Chapter I

Digitalization and
environmental
sustainability
As the evolving digital economy continues to create both opportunities and challenges
for trade and sustainable development, the Digital Economy Report 2024, for the first
time, turns its attention to the environmental implications of digitalization.
Against a backdrop of multiple environmental crises and the digital solutions leveraged
to tackle them, it is increasingly important to consider how to reduce the environmental
footprint of digitalization itself.
This chapter outlines the importance of exploring the implications that arise at the
nexus of digitalization and environmental sustainability, and stresses the need to
consider the entire life cycle of digital products.
The chapter also notes that many developing countries face a particular challenge,
as they are less equipped to harness digitalization to mitigate environmental risks
while also being exposed to many of the potential environmental costs associated
with digitalization.
© AdobeStock_.shock
 Chapter I
Digitalization and environmental sustainability

A. The digitalization and

environmental sustainability nexus

1. An area in need of more electronic commerce (e-commerce) and the

digital economy on inclusive and sustainable
attention development. They covered in particular
Sustainable development is a vital priority the increasing significance of new digital
for the United Nations and the global technologies, platformization and digital
community, articulated in successive United data (UNCTAD, 2019a, 2021a). These
Nations summits and in the 2030 Agenda reports highlighted the accelerated pace
for Sustainable Development. Sustainable of digitalization, leading to a continuously
development implies economic and social changing nature of the digital economy, Sustainable
development that is consistent with the accompanied by widening digital and development
protection of planetary boundaries – data divides and important environmental
implications. They emphasized that
implies
avoiding irreversible impacts on the
environment – and with intergenerational bridging these divides and developing economic
equity, the idea that today’s development balanced frameworks for global and social
should not jeopardize the opportunities of governance of data and digital platforms development
are essential for ensuring inclusive and
future generations (World Commission on
sustainable development outcomes.
consistent with
Environment and Development, 1987).1
In this context, three issues have become
planetary
Digital transformation of the world economy
critical: the consumption of natural boundaries
and society is taking place in parallel with
resources, the impact of climate change
growing concerns related to the depletion of
(especially resulting from fossil fuel
raw materials, water use, air quality, pollution
consumption) and pollution. The cost of
and waste generation, which are all linked
failure in these three areas threatens all
to planetary boundaries, including climate
aspects of sustainability and the future
change. Managing digital transformation
health of planet Earth.
will greatly influence the future of humanity
The Rio Declaration from the first Earth and the health of the planet. This report
Summit urged all stakeholders – explores the interconnectedness of
Governments, businesses and civil rapid digitalization and the urgent need
society – to recognize that “environmental to foster environmental sustainability
protection shall constitute an integral part against a backdrop of growing inequality
of the development process and cannot and vulnerabilities, such as increasing
be considered in isolation from it” (United socioeconomic disparity, environmental
Nations, 1993: Principle 4). Consequently, degradation and geopolitical tensions.
economic development that is not It explores ways to achieve economic
environmentally sustainable will also prove prosperity that are compatible with planetary
to be unsustainable economically. boundaries and intergenerational equity.

Recent editions of the Digital Economy The topic is timely, not to say long overdue,
Report have looked in depth at the as policy discussions on the environment
implications of the rapid growth of and digitalization in the context of

1
The concept of planetary boundaries assesses human impact on nine dimensions of the planet relative to the
time of pre-industrialization. This helps to determine the stability of the Earth system, which should support
the well-being of people and the planet. Recent research has shown that globally, six out of nine boundaries
have already been crossed (Richardson et al., 2023).

3
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

sustainable development have evolved fifteenth session of the United Nations

separately for too long. Soon after the Conference on Trade and Development
second Earth Summit in Rio de Janeiro in in 2021, member States included climate
2012, critical voices emerged, suggesting change, environmental degradation and
that the Summit had failed to recognize the digital divide among the most important
the relationship between information and development questions (UNCTAD, 2021b).
communications technologies (ICTs), the This evolution of the mandate is illustrative
Internet and sustainability, all of which are of the changing landscape of challenges
crucial elements of sustainable development faced by countries today, as well as their
policy (Souter and MacLean, 2012).
ever-increasing interconnectedness beyond
The 2030 Agenda for Sustainable
trade, which requires a policy approach
Development, which was adopted in
that breaks out of regulatory silos.
2015, did not take a cross-cutting view
of the role attributed to digitalization. There are growing references to the
To date, The word “digital” is in fact mentioned “twin transitions”, alluding to the need to
shifts towards only in reference to the “digital divide”. enable, on the one hand, the transition to
low-carbon In the Paris Agreement, adopted in the a more digital economy and, on the other,
and digital same year as the 2030 Agenda, ICTs were the transition to a low-carbon economy
(Muench et al., 2022; UNCTAD, 2023a). To
technologies primarily highlighted as a means to share
information, knowledge and good practices date, shifts towards low-carbon and digital
were considered technologies have been considered as
among countries and stakeholders; to
in parallel, yet enable the development of low carbon parallel processes. In reality, they are closely
they are closely energy technologies; to improve energy intertwined within the broader transition
intertwined efficiency and support various adaptation of the global economy. Moving towards
within the efforts, such as early warning systems more environmentally sustainable economic
(United Nations Framework Convention activities needs digital tools to become
broader on Climate Change (UNFCCC), 2016). more efficient and resilient in the long term.
economic Similarly, the 2023 outcome document At the same time, while digitalization is
transition of the twenty-eighth session of the a means to an end, it will need to be as
Conference of the Parties to the United environmentally sustainable as possible to
Nations Framework Convention on Climate avoid adding to environmental risks.
Change recognizes the importance of digital Moreover, the minerals and metal inputs
transformation and increased access to needed for digitalization and the expansion
technologies to achieve the goals set out
of renewable energy sources are largely
in the Paris Agreement (UNFCCC, 2023)
the same, creating competing demands
– without taking into consideration
and significantly influencing international
its direct environmental impact.
trade and geopolitical dynamics.
Digitalization has continued to evolve at
It is important to work towards ensuring
a high speed and, from an environmental
that no one is left behind as the world
perspective, is offering new solutions but
also obstacles to sustainability (box I.1). transitions towards a more digital and
The relationship between digitalization environmentally sustainable future. A
and environmental sustainability in all its just, low-carbon and digital technology
dimensions is starting to receive more transition requires an integrated approach
attention in policy debates with a view to to sustainable development, which brings
maximizing potential gains from digitalization, together social progress, environmental
while mitigating environmental harms and protection and economic success into a
facilitating sustainability. In the Bridgetown framework of democratic governance. This
Covenant, the outcome document of the extends to the human rights context.

4
 Chapter I
Digitalization and environmental sustainability

Box I.1
The rapidly evolving nature of digitalization

When assessing the trade and development interface between digitalization and environmental
sustainability, it is essential to acknowledge the dynamic nature of digital technologies and their
applications (German Advisory Council on Global Change, 2019; UNCTAD, 2019a; Global Enabling
Sustainability Initiative and Deloitte, 2019). Continuing digitalization creates many new opportunities
for harnessing data and digital technologies to foster trade and development and mitigate adverse
development and environmental impacts. At the same time, the importance of ensuring that the
digital ecosystem is as environmentally sustainable as possible increases further.

Higher speed. The increased use of the Internet and online services partly reflects the recent
accelerated progress in high-speed online transmissions. This opens up opportunities for
developing new digital applications, such as digital government and financial services, social
media and online purchases. The digital delivery of services, both domestically and internationally,
relies on greater bandwidth to support high-quality video calls or streaming. The extent to which
different parts of the world can seize such opportunities still varies greatly.

Shift to the cloud. Cloud computing is a key element of the evolving digital landscape (UNCTAD,
2013). It enables users to access scalable and flexible data storage and computing resources as
well as to stream video and music. The imagery of the intangible “cloud” can be misleading; cloud
computing is well anchored on the ground through hardware, networks, storage and services
needed to deliver computing as a service. A defining feature of cloud storage is the transfer of
large volumes of data to third party-owned data centres, often controlled by a small number of
very large companies (UNCTAD, 2021a).

