Technical ReportPDF Available

Data centers 2018. Efficiency gains are not enough: Data center energy consumption continues to rise significantly - Cloud computing boosts growth

Authors:

Abstract

In 2018, the power requirements of data centers in Germany rose significantly again. Compared to the previous year, the demand for electrical energy by servers and data centers increased by 6% to 14 billion kWh. This growth is primarily due to the strong expansion of cloud computing capacity in Germany. Substantial new data center capacities were built up, particularly in the greater Frankfurt area, but also at other locations. This development is expected to continue in the future. Trends such as edge computing and artificial intelligence are expected to lead to a significant expansion of data center infrastructures in Germany, Europe and worldwide. If the existing efficiency potentials are not realized , the energy consumption of data centers will continue to rise significantly. These are the results of a recent study by the Borderstep Institute on the development of the energy consumption of data centers in Germany.
Data centers 2018
Cloud computing boosts growth
Efficiency gains are not enough: Data center en-
ergy consumption continues to rise significantly
Dr. Ralph Hintemann
In 2018, the power requirements of data centers in Ger-
many rose significantly again. Compared to the previous
year, the demand for electrical energy by servers and
data centers increased by 6% to 14 billion kWh. This
growth is primarily due to the strong expansion of cloud
computing capacity in Germany. Substantial new data
center capacities were built up, particularly in the
greater Frankfurt area, but also at other locations. This
development is expected to continue in the future.
Trends such as edge computing and artificial intelligence
are expected to lead to a significant expansion of data
center infrastructures in Germany, Europe and world-
wide. If the existing efficiency potentials are not real-
ized, the energy consumption of data centers will con-
tinue to rise significantly.
These are the results of a recent study by the Borderstep
Institute on the development of the energy consumption
of data centers in Germany.
Although there have been very significant improvements
in the energy efficiency of data centers in recent years,
the sharp rise in demand for centralized computing power
has led to a further increase in the energy consumption of
data centers in Germany (Figure 1). In particular, the IT
components (servers, storage and network) will require
8.5 billion kWh of electrical energy in 2018, significantly
more than in 2010 (5.8 billion kWh). The average PUE
value1 of data centers in Germany fell from 1.98 to 1.702
between 2010 and 2018. This increased the efficiency of
the data center infrastructure by 16% on average.
1 The Power Usage Effectiveness (PUE) value indicates the ratio of
the annual energy consumption of the entire data center to the an-
nual energy consumption of the data center's IT components.
Figure 1: Energy consumption of servers and data centers in
Germany in the years 2010 to 2018 (Source: Borderstep)
If past trends continue, the energy consumption of data
centers in Germany will continue to rise, increasing by
50% from 2018 to 2030. However, this development is not
inevitable. On the basis of analyses and evaluations of
more than 60 new energy and resource-saving technolo-
gies conducted in the TEMPRO project (Hintemann &
Hinterholzer, 2018), the project partners already imple-
mented some particularly promising technologies as pro-
totypes.
If the existing technical efficiency potentials are success-
fully exploited, the energy consumption of data centers in
Germany could even be reduced by 25% by 2030 despite
the strong expansion of data center infrastructures, as
was determined in the TEMPRO project.
2 The stand-alone servers, which are normally operated without
their own air conditioning, are not included in the calculation of
these values. Taking the stand-alone servers into account, the av-
erage PUE value in Germany improved from 1.82 in 2010 to 1.63 in
2018.
Cloud and edge computing, artificial intelligence: En-
ergy consumption of data centers is expected to con-
tinue to rise
Increasing digitalization and the associated new applica-
tions, such as in the field of artificial intelligence (AI), but
also the significant increase in cloud and edge computing
capacities will lead to a growing demand for data center
infrastructures.
Growth in the data center market is driven primarily by
the rapidly increasing use of cloud services. International
cloud computing providers in particular are currently ex-
panding their data center capacity in Germany vigorously.
