CO2 emissions reduced in line with 1.5°C under scenario SSP2-19.
ICT’s future share of emissions?
40.93
32.28
23.63
17.58
11.52
7.65
3.78
0
5
10
15
20
25
30
35
40
45
2020 2025 2030 2035 2040 2045 2050
Carbon
emissions
in
GtCO2
Global CO2 emissions ICT's GHG emissions stable at 2020 level
3.2% 5.6% 11.5% 35.1%
4%
21%
42%
57%
72%
81%

Three reasons to believe ICT’s
environmental impacts will increase

What can you
do?
Scope 3
Amortise embodied cost by extending the useful life of
equipment; push for supply chain decarbonisation;
divestment; organisational practice (e.g. how grants are
costed)
Scope 2
Push for ‘real’ green energy, time computation with
abundant low emission energy; demand response (helps
integrate renewables)
Scope 1
Minimize ‘direct’ footprint: strive for efficiency, reduce
waste, e.g. with energy proportionate algorithms;
minimise fossil fuel use

Hard questions
Limits? Should there be a limit and how could this
work? How would we prioritise global compute?
Should a limit reduce over time?
Policy? In what ways do the epistemological structures
of computational biology and bioinformatics research
need to change to lower overall impacts?
Accountability? How can we provide better
transparency to drive evidence led policy in this area?
Value? Emissions isn’t everything, there are questions
of broader societal value, but then that shouldn’t mean
‘do nothing’… we need to act now!

The Climate Impact of ICT
ISMB/ECCB 2021, July 2021
Thank you.
Prof. Adrian Friday
School of Computing and Communications
a.friday@lancaster.ac.uk
Freitag, C., Berners-Lee, M., Widdicks, K., Knowles, B.,
Blair, G., and Friday, A. 2020. The climate impact of
ICT: A review of estimates, trends and regulations.
https://arxiv.org/abs/2102.02622

Socio-Digital Sustainability Team
Kathy New
PhD Student
Dr Kelly Widdicks
Lecturer
Adam Tyler
PhD Student
Dr Oliver Bates
Senior Researcher
Dr Ally Gormally
Lecturer
Prof Adrian Friday
Professor
Dr Janine Morley
Researcher
http://www.lancaster.ac.uk/scc-sds/
Matt Marsden
PhD Student
Christina Bremer
PhD Student
Iman Hussain
PhD Student
Dr Christian Remy
Senior Researcher
Dr Carolynne Lord
Senior Researcher