Platformization. Digital platforms, acting as intermediaries and infrastructure of the digital

economy, are uniquely placed to capture and extract extensive data from online actions and
Internet
interactions on the platforms. The expansion of digital platforms is directly linked to their capacity of things
to collect, analyse and monetize digital data, with businesses ranging from Internet search and connections
social media to cloud storage and e-commerce (UNCTAD, 2019a). The growing role of platforms to grow to
has led to strong market concentration, dominated by a small number of global digital platforms
from the United States and China (UNCTAD, 2021a). Platforms increasingly control all parts of
35 billion by
the global data value chain, including data collection, data transmission (installing and owning 2028, mainly in
cables and satellites), data storage (cloud and hyperscale data centres) and data analysis (machine Asia-Pacific
learning and artificial intelligence (AI)). This pivotal role in the digital economy requires high levels
of responsibility and better platform governance.

Exponential data growth and real-time sensing. The surge in Internet use, improved cloud
infrastructure and the growth of global platforms have significantly boosted interconnectedness
among people, machines and the planet. Data generated in real time from improved
interconnectedness can help to address various development challenges, including in agriculture,
energy, health, home appliances and transportation by analysing (near) real-time data. For instance,
the “Internet of things” (IoT), through sensing, automation and cloud computing, is expected to
expand from 13 billion connections in 2022 to over 35 billion by 2028, particularly in Asia and the
Pacific, and will employ various devices (sensors, meters, etc.) to collect and transmit timely data
(Global System for Mobile Communications Association (GSMA), 2023a). At the same time, this
increasing connectivity spurs the demand for digital devices, digital networks and services that
support the IoT. This translates into more demand for natural resources, more use of water and
energy, more greenhouse gas emissions from the production and use of the devices, and more
waste to handle at the end of life.

Cognitive changes. The exponential increase in data generation is amplifying the importance of
big data analytics, machine learning and AI. Global corporate investment in AI (including private
investment, mergers and acquisitions, public offerings, and minority stakes) surged from an
estimated $15 billion in 2013 to $189 billion in 2023.a Concerns are mounting that powerful AI
systems may be evolving too fast and too far, as labs compete to develop ever more sophisticated

5
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

solutions, with unknown consequences and limited regulation.b New generative AI solutions – such
as Bing, ChatGPT, Dall-e, Ernie, Gemini (formerly Bard), Gigachat, Midjourney, SenseChat and
Tongyi Qianwen – have been met with strong interest, although long-term user numbers remain
uncertain.c While offering new experiences and value to users, AI applications are computationally
costly, energy- and equipment-intensive and generate large quantities of waste (Strubell et al.,
2019).
Towards virtuality. Another new feature driven by digitalization, higher computing power and
speed is increased “virtuality”, seen in the growing use of augmented reality and virtual reality.
Virtual reality offers a three-dimensional online environment that can be entered by using a
dedicated headset connected to a computer or game console. Augmented reality shows the
real world enhanced by computer-generated items, such as graphics, enhancing the real world
by superimposing computer-generated information (Shen and Shirmohammadi, 2008). Such
technologies can enable users to access objects and experiences regardless of their physical
location. Increased adoption of virtual reality may have both positive and negative environmental
impacts, depending on the inputs required and whether it replaces or complements existing
polluting behaviour.
Distributed ledger technology. Blockchain and other distributed ledger technologies allow
multiple parties to engage in secure transactions without any intermediary. The technology
underpins cryptocurrencies and holds potential for many domains relevant to developing countries,
such as digital identification, securing property rights and disbursing aid.d Blockchain technology,
specifically cryptocurrencies that rely on proof-of-work as their mechanism to validate transactions,
demands significant resources, notably electricity and processing power. The International Energy
Agency (IEA) estimates blockchain energy demand to increase by nearly 50 per cent between
2022 and 2026 (IEA, 2024). How growth in adoption of distributed ledger technology is handled
will have environmental implications in the future, and will depend on adoption rates and efficiency
improvements.
Source: UNCTAD.

a
See https://aiindex.stanford.edu/report/.
b
See https://futureoflife.org/open-letter/pause-giant-ai-experiments/.
c
See https://www.washingtonpost.com/technology/2023/07/07/chatgpt-users-decline-future-ai-
openai/.
d
See UNCTAD (2021c) for blockchain applications in support of the Sustainable Development Goals.

United Nations General Assembly Secretary-General, Our Common Agenda,

resolution 76/300, on the human right to and its proposal for a global digital compact
a clean, healthy and sustainable and the Inter-Agency Task Team for the
environment, adopted in July 2022, Global Accelerator on Jobs and Social
recognizes that this right is “related to Protection for Just Transitions (United
other rights and existing international law” Nations, 2021a). All this is expected to
(paragraph 2) and affirms that its promotion feature prominently in the Summit for the
“requires the full implementation of the Future in September 2024.2 As part of these
multilateral environmental agreements broader efforts, new initiatives have been
under the principles of international launched. In particular, in 2022, the Coalition
environmental law” (paragraph 3). for Digital Environmental Sustainability
(CODES) developed an “Action Plan for a
The digitalization and environmental Sustainable Planet in the Digital Age” at the
sustainability nexus is to some extent Stockholm+50 Conference (CODES, 2022).
reflected in the report by the United Nations Nonetheless, considerably more attention

2
See https://www.un.org/en/common-agenda/summit-of-the-future.

6
 Chapter I
Digitalization and environmental sustainability

needs to be given to the intersection Discussions of sustainable consumption and

between the rapidly evolving digital production have increasingly focused on the
economy and environmental sustainability, desirability of a more circular economy to
and its implications for trade and reduce environmental impacts. Most goods
development. The processes involved today are produced in an essentially linear
are all complex and difficult to regulate. model that begins with the extraction of
raw materials and passes from processing,
design, manufacturing, distribution and use
2. Comprehensive life to disposal. As will be discussed later in
cycle assessments this report, the digital economy still remains
highly linear. A more circular digital economy
The relationship between digitalization would seek to reduce, reuse and recycle
and environmental sustainability is digital devices and infrastructure, including
multifaceted and can be explored from by extending their lifespan. This can be
various perspectives. There is a need to achieved through sharing, rental or donation;
consider the extent to which digitalization maintenance and repair; resale and
complies with the “planetary guardrails” redistribution; as well as remanufacturing
and refurbishing. These activities can
(Haum and Loose, 2015), related to the
help reduce emissions caused by mineral
climate, nature, soils and oceans. Key
extraction and processing, manufacturing
environmental impacts are linked to energy
or final disposal. Ideally, transitioning to
use and greenhouse gas (GHG) emissions,
a more circular digital economy would
protecting habitats, soil and water resources
help achieve at least equivalent levels of
and reducing air pollution and waste. All of
economic growth and business profitability
these are closely linked to the concept of
to those in the linear economy but with
the Anthropocene age which reflects how greater environmental sustainability.
human activity has a long-lasting impact on
the environment (The Economist, 2023). The ability to identify significant

Digital solutions are often seen as key for

environmental opportunities and risks Digital
arising from digitalization is hampered by
achieving Sustainable Development Goal 12 a lack of agreement on what specifically
solutions
which relates to sustainable consumption constitutes the ICT sector (typically, end- are often
and production. For example, they can user devices, network infrastructure and seen as key
reduce the environmental impacts of data centres; figure I.1) and associated for achieving
consumption and economic development services and what needs to be included
through the use of smart devices and when measuring environmental impact.
Sustainable
by enhancing production efficiency This together with a lack of relevant data Development
(World Economic Forum (WEF) and makes it challenging to develop targeted Goal 12 on
policy responses to minimize the
PwC, 2020; Technopolis and Institut für sustainable
ökologische Wirtschaftsforschung, 2024). environmental impacts of digitalization.
This raises a critical question of how to
consumption
To better understand these impacts,
better leverage digitalization to achieve and production
researchers use life-cycle assessments
sustainability, for which improved data (LCAs) to evaluate the environmental
and measuring approaches are needed. impacts of a product or a service
Hence, the main focus of this report is throughout its entire life span.3 International
how to make digitalization and activities standardization for LCA methodology,
related to the ICT sector more sustainable. particularly ISO 14040 and ISO 14044,
Unless adequately addressed, their has laid the foundation for a formalized,
negative impacts are likely to increase as robust and reliable approach to measuring
digitalization expands across all sectors. environmental impacts. LCA is not limited

3
LCA can be applied in different areas and sectors. Recent UNCTAD work has investigated the trade impact
from manufacturing (UNCTAD, 2021d) and of plastic substitutes on the environment (UNCTAD, 2023b).