Because of economies of scale, particularly efficient data
center infrastructure and typically high server utilization,
cloud computing data centers are often significantly more
efficient than traditional data centers (Bizo, 2019;
Shehabi et al., 2018). To date, cloud data centers in Ger-
many have been established in addition to the existing
traditional data centers. Hardly any traditional on-prem-
ise data centers have been dismantled. As a result, de-
spite the higher efficiency of cloud computing solutions,
the overall energy consumption of data centers continues
to rise (Figure 2).
Figure 2: Development of the electricity consumption of data
centers in Germany with share of cloud and edge data centers
from 2010 to 2018 and forecast until 2025 (Source: Borderstep)
In regional terms, the greater Frankfurt area in particular
is benefiting from growth in cloud computing. Investors
often decide to construct new data centers in Frankfurt
because of the good network connection through DE-CIX
and the geographical proximity to the cloud capacities al-
ready available there. This is also confirmed by the ana-
lysts of the real estate service provider CBRE, who cur-
rently see the Frankfurt area as the strongest growing
market in Europe. Most of the additional data center ca-
pacities in Europe are to be built in Frankfurt over the next
two years (CBRE, 2020).
In the future, edge data centers will also have an increas-
ing share of the energy consumption. In 2025, edge data
centers in Germany will probably require 1.5 billion kWh
of electrical energy. With the further expansion of 5G mo-
bile phone networks and edge computing applications in
areas such as Industry 4.0 applications, Autonomous Driv-
ing and Smart City, the energy consumption of edge data
centers is expected to rise to around 4.5 billion kWh/a by
2030. In a scenario with increased expansion of edge com-
puting, calculations in the TEMPRO project even indicate
that edge computing could account for 30% of the energy
consumption of all data centers.
New applications, especially in the field of artificial intelli-
gence, may cause increasing energy consumption in data
centers. AI is increasingly penetrating the human habitat
(Reinsel et al., 2018; Schneider & Ziyal, 2019; Walsh,
2018). AI solutions can be found everywhere, from the liv-
ing room with speech recognition solutions to cloud solu-
tions for deep learning applications to use in critical infra-
structures (e.g. in efficient and sustainable energy net-
work management systems). This development has a va-
riety of environmental impacts. On the one hand, AI offers
many opportunities to make our living and working envi-
ronments more sustainable. There are promising fields of
application, especially for a better understanding of the
Earth, the climate and the environment, as well as in the
areas of agriculture, energy and mobility (Jetzke et al.,
2019). As AI becomes more successful and opens up new
areas of application, its resource requirements increase
as well. Some deep learning applications, simulations and
prognoses in particular demand enormous amounts of
computing power and require large amounts of energy
and resources. Researchers at MIT have calculated that
the training of a single AI application for speech recogni-
tion generates five times as much CO2 as a car during its
entire lifetime (Hao, 2019; Strubell et al., 2019).
Precisely because of the great potential of AI and the pos-
sibility of accessing AI applications from any smartphone
or other intelligent device, it is assumed that the use of AI
technologies will continue to increase dramatically in the
future (Hintemann & Hinterholzer, 2019). In the period
from 2016 to 2021 alone, the workloads for the field of
"Database/Analytics/IOT" in data centers worldwide are
expected to increase by a factor of 2.5 (Cisco, 2018).
International development: Studies paint different
pictures
An analysis of the internationally available studies and
publications on the energy consumption of data centers
does not produce a uniform picture. Some researchers as-
sume an enormous increase in energy demand world-
wide. This could increase from 200 billion kWh in 2010 to
2,000 to 3,000 billion kWh by 2030 (Andrae, 2019; Andrae
& Edler, 2015; Belkhir & Elmeligi, 2018; The Shift Project,
2019). In contrast, other studies calculated that the en-
ergy consumption of data centers was practically constant
in recent years (IEA, 2017; Masanet et al., 2020; Shehabi
et al., 2018). For example, the calculations for 2020 range
from 200 billion kWh to 900 billion kWh.
