Coursework Header Sheet

242876-69

Course BUSI1552: Sustainability in Business Course School/Level B/UG

Coursework Essay Assessment Weight 100.00%
Tutor A Piterou Submission Deadline 09/04/2020

Coursework is receipted on the understanding that it is the student's own work and that it has not,
in whole or part, been presented elsewhere for assessment. Where material has been used from
other sources it has been properly acknowledged in accordance with the University's Regulations
regarding Cheating and Plagiarism.

000988449
Tutor's comments

Grade Awarded___________ For Office Use Only__________ Final Grade_________

Moderation required: yes/no Tutor______________________ Date _______________
 Individual Essay

Analysing the Environmental Impact of Datacentres

Student ID: 000988449

Sustainability in Business (BUSI-1552)

Lecturer: Dr. Athena Piterou

University of Greenwich

09/04/2020

1
Introduction
Datacentres, each housing thousands of servers, are the backbone of the internet, as they
process and store data and enable the delivery of shared software applications, such as email
services or video-on-demand providers (Cisco, 2020). Due to the rapid growth of internet
applications, datacentres accounted for 0.3% of global carbon emissions, whilst also
accounting for 1% of global electricity demand in 2018. Furthermore, it could reach up to 8%
of electricity demand by 2030 (James, 2018). Whilst single issues of datacentres are well
studied, research is missing on the combination of these issues in a comprehensive analysis
(Massari & Ruberti, 2013).

This paper seeks to form a more holistic approach to the environmental impact analysis of a
datacentre. At first, core sustainability concepts are evaluated, with a particular light on the
role of businesses within. Then, a lifecycle analysis, following an EPA framework, will be
executed from the acquisition of the raw materials to their disposal, primarily focussing on
the required metal resources. In the end, mitigation strategies, some of which are already in
use, are suggested to reduce the environmental impact of identified issues.

Literature Review
Sustainability concepts originated from the ‘Limits to Growth’ debate of the early 1970s,
which asked if societal collapse and severe environmental damage would be inevitable when
maintaining economic growth. By the end of the decade, it was determined that growth could
only be indefinite if modified in acknowledgment of the dependence on the natural
environment and its resources. In 1980, the concepts of sustainable development were first
publicised by the World Conservation Strategy (Pezzey, 1992). However, the so called
“Brundtland Report” of 1987 formed the most common definition of sustainable
development up to today: “development that meets the needs of the present without
compromising the ability of future generations to meet their own needs” (WCED, 1987).

The concept of sustainability is said to be carried on the triple bottom line, meaning the social,
the environmental and the economic pillars. All three of these pillars are supposed to secure
a company’s future. The social pillar is to take responsibility of the company’s impact on
society as a whole, on its stakeholders and its human resources. The environmental pillar is

2
to take responsibility in its visible and invisible environmental damage, such as waste or CO2-
emissions. The economic pillar is to take responsibility of the available financial resources, to
gain them sustainably and use them efficiently. It is argued that a sustainable business gains
a competitive advantage, or higher profits, because of optimised and cost-reduced processes
(Slack, et al., 2016). Furthermore, when the UN (2015) agreed on the 17 Sustainable
Development Goals (SDG), a framework to be reached by 2030, which cover, inter alia,
equality, innovation and energy, they specifically stressed the role of businesses, as well as
private investment, in reaching them.

Whilst there are some voices calling for the end of capitalism, with its aim for indefinite
economic growth, and promote the introduction of concepts such as “sustainable socialism”
(Pena, 2014), others are suggesting a reform to reconcile this aim with the environment. In a
reform concept, Lovins & Lovins (2001) suggested “Natural Capitalism” (NC) as a new
industrial revolution, first introduced in their book two years prior. It is to be understood as
an end of “Industrial Capitalism” (IC), which system’s flaws do not necessarily incentivise
sustainable business. In contrast to IC, NC argues for giving natural as well as human capital
an economic value. For example, most ecosystems have no available substitute, but their free
services, such as the large-scale oxygen production of trees or pollination, are necessary for
the production of goods and services. Henceforth, their value is theoretically infinite, though
the costs of ecosystems is only visualised when destruction has progressed to the extent of
service disruption (e.g. the increasing number of extreme weather scenarios, such as storms).
(Lovins & Lovins, 2001).