7
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

to any single environmental indicator, such environmental impact from end-user

as GHG emissions, but can encompass devices and ICT infrastructure (networks
multiple criteria. For instance, the LCA- and data centres), highlight potential
based product environmental footprint environmental trade-offs and assess the
methodology developed by the Joint sustainability potential of substituting
Research Centre of the European Union digital for non-digital technologies (Hilty
identifies 16 environmental impacts that can and Aebischer, 2015; Itten et al., 2020).
be assessed through LCA, with a strong link
Given data availability, LCAs in the digital
to various Sustainable Development Goals
economy typically focus on GHG emissions.
and to planetary boundaries (Joint Research
However, this focus has limitations. Such
Centre, European Commission et al., 2019).4
partial analysis can lead to production
For digital transformation, LCA can help processes that are environmentally
to identify stages with important suboptimal, potentially leading to

Figure I.1
The ICT sector is made up of three parts: Networks, data centres and
end-user devices

Satellites
Global Internet Network
connectors
infrastructure
Internet
exchange
Data centres points

Core, metro, Wired and

edge networks wireless access
networks
Cloud
data centres

Regional
(edge)
data centres

End-user
In-house
data centres devices

Source: UNCTAD, based on Pohl and Hinterholzer (2023).

4
The Joint Research Centre suggests the following impact categories for a comprehensive environmental
footprint of consumption in relation to Sustainable Development Goals: Goal 3 (good health and well-being):
human toxicity, cancer; human toxicity, non-cancer; particulate matter; photochemical ozone formation;
ionizing radiation; Goal 6 (clean water and sanitation): impacts due to water use, ecotoxicity, eutrophication;
Goal 13 (climate action): climate change; impact due to resource use; Goal 14 (life below water): eutrophication
marine and freshwater; ecotoxicity; Goal 15 (life on land): impact due to land use; eutrophication terrestrial;
acidification; impact due to mineral and metal resource use; ozone depletion (Joint Research Centre, European
Commission et al., 2019).

8
 Chapter I
Digitalization and environmental sustainability

“greenwashing”. For instance, electricity a. Direct effects Focusing only on

use by data centres can be reduced by
upgrading servers more frequently, yet
GHG emissions
Direct (or first order) effects result from digital
this leads to more electronic waste. devices and ICT infrastructure throughout
can result in
their life cycle, spanning raw material environmentally
This report examines the three phases
of the life cycle of end-user devices extraction and processing, manufacturing, suboptimal
and ICT infrastructure and seeks to transportation for distribution, use and production
assess the environmental footprint of the end-of-life phase (ITU, 2014). The processes,
digitalization in view of the interconnected direct effects on resource use, energy potentially
global challenges of digitalization, use, GHG emissions and water and soil
climate change, trade and development.
leading to
pollution constitute their “environmental
Typically, within such an assessment: footprint” (Hilty and Aebischer, 2015).
“greenwashing”
• The production phase covers the
extraction of raw materials, including As noted above, it is important to
minerals and metals, and their complex consider other direct environmental
refining process; the assembly of impacts beyond GHG emissions (Mewes,
different components of devices and 2023). For example, extraction of raw
ICT infrastructure; and their subsequent materials and handling of waste during
transportation for global distribution. production and end-of-life phases can
This phase is discussed in chapter II, have significant environmental impacts,
with a particular focus on the intensity
such as soil contamination and dangers
of mineral and metals use, as well as
to biodiversity (table I.1). Additionally, in
geopolitical, trade and development
extraction, production and cooling of
considerations of their value chain.
digital devices and infrastructure significant
• The use phase considers environmental
amounts of water are used throughout
effects generated by operating and
the life cycle (Olivié-Paul, 2022).
using end-user devices, transmission
networks and data centres. Particular GHG and water footprints, while
attention is given to energy use, GHG
interconnected, raise different issues. In
emissions and water consumption. This
one sense they go together: the more ICT
is the focus of chapter III. Chapter V
devices are built and deployed, the more
looks at a specific use case, namely the
environmental impact of e-commerce. energy is used, the more GHGs are emitted,
and the more water is consumed. There can
• The end-of-life phase at the
also be a negative correlation. For example,
treatment of digital technologies
there is often a trade-off between the energy
after use, and the importance of
moving towards a more circular and the water used for cooling. Moreover,
economy, is discussed in chapter IV. while GHG emissions are particularly relevant
for climate change, the water footprint
relates to freshwater scarcity (increasingly
3. Direct and indirect a consequence of climate change) and
effects possible impacts on biodiversity. Unlike the
The three phases of the life cycle of global impact of GHG emissions, which
digitalization have different environmental can be offset in various places, negative
impacts. In order to assess the overall impacts on water supply are highly location-
possible effects, it is important to distinguish specific. Saving water in one area cannot
between direct and indirect effects.5 compensate for the local impact in another.

5
For more details, see Berkhout and Hertin (2001); Bieser and Hilty (2018); Bremer et al. (2023); Coroamă et
al. (2020); Hilty and Aebischer (2015); Horner et al. (2016); Pohl et al. (2019); Williams (2011).

9
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

Table I.1
Direct environmental effects of digital devices and infrastructure

Life cycle phase Type of environmental impact Digital device example: Smartphone

Production Raw materials extraction. Impacts on GHG Materials, fossil fuels and water needed for
▲ emissions and the local environment from transport and processing of raw materials for
extracting and processing raw materials to smartphone production.
make digital devices and infrastructure.
Production and transportation. Impacts Energy and water to produce and ship a
on GHG emissions and water use from smartphone to market.
manufacturing and transporting digital devices
and infrastructure.
Use Impacts on GHG emissions and water Energy needed to use a phone; energy and
▲ use from operating digital devices and water needed to power the underlying digital
infrastructure. infrastructure such as data centres, mobile or fixed
broadband.
End-of-life Impacts on GHG emissions, pollution of water Negative: Energy to dispose of the smartphone;
▲▼ and soil from reuse, recycling and end-of-life impacts on water and soil from recycling and
treatment of digital devices and infrastructure. disposal of components.
Positive: Proper reuse and recycling of devices and
components reduces future negative impacts from
raw material extraction.

Source: UNCTAD, adapted from Bremer et al. (2023); Pohl et al. (2019); Horner et al. (2016).
Notes: A red upward pointing arrow indicates a negative effect (increasing environmental impact); a green
downward pointing arrow indicates a beneficial effect (avoided impact). A red upward pointing arrow next to a
green downward pointing arrow means that the net effect can be either positive or negative.

b. Indirect and rebound effects potential for significant GHG emissions

reduction through the effective use of digital
Indirect (or second and higher order) effects technologies in different industries (box I.2).
describe other environmental impacts from
the use of digital technologies and services The International Panel of Climate Change
in different sectors of the economy, thus (IPCC) acknowledges the potential role
going beyond the direct footprint of the ICT of digital technologies, including sensors,
sector. These can be both environmentally IoT and AI to mitigate climate change,
beneficial and harmful. Positive indirect improve energy management, boost energy
effects that decrease emissions or other efficiency and promote the adoption of
environmental harms are sometimes referred low-emission technologies while creating
to as “enabling effects”, “abatement” or economic opportunities (IPCC, 2022a).
“avoided emissions” (Bremer et al., 2023). Despite this, take-up of digitally enabled
Data-driven digital technologies can production processes remains limited.
be powerful tools to mitigate negative Industry estimates suggest that effective
environmental footprints from economic use of digital technologies could significantly
activities. For instance, they can enable real- reduce global GHG emissions (Global
time monitoring and adaptation in resource Enabling Sustainability Initiative and Deloitte,
use (“optimization effect”). Substituting 2019). The same study optimistically
physical goods and travel with digital concluded that digitally induced reductions
alternatives can enable decarbonization and of emissions could be nearly seven times the
dematerialization within some production size of the growth in total carbon emissions
and consumption patterns (“substitution from the ICT sector over the same period.
effect”). Various studies highlight the Researchers also recognize the potential

10
 Chapter I
Digitalization and environmental sustainability

of supply chain and business model energy use and optimizing transport and
innovations to reduce the environmental construction (Rolnick et al., 2023).
impact of the economy (Blanco et al.,
2022; Parida et al., 2019; Wang, 2017). To date, various studies have been unable
Furthermore, machine learning offers to confirm the potential for environmental
mitigation potential by improving monitoring, gains from digitalization through anticipated