The wide range of the calculation results shows that there
is still a great need for research and information in the
field of the energy consumption of data centers. From
Borderstep's point of view, neither the pessimistic calcu-
lations yielding very high energy consumption nor the op-
timistic calculations resulting in practically constant en-
ergy consumption in recent years are plausible. The pes-
simistic calculations cannot be supported by the estab-
lished figures for hardware sales and equipment in data
centers. The following facts in particular contradict that
the energy consumption of the data centers has remained
practically constant:
A large number of independent studies (CBECI,
2019; Digiconomist, 2019; Kamiya, 2019; Rauchs
et al., 2018) calculated that Bitcoin mining alone
required about 60 to 70 billion kWh of electrical
energy in 2019. If other cryptocurrencies are also
included, it can be assumed that 70 to 90 billion
kWh/a of electrical energy is currently required
for cryptocurrency mining.
Many large and medium-sized new data centers
are being built worldwide, especially by hyper-
scale cloud providers. According to analysts, new
data center construction has been setting rec-
ords for years. At the four data center locations
London, Frankfurt, Paris and Amsterdam alone,
data center capacities have quadrupled. (CBRE,
2020; CBRE Global Corporate Services, 2017).
So far, hardly any capacities at on-premise data
centers have been reduced in Europe. At pre-
sent, a trend can be seen that data is being mi-
grated from the cloud back into the company's
own data center even as hybrid cloud solutions
are seeing increased use (Alffen, 2019; Vanson-
Bourne, 2019).
Since 2010, the number of servers worldwide has
increased by about 50%. The number of server
sales worldwide has increased very significantly,
especially in 2018 and 2019 (Gartner, 2019,
2020; IDC, 2020).
Data center capacities are experiencing particu-
larly strong growth in the Asian market. A cur-
rent report indicates that data center energy
consumption in China alone reached 161 billion
kWh in 2018 (Greenpeace & North China Electric
Power University, 2019).
The European data center market is also growing
very significantly. Various scientific studies as-
sume that the energy consumption of data cen-
ters in Europe has risen markedly (Bio by Deloitte
& Fraunhofer IZM, 2016; Hintemann, 2019; Pra-
kash et al., 2014). The studies indicate that by
2020, the energy consumption of data centers in
Europe will be about 30% higher than in 2010.
According to estimates by the Borderstep Institute, the
energy consumption of data centers worldwide in 2018
was about 400 billion kWh.
Methodology of the study
The present study was conducted as part of the TEMPRO
project—"Total Energy Management for Professional
Data Centers".
According to the underlying classification, data centers
are defined as all self-contained spatial units such as
server cabinets, server rooms, parts of buildings or entire
buildings in which at least three physical servers are in-
stalled. The development of data center capacities is cal-
culated on the basis of the server equipment in the data
centers and other factors. The different performance clas-
ses of servers are also taken into account here.
The calculations are based on a comprehensive structural
model of the data center landscape in Germany, which
was developed at the Borderstep Institute and is updated
annually (Fichter & Hintemann, 2014; Hintemann et al.,
2010; Hintemann, 2017b; Hintemann & Hinterholzer,
2019; Stobbe et al., 2015). In the model, the data centers
in Germany are described in different size classes in terms
of their different server types, storage systems and net-
work infrastructures. The age structure of the servers and
the energy requirements of the various server types in dif-
ferent operating states are also taken into account. Fur-
thermore, the data center infrastructures such as air con-
ditioning, power supply, UPS, etc. are modeled for differ-
ent size and redundancy classes.
The following sources, among others, were used for the
calculations:
Study "Development of ICT-related electricity de-
mand in Germany"–Study by Fraunhofer IZM and
Borderstep on behalf of the Federal Ministry of Eco-
nomics and Energy (Stobbe et al., 2015).