NC is based on four principles: Firstly, increasing resource productivity and decreasing waste
as natural, and not human, resources are scarce today, in contrast to the times of the
industrial revolution. Secondly, eliminating waste by introducing a circular, closed-loop
economy, which produces a saleable product rather than imposes disposal costs. This concept
is also known as industrial ecology. Thirdly, shifting from a transactional to a relationship
marketing-based economy, which increases value for both parties, through a continuous
exchange of value, and performance, through more-individual solutions. Therefore, cutting
waste and increasing profits. Lastly, as more and more industries develop away from mining

3
primary inputs towards growing, it becomes increasingly important to reverse natural
destruction by reinvesting the gained profits (Lovins & Lovins, 2001).

To make these principles a reality, technological advancement, radical or incremental, in

many areas is needed. Radical sustainable innovations - entirely new technology to replace
old polluting one - will not only give companies a competitive advantage but simultaneously
drastically improve performance. Those innovations, however, are risky, highly expensive and
rare. (Kennedy, et al., 2017). Incremental innovations alone - progressive improvements to a
product - are not enough to overcome the challenge, since efficiency gains are often offset
by increased consumption or scale effects (Machiba, 2010). This is due to the so-called
rebound effect, as price reductions directly and indirectly impact both the macro- and micro-
levels (Berkhout, et al., 2000).

As most of the world has to further develop and increase consumption against poverty, the
developed world is faced with challenges from the negative effects of over-consumption, such
as obesity. The rebound effect, in addition to a growing, and increasingly wealthy, population
on this planet make supply-side initiatives alone inadequate. Henceforth, the concept of
“sufficiency” seeks to moderate the demand-side and push for consumers to live on needs
rather than wants. For example, it changes the business model from cheap, short-lived
products towards durable and reparable products, and away from aggressive promotions
towards educating consumers (Brocken & Short, 2016).

Analysis of Environmental Impacts

The analysis of this essay follows the framework shown underneath and provided by the EPA
(Keoleian & Menerey, 2004). For the production of computing technologies, the main raw
materials are rare earths (RE) (e.g. precious metals (e.g. gold, silver platinum) as well as other
metals (e.g. copper, iron, aluminium) (Ari, 2016). Due to the complexity of data centres and
size restrictions, this paper excludes possible environmental impacts of required buildings,
batteries or generators as electricity supports, or other materials needed for the production
of essential components, in addition to the software, which can heavily impact performance.

4
 Figure 1 Product Lifecylcle Framework by Keoleian & Menerey (2004)

Whilst the mining and processing of any metal ore is severely damaging to the local
environment, this is especially true for RE. Due to rapidly growing demand and substitutes
not being available, there is severe reliance on unsustainable sourcing and processing of RE
(Massari & Ruberti, 2013). Furthermore, the supply of many more metals, in addition to RE,
may become scarce in the future (Graedel, et al., 2015). Current metal extraction and refining
technologies not only require vast amounts of water, electricity and acidic substances but
also release hazardous, and in terms of RE, even radioactive, pollution in the form of gases
(incl. Co2), particulate matter, sewage and solid waste. Therefore, mitigation and safety
standards represent a major capital expense. Additionally, the carbon footprint and other
emissions of used electricity depend on the fuel mix of each respective country, and transport
emissions are relatively low, as mining and processing are often done within the same country
(Massari & Ruberti, 2013; Dudka & Adriano, 1997).

Today, despite worldwide deposits, the extraction and processing of metals is mainly done in
developing countries, such as China, Brazil, India, Russia and the Congo due to often richer,
more easily accessible deposits, and looser environmental regulations and safety standards
(i.e. providing a price advantage). Main exceptions include Canada, the USA and Australia
(Massari & Ruberti, 2013; ICMM, 2018). Using the RE as an example, most of the mines in
China are unlicensed and unregulated, using cheap labour and causing environmental
damage, whilst in Mongolia, cases of cancer, dental loss and respiratory diseases have been
increasing (Massari & Ruberti, 2013) and in the Congo, a country without safe drinking water
and available electricity, there were reports of child labour (Amnesty International, 2016).