Box I.2
Opportunities for digital technologies to mitigate carbon emissions

Digital technologies can be applied across sectors with a view to reducing negative environmental Empirical
effects. This box provides examples of potential opportunities including in global value chains,
transportation, construction, agriculture and energy. However, in most areas, empirical evidence
evidence
on actual gains realized remains limited. on actual
Digital technologies can be used to make global value chains more environmentally sustainable by environmental
enhancing productivity, reducing environmental impacts of current production and consumption
gains from
modes, introducing new, more environmentally friendly technologies and eco products, and
enhancing the diffusion of business models based on circular economies (UNCTAD, 2023c). digitalization
The use of advanced robotics, three-dimensional printing, sensors and wireless technologies remains
can enable automation and the decentralization of tasks to potentially reduce emissions from
transport. Digitalization can also help to better monitor environmental standards, optimize logistics,
limited
boost operational efficiency and thereby reduce carbon emissions and energy consumption. Data
processing technologies, such as big data analytics, cloud computing and AI, further contribute
to environmentally sustainable production processes.
The transport sector accounts for about one-quarter of global energy-related GHG emissions;
varying from below 3 per cent in some least developed countries (LDCs) to more than 30 per cent
in high-income countries, although growth rates in transport-related emissions have been larger
in developing regions in recent years.a Smartphone applications can help to optimize routes and
vehicle efficiency (GSMA, 2019). However, the effect of circular and shared economy initiatives as
well as other aspects of digitalization is uncertain (IPCC, 2022a). Dematerialization could reduce
demand for transport services, while an increase in e-commerce with priority delivery may raise
demand for freight transport.
Another major contributor to emissions is the buildings and construction sector. In 2021, this sector
accounted for 37 per cent of energy and process-related CO2 emissions.b Digital technologies may
be leveraged to reap benefits from optimizing energy use through automation in smart buildings
and cities (Global Enabling Sustainability Initiative and Deloitte, 2019).
The agricultural sector accounts for 10–12 per cent of global anthropogenic (human-generated)
GHG emissions. Precision agriculture, improved weather prediction and the IoT in smart water
infrastructure can notably reduce CO2 emissions and improve irrigation efficiency (Global
Enabling Sustainability Initiative and Deloitte, 2019; Technopolis and Institut für ökologische
Wirtschaftsforschung, 2024). At the same time, precision farming has been found to only slightly
reduce pesticide use (Bovensiepen et al., 2016).
According to the IPCC (2022a), improvements in energy efficiency from digital technologies can
help to reduce energy demand in all end-use sectors. This includes material input savings and
increased coordination. For example, smart appliances and energy management can effectively
reduce energy demand and associated GHG emissions without reducing service levels; similarly,
district heat systems can use waste heat from nearby data centres.
Source: UNCTAD, based on cited sources.

a
See IPCC (2014, 2022a).
b
See https://www.unep.org/news-and-stories/press-release/co2-emissions-buildings-and-
construction-hit-new-high-leaving-sector.

11
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

efficiency and substitution gains from ICT anticipated beneficial environmental effects
(Clausen et al., 2022; Schultze et al., 2016). (Digitalization for Sustainability, 2022).
In fact, one review found no significant shift In the case of e-commerce, for example,
towards sustainable energy consumption buying a product online can be more
levels in any sector after introducing digital energy efficient under certain conditions
tools (Lange et al., 2020). Similarly, the than driving to a physical store to buy the
IPCC (2022a) stresses that potential gains same product, thereby reducing GHG
may be reduced or counterbalanced by emissions. But if the convenience of online
“rebound effects”, leading to increased shopping encourages increased purchasing
demand for and use of goods and services. frequency, volume and returns that are not
always resold, any initial emission reductions
Rebound effects in digitalization, where initial
may be diminished or counterbalanced.
positive impacts are offset by increased
Higher order indirect effects, or societal
demand and use, can undermine the
effects, stem from behavioural changes
benefits of more efficient goods and services
triggered by the interaction of direct and
(Vickery, 2012; Coroamă and Mattern, 2019;
indirect effects, including rebound effects,
Technopolis and Institut für ökologische
as digital technologies are widely adopted,
Indirect Wirtschaftsforschung, 2024). Rebound
leading to changes in lifestyles and value
effects can occur for the same good or
environmental service because the efficiency gains made
systems (Hilty and Aebischer, 2015; Horner
effects could it cheaper or more convenient to consume
et al., 2016; Pohl et al., 2019; Williams,
2011). For example, digitally enabled
be significantly more of it. The money or time saved through
teleworking reduces transport-related
greater than digitally induced efficiency, however, can
energy use but increases energy use in the
also lead to the increased consumption of
the direct places in which the telework is performed.
other goods and services, two phenomena
environmental often referred to as “income effect”
It may induce secondary changes such as
living locations (for instance, relocating
footprint (Coroamă and Mattern, 2019) and “time further away from urban centres into
from digital rebound” (Binswanger, 2001), respectively. larger houses), communication methods
technologies Digitalization is also decreasing the skill
(more remote communication through
social media) and purchasing habits
thresholds needed to perform various
(online rather than offline) (table I.2).
activities, thus likely increasing their use
(“induction effect”) – a phenomenon that Challenges in measuring indirect effects
may be particularly visible for autonomous often lead to these being excluded when
vehicles (Coroamă and Pargman, 2020) assessing the true environmental impact of
and the use of data analysis through large digitalization. This underlines the importance
language AI models, such as ChatGPT, of developing better standardized
which previously required specialized frameworks to more adequately account
training. Additionally, an “obsolescence for indirect and rebound effects to ensure
that efficiencies are correctly estimated
effect” may arise as certain unconnected
in the future (Widdicks et al., 2023).
products become less useful because they
are not connected to newer generations
c. Combined effects of
of technology (Hilty and Aebischer, 2015).
digitalization are uncertain
Even if it is possible to achieve efficiency
improvements and substitute physical Understanding the cumulative environmental
goods with digital services,6 behavioural effects is crucial for policymakers,
changes due to rebound effects and researchers, the private sector and
increased overall consumption may mitigate consumers to determine the net impact of

6
While this substitution from physical goods to digital services may appear to reduce the need for materials,
this is not necessarily the case, as any digital service is enabled by devices, transmission networks and data
centres.

12
 Chapter I
Digitalization and environmental sustainability

digital technologies. The cumulative effect and substitution, or negative impacts by

depends on whether ICT is considered part inducing more consumption or making
of the problem or solution for environmental existing devices obsolete. Furthermore, more
sustainability (figure I.2): In terms of direct systemic indirect effects due to behavioural
effects, negative impacts arise from the or structural changes can either reduce or
production, use and end-of-life phases of increase the impact on the environment.7
digital devices and infrastructure. Applying
Indirect environmental effects could be
digitalization in other sectors, however, can
significantly greater than the direct
have both positive indirect effects, limiting
environmental footprint from using digital
environmental impacts through optimization

Table I.2
Indirect environmental effects from the use of digital devices and
infrastructure

Type of indirect Potential environmental Digital device example:

effect impact Use of maps on a smartphone

Substitution Products are replaced by their digital Replacement of paper-based maps and dedicated GPS-only devices.
▲▼ equivalents (with lower or higher
environmental impacts).

Optimization Adoption of digital technologies leads to Enhanced traffic and energy efficiency through real-time routing,
▼ efficiency improvements. reducing travel due to optimized routes.

Rebound Time and income effects. Optimization Same good or service: additional use of device compared to traditional
▲▼ gains from digital technologies enable cost paper-based maps, increased data consumption.
reductions (in terms of money or time), Other good or service: energy consumed during time/with resources
boosting the consumption of the good or saved by more efficient travel.
service or of other goods or services.

Induced consumption Digital technologies induce an increase Increased travel as smartphone-enabled routing eases and aids driving
▲ in the consumption or use of a product, in unfamiliar areas.
process or service.

Transformational (societal) Introduction of digital technologies causes Growth in location-based services and advertising; GPS technology
rebound macroeconomic adjustments across in smartphones boosts autonomous vehicles and expands intelligent
▲▼ sectors. transportation system manufacturing.

Sustainable lifestyle Digital technologies enable or encourage Smartphone maps and routing promote sustainable travel methods,
and practices more sustainable lifestyles and practices. such as walking or biking in unfamiliar areas.
▼

Systemic transformation and Digital technologies generate systemic Digital maps change transportation consumption boosting demand for
structural economic change society-wide transformations. car-sharing and ride-sharing such as Uber; long-term, GPS-enabled
▲▼ autonomous vehicles shift living and working location choices.
Improved navigation efficiency may enable more private vehicle use
over public transportation, and delay structural changes needed to
reduce carbon emissions and traffic congestion.