Current results of studies on the development of the
data center market (CBRE, 2018, 2020; CBRE Global
Corporate Services, 2017; Cisco, 2015, 2016; Gartner,
2020; Hintemann, 2014, 2017a; Hintemann et al.,
2014; Hintemann & Clausen, 2018a, 2018b; Howard-
Healy, 2018)
Data from the market research institute Techconsult
on market development for servers, storage and net-
work components (eanalyzer) (Techconsult, 2014,
2015, 2016)
Data from the market research institutes IDC and
EITO on the market development for servers in Ger-
many and Europe (EITO, 2014; IDC, 2018)
Scientific literature and manufacturer information on
the evolution of the energy consumption of servers,
storage and networking products as well as emerging
data center efficiency technologies.
References:
Alffen, G. (2019, May 20). Cloud-Repatriation—Warum migrieren Un-
ternehmen aus der Public Cloud zurück? silicon.de.
https://www.silicon.de/experten-tipp/cloud-repatriation-
warum-migrieren-unternehmen-aus-der-public-cloud-zu-
rueck
Andrae, A. S. G. (2019). Projecting the chiaroscuro of the electricity use
of communication and computing from 2018 to 2030.
Andrae, A. S. G., & Edler, T. (2015). On Global Electricity Usage of Com-
munication Technology: Trends to 2030. Challenges, 6(1),
117–157. https://doi.org/10.3390/challe6010117
Belkhir, L., & Elmeligi, A. (2018). Assessing ICT global emissions foot-
print: Trends to 2040 & recommendations. Journal of
Cleaner Production, 177, 448–463.
Bio by Deloitte, & Fraunhofer IZM. (2016). Ecodesign Preparatory Study
on Enterprise Servers and Data Equipment. https://publica-
tions.europa.eu/en/publication-detail/-/publica-
tion/6ec8bbe6-b8f7-11e5-8d3c-01aa75ed71a1
Bizo, D. (2019). The Carbon Reduction Opportunity of Moving to Ama-
zon Web Services. https://d39w7f4ix9f5s9.cloud-
front.net/e3/79/42bf75c94c279c67d777f002051f/carbon-
reduction-opportunity-of-moving-to-aws.pdf
CBECI. (2019). Methodology—Cambridge Bitcoin Electricity Consump-
tion Index (CBECI). https://www.cbeci.org/methodology/
CBRE. (2018, March 21). Nachgefragte Leistung europäischer Rechen-
zentren übersteigt erneut 100 MW-Wert. Nachgefragte Leis-
tung europäischer Rechenzentren übersteigt erneut 100
MW-Wert. http://news.cbre.de/nachgefragte-leistung-eu-
ropaischer-rechenzentren-ubersteigt-erneut-100-mw-wert
CBRE. (2020). Europe Data Centres Q4 2019.
https://www.cbre.de/en/global/research-and-reports/fea-
tured-reports-global/featured-reports-emea
CBRE Global Corporate Services. (2017). European Data Centres Market
Review. Q4 2016. https://www.cbre.de/de-de/research/Eu-
ropean-Data-Centres-MarketView-Q4-2016
Cisco. (2015). Cisco Global Cloud Index: Forecast and Methodology
2014-2019. http://www.cisco.com/c/en/us/solutions/collat-
eral/service-provider/global-cloud-index-gci/Cloud_In-
dex_White_Paper.pdf
Cisco. (2016). Cisco Global Cloud Index: Forecast and Methodology
2015-2020. https://www.cisco.com/c/dam/en/us/solu-
tions/collateral/service-provider/global-cloud-index-
gci/white-paper-c11-738085.pdf
Cisco. (2018). Cisco Global Cloud Index: Forecast and Methodology
2016-2021. https://www.cisco.com/c/en/us/solutions/col-
lateral/service-provider/global-cloud-index-gci/white-pa-
per-c11-738085.pdf
Digiconomist. (2019). Bitcoin Energy Consumption Index. Digiconomist.
https://digiconomist.net/bitcoin-energy-consumption
EITO. (2014). EITO Costumized Report for Borderstep. EITO.