5
The third step (engineered and specialty materials) and part of the fourth step (the
manufacturing of components) in the cycle are most likely to be done in East Asia. This is due
to the fact that Silicon Valley companies, which are a powerful player on the electronics
market, outsourced standardised, low value-added activities (i.e. production) primarily to
developing economies, especially in East Asia. This means that together with the rise of local
global players, such as Samsung, LG or Huawei, the region consisting of Japan, South Korea,
China and Taiwan has a dense network of suppliers and plays a significant role in the global
value chain, resulting in significant transport emissions (Gong, et al., 2018).

Since the core components of a computer are based on nanoparticle technology, necessary
materials, such as metals, have to be cut down into nanoparticles. There is not enough
research at the moment to know how hazardous nanoparticles are. Nonetheless, studies do
show ecotoxicity in certain circumstances, posing a risk for workers and, if the waste is not
disposed properly, the environment. For this process, electricity is used. (Bhatt & Tripathi,
2011). Once the nanotechnology and other components, such as cooling fans, are
manufactured, they will be packaged and sent to the server location, where they will be
assembled together. This causes, once again, greenhouse gas emissions.

As already mentioned earlier, this paper excludes the building of datacentres and assumes
the location to be the UK. Environmental impacts of assembly are negligible as compared to
the operation itself, only a fraction of electricity is needed. In the operating phase, the main
driver of environmental impact is the electricity consumption. Due to heavy reliance on power
and rising energy costs, datacentres historically have used operational efficiency as a measure
for sustainability (Whitehead, et al., 2014). Hereby, it should be differentiated between the
energy consumption of the required cooling and the actual IT. Depending on the method and
the climate, cooling can account for up to 40% of total energy consumption. Although, cooling
costs can be assumed to be relatively low due to the UK’s relatively cool climate (Zhang, et
al., 2017). Whilst most energy will come from the main power supply, datacentres sometimes
need to use diesel generators or batteries to cover peak demand and blackouts. This is not
considered in this paper. In case of failure or a decrease in performance, components in a

6
single server will be exchanged whilst other servers take over the workload. Generally, servers
will be completely exchanged every three to five years (Whitehead, et al., 2015).

The disposal of electronic waste (EW) has been an issue for a long time. EW will drastically
increase in the future. It is hazardous but also valuable at the same time because of certain
metals. Metals are generally recyclable, but models around EW are still evolving globally (Ari,
2016). An open-loop recycling model is the norm in the industry, as exchanged components
are most often technically obsolete, and therefore cannot be used elsewhere. This processes
the EW back into reusable raw materials and other waste (Whitehead, et al., 2015). Generally,
the environmental impact is substantially reduced when using recycled instead of primary
materials, in terms of resource, as well as energy usage. (Reck & Graedel, 2012). As the
Western norm of waste and recycling “outsourcing” often gets criticised internationally, it is
forbidden under UK law, based on an EU directive, to export hazardous waste to any country
outside the OECD. However, in a report of the BAN (2018), it has been found that the UK has
the most illegal exports of EW compared to another nine EU countries. This poses risk to the
health of workers and the environment, as correct waste processing cannot be guaranteed
(BAN, 2018).

Overall, the environmental impact of datacentres is complex because of the plurality of

needed resources and manufacturing, as well as service locations. This holds especially true
since all manufacturing processes, and the operation itself, are electricity and energy
intensive, but the fuel mix varies strongly between countries (e.g. highly polluting coal is still
the major source of energy in China (Lin, et al., 2018). Moreover, most academic analyses of
the environmental impacts of data centres often only concentrate on one issue, for example,
electricity consumption. A complete life cycle assessment of a UK datacentre was done by
Whitehead, et al. (2015), who concluded that the operation of the IT has a larger impact than
the manufacturing, due to the high impact of burning fossil fuels for the production of
electricity.

However, this paper disagrees with this point. The pollution and carbon intensity of electricity
production have drastically fallen in recent years, as coal has largely been replaced with
renewables, and this trend will continue in the future (Ofgem, 2019). This aligns this paper

7
closer with others that estimate the damage of manufacturing electronics as much higher
(Deng & Williams, 2011; Vasan & Pecht, 2014). The reason for this is the immense direct
impact of mining onto communities and the environment of developing countries, without
an immediate substitute available for RE, such as in the same way electricity production with
renewables substitutes fossil fuels.