Source: UNCTAD, adapted from Bremer et al. (2023); Pohl et al. (2019); Horner et al. (2016).
Notes: A red upward pointing arrow indicates a negative effect (increasing environmental impact); a green
downward pointing arrow indicates a beneficial effect (avoided impact). A red upward pointing arrow next to a
green downward pointing one means that the net effect can be either positive or negative.

7
For more information, see IEA (2017); Bergmark et al. (2020); Coroamă et al. (2020); Global Enabling
Sustainability Initiative (2020); The Royal Society (2020); Bieser et al. (2023); Bremer et al. (2023); Kaack et al.
(2022); Technopolis and Institut für ökologische Wirtschaftsforschung (2024).

13
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

technologies.8 For example, direct life cycle transformation” in the figure) can have
GHG emissions (“ICT footprint” in the figure) larger, albeit uncertain, positive or negative
from teleworking using a computer, data indirect impacts, depending on how policy,
transmission networks and data centres are technology and behaviour interact and
likely to be less than 0.4 kg carbon dioxide evolve. However, to date, options to
equivalent emissions (CO2e)9 when a global comprehensively measure indirect effects
average electricity grid mix is used.10 This is remain limited, though the International
roughly one tenth of the emissions arising Telecommunication Union (ITU, 2022)
from a 20 kilometre commute to work by has put forward a first recommendation
car.11 Hence, using digital technologies (L.1480) on assessing the impact of ICT
can lead to a positive indirect effect of on GHG emissions in other sectors.
avoiding a commute equivalent to 4 kg of
CO2e emissions (“applications of digital This report thus focuses primarily on the
technology” in the figure). Longer-term direct environmental effects of digital devices
behavioural and lifestyle changes (“structural and infrastructure, encompassing the entire
effects and economic changes” and life cycle. Chapter V is an exception as it
“systemic and societal-level effects and looks at a specific use case of digitalization,

Figure I.2
Digitalization as a problem or a solution for promoting environmental
sustainability

ICT as a part of the ICT as a part of the

problem solution

Production
Life
Technology cycle of Use n/a by definition Direct effects
ICT
End-of-life

enables
Indirect effects
Induction effects Substitution effects
Application Second-order
Obsolescence effects Optimization effects

enables

Transition towards
Rebound effects sustainable patterns
Behavioural and Higher-order
structural changes Emerging risks of consumption and
production

Source: UNCTAD, based on Hilty and Aebischer (2015).

8
Indirect effects are also considered when categorizing scope 1, scope 2 and scope 3 emissions. Scope 1
covers direct GHG emissions owned or controlled by a producing entity; scope 2 covers indirect GHG
emissions from electricity, heating or cooling used; and scope 3 are indirect emissions linked to all other
indirect effects, e.g., from mining, production, inputs, transportation and end-of-life treatment (Allwood et al.,
2014). Scope 3 is understood to have the largest emissions impact, and is the most complex to measure.
9
CO2 equivalent emissions serve as a proxy measure that allows emissions from various GHGs to be compared
in terms of their potential for global warming. For this, an amount of a GHG is converted to an amount of CO2
which has the same global warming potential as the original GHG (Eurostat, 2023; IPCC, 2023).
10
Based on an eight-hour workday using a laptop (30W), 24-inch LED monitor (30W), 50 per cent allocation of
a router (5W), fixed access and core networks (<5W), data centre services (<2W) and associated embodied
emissions.
11
Based on the life cycle GHG emissions for an average new vehicle in 2017 including raw material extraction,
production, use (fuels included), based on IEA (2019).

14
 Chapter I
Digitalization and environmental sustainability

namely e-commerce. E-commerce has both Regardless of indirect environmental

positive and negative direct and potential impacts of digitalization, including societal
indirect environmental impacts, and effects, minimizing the direct footprint of
these can be influenced by policymaking. the digital economy remains essential.

B. Assessing the overall direct

environmental footprint of
digitalization
As noted above, accurately assessing the to rely on hugely varying and potentially Comprehensive
direct environmental impacts of the ICT outdated data sources, given the speed of assessments
sector is difficult. Rapid technological and change in the digital economy (Freitag et of the
economic changes further complicate al., 2021). There is also no standardized
measurement, with numerous factors approach for converting ICT energy use
environmental
affecting environmental impacts, such as in kilowatts per hour (kW/h) into tons footprint of
resource depletion, GHG emissions, water of GHGs emitted, as these depend on digitalization
consumption, biodiversity and noise. Taking the technologies and source of energy are scarce
a broad, multicriteria perspective on the used. Consequently, estimates vary
environmental footprint, available research significantly between countries and sectors
suggests that the production phase has (Chiarella et al., 2022). Nevertheless, as
the greatest impact. This is due to mineral energy use and GHG emissions data are
and metal depletion, the volume of GHG still the most frequently available, much
emissions generated and water-related research has focused on these areas.
impacts (Duporte et al., 2022). During the Second, the scope of the ICT sector varies
use phase, GHG emissions and water between studies. For instance, televisions
consumption are the main concerns and consumer electronics are included
(Agence de la transition écologique as part of the sector in some studies
(Ademe) and Autorité de régulation des (Andrae and Edler, 2015; Malmodin and
communications électroniques, des postes Lundén, 2018), but not in others (Belkhir
et de la distribution de la presse (Arcep), and Elmeligi, 2018). More importantly,
2022; Bordage, 2019; Freitag et al., 2021). new applications, such as AI, blockchain
and the IoT, are often not yet considered,
1. Measurement likely underestimating the overall sectoral
challenges impact (Freitag et al., 2021). The increasing
Comprehensive assessments of the integration of digital technologies into other
environmental footprint of digitalization sectors further complicates the ability to
are scarce, due to five factors. First, set clear boundaries when assessing
there is a lack of timely, comparable and the sector’s environmental footprint.
accessible data regarding the energy Third, studies also vary in the definition of
and environmental impacts of the ICT the life-cycle stages of the ICT sector. ITU
sector, with no harmonized reporting has introduced standards on conducting life-
standards. Additionally, there is often cycle analyses of the ICT sector (such as the
limited disclosure of impacts such as the ITU-T L.1410 and L.1450), but they have not
effect on local watersheds (Koomey and been consistently followed by researchers,
Masanet, 2021; Pasek et al., 2023). Data with some exceptions (ITU, 2020; Malmodin
scarcity leads to analytical studies having and Lundén, 2018; Malmodin et al., 2024).