Fichter, K., & Hintemann, R. (2014). Beyond Energy: Material Stocks in
Data Centers, Taking Resource Efficiency into account in
Green IT Strategies for Data Centers. Journal of Industrial
Ecology, im Erscheinen. https://doi.org/DOI:
10.1111/jiec.12155
Gartner. (2019, March 18). Gartner Says Worldwide Server Revenue
Grew 17.8 Percent in the Fourth Quarter of 2018, While
Shipments Increased 8.5 Percent. Gartner.
https://www.gartner.com/en/newsroom/press-relea-
ses/2019-03-18-gartner-says-worldwide-server-revenue-
grew-17-8-per-c
Gartner. (2020, March 19). Gartner Says Worldwide Server Revenue
Grew 5.1% in the Fourth Quarter of 2019, While Shipments
Increased 11.7%. Gartner.
https://www.gartner.com/en/newsroom/press-relea-
ses/2020-03-19-gartner-says-worldwide-server-revenue-
grew-5-percent-in-the-fourth-quarter-of-2019-while-
shipments-increased-11-percent
Greenpeace, & North China Electric Power University. (2019). Powering
the Cloud: How China’s Internet Industry Can Shift to Renew-
able Energy (Summary). https://secured-static.green-
peace.org/eastasia/PageFiles/299371/Power-
ing%20the%20Cloud%20_%20English%20Brief-
ing.pdf?_ga=2.134490865.1643020916.1584627591-
1230699852.1584179778
Hao, K. (2019, June 6). Training a single AI model can emit as much car-
bon as five cars in their lifetimes—MIT Technology Review.
https://www.technologyreview.com/s/613630/training-a-
single-ai-model-can-emit-as-much-carbon-as-five-cars-in-
their-lifetimes/
Hintemann, R. (2014). Consolidation, Colocation, Virtualization, and
Cloud Computing – The Impact of the Changing Structure of
Data Centers on Total Electricity Demand. In L. M. Hilty & B.
Aebischer (Eds.), ICT Innovations for Sustainability. Ad-
vances in Intelligent Systems and Computing. Springer.
Hintemann, R. (2017a). Energieeffizienz und Rechenzentren in Deutsch-
land – weltweit führend oder längst abgehängt? - Präsenta-
tion. Netzwerk energieeffiziente Rechenzentren - NeRZ.
https://www.borderstep.de/wp-content/uplo-
ads/2017/07/NeRZ-Studie-Rechenzentrumsmarkt-30-06-
2017.pdf
Hintemann, R. (2017b). Rechenzentren 2016. Trotz verbesserter Ener-
gieeffizienz steigt der Energiebedarf der deutschen Rechen-
zentren im Jahr 2016. Borderstep Institut für Innovation und
Nachhaltigkeit. https://www.borderstep.de/wp-content/up-
loads/2017/03/Borderstep_Rechenzen-
tren_2016_Stand_07_03_2017_finaln-1.pdf
Hintemann, R. (2019, September 10). Energy demand of cloud compu-
ting, development and trends: Data center energy demand.
Workshop on research and technological development
(R&TD) of energy efficiency in cloud computing.
https://www.cloudefficiency.eu/workshop1
Hintemann, R., & Clausen, J. (2018a). Bedeutung digitaler Infrastruktu-
ren in Deutschland. Sozioökonomische Chancen und Heraus-
forderungen für Rechenzentren im internationalen Wettbe-
werb. Berlin. Verfügbar unter. https://www.eco.de/wp-con-
tent/uploads/dlm_uploads/2018/06/DI_Studie.pdf
Hintemann, R., & Clausen, J. (2018b). Potenzial von Energieeffizienz-
technologien bei Colocation Rechenzentren in Hessen. Bor-
derstep Institut für Innovation und Nachhaltigkeit.