Mitigation of Environmental Impact

Since the environmental impacts have now been explained, this paper will further introduce
mitigation strategies. First it should be clarified that there are no substitute services or
products for datacentres, as they have a central position in the function of the internet.
Instead, certain datacentre services could have positive impacts that potentially offset some
negative ones. For example, cloud computing offers the chance to centralise computing
power, which increases resource efficiency (as several users share hardware) and decreases
hardware turnover on the consumer-side (as older hardware could cope more easily with new
requirements) (Manyika, et al., 2013). This is supported by Koomey & Taylor (2015), who
argue that at least 30% of servers in the US have not been utilisied for at least six months,
despite being online. If companies centralised their IT services and outsourced it to
datacentres, energy consumption and resource usage would be greatly reduced.

Furthermore, an increased circular economy thinking would vastly increase resource

efficiency, decrease costs, increase revenue and have, therefore, positive impacts on the
environment. For example, the waste heat of datacentres is often unused nowadays. Heat
pumps could cool down the datacentres, whilst feeding the energy into a district heating
network to provide heating and warm water for housing (Bloomberg NEF; Eaton; Statkraft,
2020). Another example is the increased use of recycling, which should be aimed for. At the
moment, the recycling rates of precious metals and rare earths are very low. However,
especially the problematic world market situation of the latter (due to dependency on China’s
monopoly) should be incentive enough to develop recycling concepts (Liu, et al., 2011).
Recycling could save up to 87% of natural resources, compared to the landfill (Eygen, et al.,
2016). Additionally, the problem of exported EW could be fixed by tightened regulation, as
advised in BAN (2018).

8
Innovations could further alleviate the situation in three aspects. First, increased energy
efficiency of servers would have a positive impact on the IT power consumption. This would
not cause the rebound effect, as long as cost reductions are not given further to the
customers. Although, this is not likely due to rising electricity bills, which minimise the savings
(Ofgem, 2019). Second, cooling innovations may significantly decrease power consumption.
An example of this are underwater servers currently tested by Microsoft, which do not need
extra cooling (Roach, 2018). Third, further innovations into the possibility of reduction or
substitution of RE and other metals. Since the price hikes around 2010, research already has
concentrated, and achieved, a reduction of RE usage. Further uncertainty of the market may
increase willingness to find alternatives (Skokov & Gutfleisch, 2018).

In the end, however, the overall impact of mining can only be reduced by tightened regulation
in each respective country, under pressure from NGOs (such as Amnesty International) and
IT businesses, which become increasingly aware of sustainable practices and strive for carbon
neutrality (Kurp, 2008). In fact, the IT industry is often a pioneer, as electricity savings is their
main driver behind cost savings. Hence, there are many examples where carbon, and cost,
savings were achieved through efficient cooling systems or use of local low-carbon electricity,
mostly done off-site close to the datacentre since on-site generation would not be able to
meet the power demand (Kester, et al., 2014). Additionally, the concept of sufficiency can
only be partly applied here, depending on the service provided by the datacentre. For
example, in the case of video-streaming services, providers could argue for a decreased
resolution. This would decrease required IT power, bandwidth and therefore power
consumption. However, the same may not be possible with an online office suite.

Conclusion
In conclusion, this paper explained the origin of the sustainability discussions and its core
concept of meeting current societal needs without compromising the needs of future
generations. Due to the important role of private investment to achieve the UN SDGs, it
argued along the theory of natural capitalism, which seeks to use resources more efficiently,
eliminate waste and preserve natural as well as human capital through assigning it economic
value. It went on to explain the role of innovation within sustainability, and the interaction

9
between businesses and consumers, through concepts such as sufficiency and the rebound
effect.

This paper recognised the case of datacentres as a rapidly growing carbon emitter, and
explored it with a life-cycle assessment, concentrating on the datacentre operation in the UK
in addition to the production and disposal of metal. Severe sustainability issues have been
found with the mining of metals, especially RE. Metal mining and processing is not only energy
intensive, with the majority of energy provided through the burning of fossil fuels in
developing countries like China, but also causes local environmental degradation, impacting
the health of workers and communities. Further, still understudied issues were described
through the use of the necessary nanotechnology in the steps of producing engineered
materials and manufacturing their components. Moreover, the transport through globalised
value chains add to the negative balance.