15
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

Fourth, even those studies that look at measurement methodologies that can help
similar life-cycle stages have reached in policymaking.
different conclusions due to varying
assumptions and models adopted to
estimate the environmental impact. For 2. Estimates of the carbon
example, variations include anticipated footprint of the ICT
growth of the ICT sector, its correlation sector
with energy consumption (reflecting
assumptions on efficiency gains in As noted, energy use and GHG emissions
computing power) and the extent to which are the most researched aspects of the
ICT will contribute to emissions reductions ICT sector’s environmental footprint. The
in other sectors (Freitag et al., 2021). energy use of devices, data centres and
networks has been estimated to account
Moreover, existing literature mainly looks at
for approximately 6 to 12 per cent of global
the global environmental impact, overlooking
electricity use (about 1 to 2 per cent of
location-specific effects.12 As such, studies
global energy use), depending on use
neglect consequences that are highly region-
patterns, number of devices and associated
or country-specific, such as mining for raw
energy consumption (IPCC, 2022a).
materials, which primarily affects developing
Still, since 2015, studies assessing total
countries, and water use, both of which
GHG emissions of the ICT sector have
have profound environmental implications
arrived at vastly different results (table I.3).
that extend beyond the generalized
Estimates of life cycle emissions for 2015
impact of global GHG emissions.
range from 0.73 to 1.1 metric gigatons
Such methodological challenges have of CO2 equivalent (GtCO2e) emissions
led to considerable variation in estimates (1.4–2.2 per cent of global GHG emissions),
of the ICT sector’s environmental impact and for 2020 from 0.69 to 1.6 GtCO2e
(Koomey and Masanet, 2021) and of its emissions (1.5–3.2 per cent of global
subsectors. For example, to calculate GHG emissions). Differences are even
network energy intensity (i.e. the energy greater if the most optimistic and most
needed per amount of data sent across pessimistic estimates are also considered.
the Internet), existing estimates differed
These differences become more
by a factor of 20,000 a decade ago
pronounced in longer-term projections. For
(Coroamă and Hilty, 2014). Disagreement
example, Andrae and Edler (2015) estimate
also persists on whether overall impact
in their “best case” scenario that the ICT
is overestimated – due to outdated data,
The need excessive growth assumptions and
sector (excluding televisions and associated
to improve projections that extrapolate too far into the
devices) could emit 1.3 GtCO2e in 2030 but
as much as 19 GtCO2e in the “worst case”
availability of future (Koomey and Masanet, 2021) – or
scenario – representing a 15-fold difference.
quality data underestimated, because these estimates
exclude relevant technologies and trends Many of the studies in table I.3 are widely
and common
(Freitag et al., 2021). As it is vital to cited, but this does not mean they are
measurement estimate and analyse impacts to inform necessarily robust to changes in model
methodologies policy actions, the need to improve the assumptions and underlying data. For
must not be availability of quality data must not be an example, Andrae and Edler (2015) and
excuse for inaction. However, more work Belkhir and Elmeligi (2018) largely rely on
an excuse for
is needed to develop commonly accepted relatively simplistic extrapolations.13 More
policy inaction
12
A recent study by ITU and the World Bank provides estimates in country case studies, highlighting the variation
in data collection approaches for climate data in the ICT sector (Ayers et al., 2023).
13
The latter study extrapolates GHG emissions from data centres using a study from 2009–2010 (Vereecken et
al., 2010), applying an assumed compound annual growth rate from an industry report, implicitly disregarding
underlying drivers of data centre demand growth and efficiency improvements.

16
 Chapter I
Digitalization and environmental sustainability

recently, Andrae (2019a, 2020) significantly efficient battery-powered devices (such

revised downwards earlier estimates from as smartphones and tablets). Around
Andrae and Edler (2015) – in some cases 80 per cent of the GHG impacts of a
by more than half for 2020 – indicating smartphone’s life cycle can be attributed to
the limited usefulness of extrapolations the production phase (Ercan et al., 2016;
beyond a few years.14 The rapidly evolving Lhotellier et al., 2018; Clément et al., 2020;
nature of digital technologies makes long- Ademe and Arcep, 2022). Meanwhile,
term projections highly uncertain and the use phase dominates the GHG
further complicates defining the scope of impact of life cycles of data centres and
the ICT sector’s footprint as more objects networks due to their high energy intensity
become connected to the Internet.15 and constant operation (Andrae, 2020;
Bordage, 2019; Malmodin and Lundén,
The methodological approach of
2018; Malmodin et al., 2024; Masanet
Malmodin and Lundén (2018) involves
et al., 2013; Whitehead et al., 2015).
a more comprehensive combination of
bottom-up data (e.g. shipment data of
devices, servers, other hardware), detailed 3. Environmental footprint
life-cycle analyses, reported operator
data and benchmarking with other high-
beyond emissions and
quality studies that have focused on energy
specific ICT subsectors (e.g. data centres).
However, their methodology has also Direct environmental impacts of digital
been criticized in Freitag et al. (2021) for technologies also concern, among other
lacking transparency and replicability. impacts, raw material depletion, water The
consumption and quality, local air quality,
ITU (2020), largely based on Malmodin and
soil, biodiversity and waste. The importance
production
Lundén (2018), and Malmodin et al. (2024)
provide greater transparency regarding the of these impacts differs across ICT products phase emerges
methodologies and assumptions applied. and the different life cycle stages. For as the stage
In the case of energy use and associated
example, material use, water and air quality with the most
and biodiversity impacts are particularly
GHG emissions, different studies have adverse
important in the production phase, while
estimated that 56–80 per cent of the ICT waste generation is most important, environmental
sector’s total life cycle emissions come from but not exclusively, in the end-of-life effects
the use phase (Andrae, 2020; Bordage, phase. In studies applying comprehensive,
2019; Malmodin and Lundén, 2018; multicriteria analyses, the production phase
Malmodin et al., 2024; Masanet et al., emerges as the life-cycle stage that has the
2013; Whitehead et al., 2015). However, the most adverse effects on the environment
relative shares of each phase differ greatly (Ademe and Arcep, 2022; Bordage, 2019).
between data centres, data transmission
networks and connected devices. The Biodiversity and livelihoods can be
production phase is the most important severely affected by the water use of
for devices, especially for highly energy- digital technologies and infrastructure,16

14
The Shift Project’s 2019 report, largely based on modelling by Andrae and Edler (2015), has been widely cited
despite similar methodological issues. A review of data centre energy estimates by Mytton and Ashtine (2022)
also noted its methodological problems.
15
Some studies, such as Andrae and Edler (2015), used exponential growth rates to arrive at alarming figures
that have been widely quoted in the media (Koomey and Masanet, 2021). They projected that the ICT sector
could end up using half of the world’s electricity consumption by 2030, while accounting for nearly one quarter
of global GHG emissions. This, however, is an improbable scenario given the time required to build ICT and
energy infrastructure and the high costs of energy.
16
Low-income countries have recently experienced a loss in their biodiversity likely linked to important land
degradation from activities such as mining of critical raw materials in resource-rich countries, while other
countries have gained in per capita terms through accelerated conservation efforts (IPBES, 2019; Balvanera
et al., 2019).

17
 Table I.3
Overview of selected recent assessments of global greenhouse gas emissions

Institution and Greenhouse gas emissions

Approach Strengths Limitations
studies estimates (Per year)

Ericsson; Telia Hybrid approach combining bottom-up data (e.g. Strives to follow the ITU-T L.1450 Main limitation is the lack of transparency 0.73 GtCO2e in 2015
(Malmodin et al., 2024; shipments of devices, servers, other hardware), standard. The analysis is based on regarding some datasets (e.g., operator energy

18
0.76 GtCO2e in 2020
Malmodin and Lundén, previous life cycle analyses, top-down reported data rich and recent sources of bottom-up consumption). However, some of this data can
2018) from data centre operators, network operators, and data, use of reported operator data, be accessed through companies’ sustainability 1.4% of global emissions
major equipment manufacturers, and benchmarking to and benchmarking to established reports or other disclosures. The 2023 study
other studies such IEA (2017). studies implies high degree of frequently refers to supplementary material that is
analytical quality. not easily accessible.

GreenIT.fr For data centres, based on estimated number of Reports environmental impacts While the scope appears to be comprehensive, 1.4 GtCO2e in 2019
(Bordage, 2019) servers in operation and LCA of three data centres across multiple impact areas, the lack of detail regarding assumptions and data
from previous studies. For networks, based on data including energy consumption, GHG sources makes it difficult to evaluate its quality. 3.8% of global emissions (56% from use, 44%
traffic and access types. For devices, bottom-up emissions, water consumption and The scope of the study includes televisions and from manufacturing)
estimates, based on sales figures and average abiotic resource depletion. peripherals and is limited to manufacturing and
lifespans. use.