https://www.digitalstrategie-hessen.de/rechenzentren
Hintemann, R., Fichter, K., & Schlitt, D. (2014). Adaptive computing and
server virtualization in German data centers—Potentials for
increasing energy efficiency today an in 2020. In Marx
Gómez, Sonnenschein, Vogel, Winter, Rapp, & Giesen (Eds.),
Proceedings of the 28th Conference on Environmental Infor-
matics—Informatics for Environmental Protection, Sustaina-
ble Development and Risk Management (pp. 477–484). BIS.
http://enviroinfo.eu/sites/default/fi-
les/pdfs/vol8514/0477.pdf
Hintemann, R., Fichter, K., & Stobbe, L. (2010). Materialbestand der Re-
chenzentren in Deutschland-Eine Bestandsaufnahme zur Er-
mittlung von Ressourcen-und Energieeinsatz. Studie Im Rah-
men Des UFO-Plan-Vorhabens “Produktbezogene Ansätze in
Der Informations-Und Kommunikationstechnik “(Förder-
kennzeichen 370 893 302), Beauftragt Vom Umweltbundes-
amt.
Hintemann, R., & Hinterholzer, S. (2018, 13.12). Technology radars for
energy-efficient data centers: A transdisciplinary approach
to technology identification, analysis and evaluation. Sus-
tainable Technologies. World Congress. 2018. (WCST 2018).
World Congress on Sustainable Technologies, Cambridge.
https://www.researchgate.net/publica-
tion/330359801_Technology_radars_for_energy-effi-
cient_data_centers_A_transdisciplinary_approach_to_tech-
nology_identification_analysis_and_evaluation
Hintemann, R., & Hinterholzer, S. (2019). Energy Consumption of Data
Centers Worldwide—How will the Internet become Green?
ICT4S, Lappeenranta, Finland. http://ceur-ws.org/Vol-
2382/ICT4S2019_paper_16.pdf
Howard-Healy, M. (2018). Co-location Market Quarterly (CMQ) brief—
Vortrag auf dem BroadGroup’s Knowledge Brunch in Frank-
furt. Broadgroup.
IDC. (2018). Server Market and Enterprise Storage Systems By Country
2014-2017.
IDC. (2020, March 12). Worldwide Server Market Revenue Grew 7.5%
Year Over Year in the Fourth Quarter of 2019, According to
IDC. IDC: The Premier Global Market Intelligence Company.
https://www.idc.com/getdoc.jsp?contain-
erId=prUS46132420
IEA. (2017). Digitalization & Energy. https://www.iea.org/reports/digi-
talisation-and-energy
Jetzke, T., Richter, S., Ferdinand, J.-P., & Schaat, S. (2019). Künstliche In-
telligenz im Umweltbereich: Anwendungsbeispiele und Zu-
kunftsperspektiven im Sinne der Nachhaltigkeit (56/2019;
UBA-Texte). https://www.umweltbundesamt.de/publikatio-
nen/kuenstliche-intelligenz-im-umweltbereich
Kamiya, G. (2019, July 5). Bitcoin energy use: Mined the gap.
https://www.iea.org/newsroom/news/2019/july/bitcoin-
energy-use-mined-the-gap.html
Masanet, E., Shehabi, A., Lei, N., Smith, S., & Koomey, J. (2020, Febru-
ary 28). Recalibrating global data center energy-use esti-
mates | Science. Science. https://science.science-
mag.org/content/367/6481/984
Prakash, S., Baron, Y., Ran, L., Proske, M., & Schlösser, A. (2014). Study
on the practical application of the new framework method-
ology for measuring the environmental impact of ICT -
cost/benefit analysis (p. 373) [Studie]. European Commis-
sion.
Rauchs, M., Blandin, A., Klein, K., Pieters, G. C., Recanatini, M., &
Zhang, B. Z. (2018). 2nd Global Cryptoasset Benchmarking
Study. Available at SSRN 3306125. https://cdn.crowdfundin-
sider.com/wp-content/uploads/2018/12/2018-ccaf-2nd-
global-cryptoasset-benchmarking-study.pdf
Reinsel, D., Gantz, J., & Rydning, J. (2018). The Digitization of the
World. From Edge to Core. IDC (An IDC White Paper,
#US44413318).