The environmental impact of assembly was considered negligible compared to the operation
phase, as the latter uses significantly more electricity due to the high concentration of IT
power and its required cooling. For disposal, large potential for a closed-loop recycling system
was determined, which is being undermined by low exploitation and illegal export of
hazardous, though valuable, waste. Therefore, it was concluded overall that the production
and disposal of the components considers a bigger environmental threat than the operation
itself for two reasons: First, the on-going decarbonisation of the UK’s electricity coupled with
a relatively low cooling electricity demand; Second, the missing alternatives to the usage of
metals, such as RE, their severely polluting mining technologies worsen by the missing safety
regulation in developing countries.

In the end, possible mitigation strategies were discussed, some of which are already in use
today. In fact, the paper underlined the pioneering function of the IT industry, and its natural
incentive to reduce energy consumption. Whilst there are no substitutes available, some
services may have a positive environmental impact, due to increased resource efficiency.
Suggested strategies included to increase circular economy thinking by increasing recycling,
utilising waste heat as well as local low-carbon energy sources. Additionally, this paper

10
discussed the importance of innovation to decrease lifetime resource usage, in addition to
the unlikelihood of a rebound effect and the impracticability of sufficiency concepts.

References
Amnesty International, 2016. "This Is What We Die For": Human Rights Abuses in the
Democratic Republi of the Congo Power The Global Trade in Cobalt, London: Amnesty
International.

Ari, V., 2016. A Review of Technology of Metal Recovery from Electronic Waste. In: F. Mihai,
ed. E-Waste in Transition. Rijeka : InTech, pp. 121-158.

BAN, 2018. Holes in the Circular Economy: WEEE Leakage from Europe, Seattle: Basel Action
Network.

Berkhout, P. H., Muskens, J. C. & Velthuijsen, J. W., 2000. Defining the rebound effect.
Energy Policy, 28(6-7), pp. 425-432.

Bhatt, I. & Tripathi, B. N., 2011. Interaction of engineered nanoparticles with various
components of the environment and possible strategies for their risk assessment.
Chemosphere, Volume 82, pp. 308-317.

Bloomberg NEF; Eaton; Statkraft, 2020. Sector Coupling in Europe: Powering

Decarbonization, London: Bloomberg Finance L.P..

Brocken, N. & Short, S., 2016. Towards a sufficiency-driven business model: Experiences and
opportunities. Environmental Innovations and Societal Transitions, Volume 18, pp. 41-61.

Cisco, 2020. What is a Data Centre?. [Online]

Available at: https://www.cisco.com/c/en_uk/solutions/data-center-virtualization/what-is-
a-data-center.html
[Accessed 9 April 2020].

Deng, L. & Williams, E. D., 2011. Economic-balance hybrid LCA extended with uncertainty
analysis: case study of a laptop computer. Journal of Cleaner Production, 19(11), pp. 1198-
1206.

Dudka, S. & Adriano, D. C., 1997. Environmental Impacts of Metal OreMining and
Processing: A Review. Journal of Environmental Quality, 26(3).

Eygen, E. v., Meester, S. D., Tran, H. P. & Dewulf, J., 2016. Resource savings by urban mining:
The case of desktop and laptop computers in Belgium. Resources, Conservation and
Recycling, Volume 107, pp. 53-64.

11
Gong, P., Liu, W., Tang, Z. & Yin, G., 2018. Electronics industry interregional linkages in East
Asia and their regional development impacts in China. Area Development and Policy, 3(1),
pp. 70-92.

Graedel, T. et al., 2015. Criticality of metals and metalloids. PNAS (Proceedings of the
National Academy of Sciences of the United States of America), 112(14), pp. 4257-4262.

ICMM, 2018. Role of Mining in National Economies, London: International Council on Mining
and Metals.

James, N., 2018. How to stop data centres from gobbling up the world’s electricity (online).
Nature, 12 September.

Kennedy, S., Whiteman, G. & Ende, J. v. d., 2017. Radical Innovation for Sustainability: The
Power of Strategy and Open Innovation. Long Range Planning, 50(6), pp. 712-725.

Keoleian, G. & Menerey, D., 2004. Life cycle design guidance manual. Environmental
requirements and the product system, Washington, D.C.: U.S. Environmental Protection
Agency.