Huawei Data centres and networks estimated based on High degree of transparency, with Assumptions are transparent but lack sources Andrae (2020) “Expected” scenario:
(Andrae, 2019a, 2019b, Internet Protocol traffic estimates from Cisco with a supplementary model disclosing or explanations, making some appear arbitrary, 0.66 GtCO2e in 2015
2020, 2017; Andrae and varying assumptions for energy intensity per unit of assumptions and detailed results. such as energy intensity assumptions, efficiency
0.69 GtCO2e in 2020
Edler, 2015) Internet Protocol traffic (for data centres, fixed access improvement factors and assumed lifespan of
networks, and mobile access networks). Devices devices. 0.71 GtCO2e in 2025
estimated based on bottom-up data combining number The 2015 study’s long-term projections, especially 1.3 GtCO2e in 2030
of devices per year, expected electricity use, and the “worst case” scenario, have been widely Andrae and Edler (2015) “Expected” scenario:
expected lifespans. cited, suggesting that the ICT sector would
Projects three scenarios (best, expected, worst) to consume half of the world’s electricity and 0.94 GtCO2e in 2015
2030 with differing assumptions for electricity use, generate a quarter of global GHG emissions. 1.3 GtCO2e in 2020
pace of energy efficiency improvements, device These implausible long-term projections have 2.1 GtCO2e in 2025
lifespans, and share of embodied emissions. created confusion regarding the actual climate 4.4 GtCO2e in 2030
impacts of digital technologies and have since
been amended.
Digital Economy Report 2024

International Similar approach and data sources as Malmodin and Similar to Malmodin and Lundén Projections aim to align ICT sector emissions 0.74 GtCO2e in 2015
Telecommunication Lundén (2018) aligned with ITU-T L.1450 and linked to (2018), strives to follow the ITU-T with the 1.5°C objective which adds additional 0.69 GtCO2e in 2020
Union the 1.5°C objective of the Paris Agreement. L.1450 standard. Provides significant uncertainty to the estimations, as outlined in the 0.53 GtCO2e in 2025
(ITU, 2020) methodological detail in an annex. section “Risks”.
0.39 GtCO2e in 2030
Shaping an environmentally sustainable and inclusive digital future

ITU; World Bank Bottom-up extrapolation based on environmental, Clear on which life cycle stages Due to data limitations, estimates of data centres 0.57 GtCO2e in 2022
(Ayers et al., 2023) social and governance reports from the largest are included and which are not (no are limited to relatively efficient co-location 1.7% of global emissions
telecommunications operators, data centre operators, transport, end-of-life). data centres, cloud computing and social media (Telecom operators: 0.133 GtCO2e
equipment manufacturing and use. Takes into account variations storage, leading to a lower bound estimation of Co-location data centres: 0.042 GtCO2e
in electricity mix into its global the impact from this section of the ICT sector. Cloud and content data centres: 0.032 GtCO2e
aggregates. Not all life-cycle phases included. PC manufacturing: 0.065 GtCO2e; PC use:
0.187 GtCO2e
Smartphone manufacturing: 0.057 GtCO2e;
their use: 0.018 GtCO2e;
Network manufacturing: 0.033 GtCO2e)
 Institution and Greenhouse gas emissions
Approach Strengths Limitations
studies estimates (Per year)

Lancaster University Detailed review of recent studies estimating the global Provides a useful comparative Includes televisions and related peripherals in top- 1.2–2.2 GtCO2e in 2020 (including televisions)
(Freitag et al., 2021) GHG footprint of the ICT sector, focusing on Malmodin review of the three major global ICT level results, contrary to the ITU recommendation 2.1–3.9% of global emissions
and Lundén (2018), Andrae and Edler (2015) and GHG studies published since 2015, to categorize them under the entertainment and 0.8–1.7 GtCO2e in 2020 (excluding televisions)
Belkhir and Elmeligi (2018). comparing scopes, methods, and media sector.
assumptions. Lacks critical assessment of the reviewed studies,
implying equal validity of all three studies despite
varying levels of quality.

McMaster University Data centres and networks are extrapolations of The main strength of this analysis is Emissions from data centres and networks ignore 0.87 GtCO2e in 2015
(Belkhir and Elmeligi, trends from previous studies and industry reports on the bottom-up analysis of devices. their embodied carbon (emissions from materials 1.1–1.3 GtCO2e in 2020
2018) demand growth (e.g. Vereecken et al. (2010) for data and manufacturing). 1.4–1.8 GtCO2e in 2025
centres and Van Heddeghem et al. (2014), Lambert The linear extrapolation of GHG emission trends
7 GtCO2e in 2040
et al. (2012) and Pickavet et al. (2008) for networks). for data centres and networks to 2040 does not
Devices are estimated using a bottom-up approach take account of potential efficiency improvements
based on published academic and industry estimates and underlying drivers of demand growth, and
for units and life cycle impacts. likely overestimates future emissions if the
Projections are based on linear extrapolation of carbon historical trend is taken from a period of rapid
emissions for data centres and networks based on growth. Three years after publication, the author
trends from around 2008 to 2012. notes in Freitag et al. (2021) that: “regarding data
centres, Belkhir himself noted that his projection
Chapter I

of 495 MtCO2e for data centres in 2020 is

overestimated”.

Schneider Electric Bottom-up estimation of number of devices and As a post-COVID-19 study, this takes Study is not peer-reviewed. No information on 0.96 GtCO2e in 2020
Sustainability Research infrastructure. Given varying estimates of the existing into account recent changes in the distribution of emissions from manufacturing 2.8% of global emissions; (27% from
Institute stock of devices, the study focuses on growth rates development of the ICT sector. and embodied emissions, limited information manufacturing)
(Petit et al., 2021) rather than quantities. on definition of embodied emissions. Does
Digitalization and environmental sustainability

0.89–1.2 GtCO2e in 2030

not consider end-of-life and transportation.
Or 2.6–3.4% of global emissions (with
Projections for 2030 are based on the sector’s aim
manufacturing accounting for 32% of 0.89
to reduce emissions by 50 per cent by that point.
and 24% of 1.2 GtCO2e)

The Shift Project Uses model from Andrae and Edler (2015) (same Given that this analysis is based on Given that this analysis is based on the model 1.8 GtCO2e in 2015
(The Shift Project, 2019a, scope of technologies and life-cycle stages) with the the model developed by Andrae and developed by Andrae and Edler (2015), the same
2021) institution’s assumptions on data traffic and number of Edler (2015), the same strengths weaknesses outlined in Huawei also apply here. 2019 study, “Expected updated”:
devices and introduces new scenarios. outlined in Huawei also apply here. 1.6 GtCO2e in 2020
Projects four scenarios: expected updated; sobriety; 2.7 GtCO2e in 2025
higher growth higher energy efficiency; superior
growth peaked energy efficiency. 2021 study “Expected updated”:
1.4 GtCO2e in 2020
1.8 GtCO2e in 2025

19
Source: UNCTAD, based on Andrae (2017, 2019a, 2019b, 2020); Andrae and Edler (2015); Belkhir and Elmeligi (2018); Bordage (2019); Freitag et al. (2021); ITU (2020); Malmodin et al. (2024); Malmodin
and Lundén (2018); Petit et al. (2021); The Shift Project (2019a, 2021).
Notes: Refer to Freitag et al. (2021) for a detailed review and critique of most of the studies presented in this table. All GHG estimates are for the ICT sector only and exclude televisions unless specified.
Emissions excluding televisions in Freitag et al. (2021) were estimated by UNCTAD based on the primary data from the three studies reviewed by Freitag et al. GHG estimates for Andrae and The Shift
Project have been recalculated using their publicly available spreadsheet models to exclude televisions and associated peripherals (set-top boxes, DVD players, etc.). The Shift Project did not publish their
2021 model, so those figures are estimated based on the relative share of televisions from their 2018 model (2019 study). ITU (2020) projections for 2020–2030 are based on decarbonization efforts of the
electricity supply in line with the 1.5°C trajectory.
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

potentially threatening the balance of impacts can have varying effects at

ecosystems (Mewes, 2023). Estimates of local, regional and global levels. For
how much water is required to produce example, air pollutants have adverse
digital devices vary widely. For instance, impacts on local air quality and human
for smartphones, estimates range from health, whereas the impacts of climate
100 to 13,000 litres of water per device change from GHG emissions are global.
depending on the underlying assumptions
and modelling approaches (Friends of the Moreover, digital technologies can also
Earth, 2015; Leahy, 2014; Merchant, 2017). affect other dimensions of sustainability,
notably gender equity and human rights.
Beyond production, which includes mining, It is important to ensure that the human
using ICTs requires large amounts of water rights impacts and the unique challenges
for data centre cooling, with very limited confronting women and girls, youth,
water being reused (Monserrate, 2022). indigenous peoples and other groups at risk
This effect is aggravated by the fact that of being left behind, are not overlooked. For
many production hubs and data centres
Adverse instance, nearly 12.9 million women and
are located in areas under water stress many children work in the informal sector
effects (Jones, 2018; Farfan and Lohrmann, 2023; managing waste from digitalization, which
of device The Guardian, 2023).17 Similarly, the makes them significantly more likely to be
production and end-of-life phase is linked to significant exposed to potential negative consequences
digitalization- impacts on the water supply in some for their health (Parvez et al., 2021; World
locations. Groundwater contamination
related waste Health Organization (WHO), 2021a).
from leaching, dumping and digitalization-
often impact related processing activities can adversely From an equity perspective and in view of
regions far affect biodiversity and human health. today’s highly complex global supply
from where chains, it is important to recognize that
Mining for digital technologies comes with
the devices a significant environmental footprint. The
adverse impacts associated with device
production and waste generation at end-
are mainly specific impact depends on the local
of-life often affect regions located far away
used ecosystem as well as on the mining
from where the devices are predominately
technology used. As the overwhelming
used. While developed countries remain
majority of earth and rock removed in mining
the primary users of many aspects of the
is eventually discarded, this can lead to high
ICT sector, considerable harm may accrue
levels of toxicity from mining by-products
in regions that currently use and benefit
and soil damage (Dwivedi et al., 2022;
less from digitalization. However, to date,
The Shift Project, 2019a). Moreover,
research specific to the environmental
mining can be very water-intensive, often
impact of the digital economy on developing
leading to competition for water between
countries remains scarce.18 This results in
mining operations, agriculture and direct
policy discussions being skewed towards
consumption (UNCTAD, 2020).
the concerns of high-income countries
Most studies position themselves as global that are better positioned to harness
analyses. However, the environmental the benefits of digital technologies.

17
Overall, the share of global population affected by water stress is rising. In 2018, about 10 per cent of the global
population – more than 733 million people – lived in countries with high water stress (Food and Agriculture
Organization (FAO) and United Nations Water, 2021), with projections predicting a 40 per cent shortfall of
freshwater by 2030 (Global Commission on the Economics of Water, 2023), triggered by human activity (Yao
et al., 2023) and leading to increasing tensions within and between countries and the displacement of affected
populations.
18
This mirrors the scarcity in environmental research, especially on climate impacts, for low-income countries. In
these countries, 23 per cent of the population live in areas that remain uncovered by research on local climate
impacts, compared to only 3 per cent in high-income countries (Callaghan et al., 2021). This is also the case
in dimensions such as water access.

20
 Chapter I
Digitalization and environmental sustainability

4. Environmental the quality of infrastructure and the digital

literacy of individuals and businesses. The
sustainability in the
disparities in Internet use intensity – the
context of digital and data divide – are driven by varying levels
development divides of development and highlight missed
chances for leveraging digitalization for the
In the digitalization and environmental Sustainable Development Goals (UNCTAD,
sustainability nexus, the distribution of 2021a). Particularly pronounced in LDCs
environmental impact is linked to and remote Small Island Developing
countries’ geographical location and States, the digital divide is exacerbated
socioeconomic status. The disparities in by factors such as socioeconomic
income, wealth, digital access and use, status, location, age and gender.
and development have been further
exacerbated by the COVID-19 pandemic Overall, the divides in terms of development, Divides in
and recent geopolitical tensions. This environment and digitalization are
development,
underscores the need for nuanced policy interrelated, emphasizing the need to
responses to address these divides. address them holistically. Developing
environmental
regions are primary providers of many of responsibility
Developed countries have generated the the raw materials required for digitalization, and impact, and
bulk of emissions while propelling their
economic development, with Europe and
with extractive processes that can lead to digitalization
land degradation. Furthermore, developing
North America responsible for approximately
countries contribute to the part of global
are interrelated
40 per cent of anthropogenic CO2 emissions
value chains where value addition is and need to
since 1850 (Chancel et al., 2023;
relatively small and therefore have limited be addressed
Diffenbaugh and Burke, 2019; IPCC, 2023).
The Paris Agreement of 2015 (UNFCCC,
scope for accelerated economic growth. holistically
At the end of the life cycle of digital
2016) acknowledged this historical fact and
technologies, developing countries are
placed a greater responsibility on these
the destination for an important share of
countries for future GHG reduction efforts.
waste from global digitalization, which
However, the inequality in emissions opens up another dimension of the digital
transcends national borders, reflecting a divide. As noted above, these countries
stark divide in consumption patterns across are also more affected by climate change,
different income groups. The wealthiest 10 which directly impacts their options for
per cent of the population in every region socioeconomic development. Moreover,
emit significantly more than the global low-income countries are less able to
average (Chancel et al., 2023), associated afford and harness digital tools to mitigate
with overconsumption by wealthy individuals. various environmental impacts. Thus,
the opportunities for technologies to
While global Internet use surged from 35 address these environmental concerns in
to 67 per cent between 2013 and 2023, the short term are possibly overstated.
the digital divide remains a significant
barrier to socioeconomic development in By contrast, consumption patterns in
an increasingly digitalized world.19 Despite developed countries and of wealthy
advances in ICT infrastructure, disparities individuals everywhere are increasingly
in access and use persist, particularly marked by overconsumption. This is both
between high-income and low-income in terms of digitalization, for instance
countries, including LDCs. These divides measured by the number of devices per
encompass not just the number of devices person, and the environment, measured
and Internet connections per capita, but in terms of the multiples of CO2 emissions
also the affordability of digital services, per capita. Additionally, this group causes

19
ITU (2023). Key ICT indicators, available at https://www.itu.int/en/ITU-D/Statistics/Pages/facts/default.aspx.

21
 Digital Economy Report 2024
Shaping an environmentally sustainable and inclusive digital future

environmental externalities in developing towards a more environmentally sustainable

countries due to the production of digital economy that can generate inclusive
devices used in developed countries. development. At the same time, efforts are
needed to strengthen the ability of many
These factors point to the need for
developing countries to better harness
developed countries and digitally
opportunities from digitalization in an
advanced economies to assume particular
environmentally sustainable manner.
responsibility for ensuring a transition

C. Conclusions and roadmap for the

rest of the report
This chapter has highlighted the need to Trade and technological change are integral
give more attention to the interlinkages parts of the significant transformation
between the rapidly evolving digital economy process that the world is undergoing.
and environmental sustainability, and how This is underscored by the urgent need to
they relate to trade and development. The reduce carbon emissions, address widening
expanding scale and changing nature of economic inequalities and enable economic
digitalization have environmental implications diversification and structural transformation.
at all three stages of the life cycle of digital In the context of the interrelated nature of
devices and infrastructure. Depending on the Sustainable Development Goals, this
their positioning, countries will encounter requires policy integration and coherence at
different opportunities and challenges at the national, regional and international levels.
Amid Against this background, this report seeks
each stage. There is a need to improve
environmental the understanding of how countries at to contribute to a better understanding
crises and different levels of development are affected of the environmental impact of the
digitalization, and how this affects global trade dynamics. production, use and end-of-life phases
of digital devices and ICT infrastructure
developing The relationship between digitalization
with a view to informing policy debates on
countries and sustainability is bidirectional. Against
digitalization, trade and environmentally
a backdrop of multiple environmental
face a double sustainable and inclusive development.
crises and the importance of leveraging
bind… digital solutions for economic development While digital tools and solutions can be
and to tackle these challenges, it is used to reduce the global environmental
increasingly important to consider how impact of various sectors and bring the
to reduce the environmental footprint of 2030 Agenda for Sustainable Development
digitalization. However, this comes with back on track, positive outcomes cannot
…they are most a double bind for developing countries, be taken for granted. As shown in this
vulnerable to in particular LDCs. On the one hand, chapter, the overall environmental footprint
they are often the most vulnerable to of the digital economy is hard to assess
digitalization’s potential negative environmental and and remains largely unknown. Identifying
negative impacts social effects arising from digitalization, opportunities and risks from digitalization
yet least relating to raw material extraction, carbon is hampered by a lack of agreement on
equipped to emissions, water consumption and waste what constitutes the ICT sector and its
from digitalization. On the other hand, associated services, what criteria to include
use digital
they are less equipped to harness digital in an environmental impact assessment, a
tools for technologies to mitigate risks from climate lack of broadly agreed methodologies to
mitigation change and other environmental crises. measure impact, and a lack of data.

22
 Chapter I
Digitalization and environmental sustainability

The remainder of this report explores the applications. Chapter IV focuses on the end-
direct environmental impacts along the three of-life phase and the potential for fostering
main stages of the ICT sector life cycle. more circularity related to digital devices and
Chapter II focuses on the environmental infrastructure. Chapter V explores a case
impacts of the production phase, from of indirect and rebound effects from ICT
raw materials extraction and processing, use, notably in the context of e-commerce.
as well as manufacturing of ICT devices Finally, chapter VI discusses actions and
and infrastructure. Chapter III turns to policies to facilitate a more environmentally
the use phase, giving special attention sustainable digital economy which is
to the environmental impacts related to conducive to inclusive development.
data centres and emerging technological

23