Kester, C., Schuchard, R. & Mohapatra, A., 2014. Intelligent Low-Carbon Power Sourcing for
Data Centers (Working Paper), New York City: BSR.

Koomey, J. & Taylor, J., 2015. New data supports finding that 30 percent of servers are
‘Comatose’, indicating that nearly a third of capital in enterprise data centers is wasted,
London: Anthesis Group.

Kurp, P., 2008. Green Computing. Communication of the ACM, 51(10), pp. 11-12.
Lin, J. et al., 2018. Has coal use peaked in China: Near-term trends in China's coal
consumption. Energy Policy, Volume 123, pp. 208-214.

Liu, R. et al., 2011. Application of rare earths in consumer electronics and challenges for
recycling. Berlin, ICCE International Conference on Consumer Electronics.

Lovins, L. H. & Lovins, A. B., 2001. Natural Capitalism: Path to Sustainability?. Corporate
Environmental Strategy, 8(2), pp. 99-108.

Machiba, T., 2010. Eco-innovation for enabling resource efficiency and green growth:
development of an analytical framework and preliminary analysis of industry and policy
practices. International Economics and Economic Policy, Volume 7, pp. 357-370.

Manyika, J. et al., 2013. Disruptive technologies: Advances that will transform life, business,
and the global economy, New York: McKinsey Global Institute.

Massari, S. & Ruberti, M., 2013. Rare earth elements as critical raw materials: Focus on
international markets and future strategies. Resources Policy, 38(1), pp. 36-43.

12
Ofgem, 2019. Wholesale electricity generation market shares by company in 2018 (GB).
[Online]
Available at: https://www.ofgem.gov.uk/data-portal/wholesale-electricity-generation-
market-shares-company-2018-gb
[Accessed 28 March 2020].

Pena, D. S., 2014. The Six Essential Components of Sustainable Socialism: From Building the
Productive Forces to Combating Bourgeois Liberalization. International Critical Thought,
4(3), pp. 267-288.

Pezzey, J., 1992. Sustainable Development Concepts: An Economic Analysis, Washington

D.C.: The World Bank.

Reck, B. K. & Graedel, T. E., 2012. Challenges in Metal Recycling. Science, Volume 337, pp.
690-695.

Roach, J., 2018. Under the sea, Microsoft tests a datacenter that’s quick to deploy, could
provide internet connectivity for years: Microsoft. [Online]
Available at: https://news.microsoft.com/features/under-the-sea-microsoft-tests-a-
datacenter-thats-quick-to-deploy-could-provide-internet-connectivity-for-years/
[Accessed 9 April 2020].

Skokov, K. & Gutfleisch, O., 2018. Heavy rare earth free, free rare earth and rare earth free
magnets - Vision and reality. Scripta Materialia, Volume 154, pp. 289-294.

Slack, N., Brandon-Jones, A. & Johnston, R., 2016. Operations Management. 8th Edition ed.
Harlow: Pearson Education Limited.

United Nations (UN), 2015. Sustainable Development in Action, New York City: Division for
Sustainable Development, United Nations Department of Economic and Social Affairs.

Van Eygen, E. & Dewulf, J., 2016. Resource savings by urban mining: The case of desktop and
laptop computers in Belgium. Resources, Conservation and Recycling, Volume 107, pp. 53-
64.

Vasan, A. & Pecht, M., 2014. Carbon footprinting of electronic products. Applied Energy,
Volume 136, pp. 638-648.

WCED (World Commission on Environment and Development), 1987. Our Common Future
(Brundtland Report), Oxford: Oxford University Press.

Whitehead, B., Andrews, D., Shah, A. & Maidment, G., 2014. Assessing the environmental
impact of data centres part 1: Background, energy use and metrics. Building and
Environment, Volume 82, pp. 151-159.

Whitehead, B., Andrews, D., Shah, A. & Maidment, G., 2015. The life cycle assessment of a
UK data centre. International Journal of Life Cycle Assessment, Volume 22, pp. 332-349.

13
Zhang, X. et al., 2017. Cooling Energy Consumption Investigation of Data Center IT Room
with Vertical Placed Server. Energy Procedia, Volume 205, pp. 2047-2052.

14