Schneider, J., & Ziyal, L. K. (2019). We Need to Talk, AI.
Shehabi, A., Smith, S. J., Masanet, E., & Koomey, J. G. (2018). Data cen-
ter growth in the United States: Decoupling the demand for
services from electricity use. Environmental Research Let-
ters, 13(12). http://iopscience.iop.org/arti-
cle/10.1088/1748-9326/aaec9c
Stobbe, L., Hintemann, R., Proske, M., Clausen, J., Zedel, H., & Beucker,
S. (2015). Entwicklung des IKT-bedingten Strombedarfs in
Deutschland—Studie im Auftrag des Bundesministeriums für
Wirtschaft und Energie. Fraunhofer IZM und Borderstep
Institut. http://www.bmwi.de/BMWi/Redaktion/PDF/E/ent-
wicklung-des-ikt-bedingten-strombedarfs-in-deutschland-
abschlussbericht,property=pdf,bereich=bmwi2012,spra-
che=de,rwb=true.pdf
Strubell, E., Ganesh, A., & McCallum, A. (2019). Energy and Policy Con-
siderations for Deep Learning in NLP.
https://arxiv.org/abs/1906.02243v1
Techconsult. (2014). Daten des eanalyzers. www.eanalyzer.biz
Techconsult. (2015). Daten des eanalyzers. www.eanalyzer.biz
Techconsult. (2016). Daten des eanalyzers. www.eanalyzer.biz
The Shift Project. (2019). LEAN ICT- Towards digital sobriety.
https://theshiftproject.org/en/article/lean-ict-our-new-re-
port/
VansonBourne. (2019). Nutanix Enterprise Cloud Index—Application re-
quirements to drive hybrid cloud growth. https://www.nu-
tanix.com/enterprise-cloud-in-
dex?utm_source=sprout&utm_medium=social
Walsh, T. (2018). Machines that Think: The future of artificial intelli-
gence. Prometheus Books.
Contact:
Dr. Ralph Hintemann
Partner and Senior Researcher
Borderstep Institute for Innovation and Sustainability
Clayallee 323
D-14169 Berlin, Germany
Tel. +49.30.306 45-1005
Fax +49.30.306 45-1009
E-Mail: hintemann@borderstep.de
www.borderstep.de
... For example, Brown et al. 21 provided estimates for 2006 and 2011, Hintemann 46 provided estimates for TWh with a range of 2,000-8,253 TWh to allow for effective scaling of the visualization. All estimates can be found in Table S2. ...
... 5 This US focus has been suggested as a reason why data center energy consumption continues to rise in regions outside of the United States, because the United States has benefited from the improved efficiencies of these facilities. 46,82 The rise of cryptocurrency is another trend that affects energy estimates. From 2008 when the initial Bitcoin whitepaper was published, 83 the energy consumption of proof-of-work crypto technologies has become a highly debated topic. ...
... Are the assumptions used for assessing systems in the past still relevant today? A good example of this type of assessment can be found in Hintemann 46 where the results under discussion are compared with other contradictory publications. This highlights factors such as the relevance of cryptocurrency energy consumption for inclusion within the system boundary and how regional growth may have local effects that are not represented in global statistics. ...
Article
Data centers are a critical component of information technology (IT), providing an environment for running computer equipment. Reliance on data centers for everyday activities has brought increased scrutiny of their energy footprint, yet the literature presents a wide range of estimates with challenging-to-validate calculations that make it difficult to rely on their subsequent estimates. In this review, we analyze 258 data center energy estimates from 46 original publications between 2007 and 2021 to assess their reliability by examining the 676 sources used. We show that 31% of sources were from peer-reviewed publications, 38% were from non-peer-reviewed reports, and many lacked clear methodologies and data provenance. We also highlight issues with source availability—there is a reliance on private data from IDC (43%) and Cisco (30%), 11% of sources had broken web links, and 10% were cited with insufficient detail to locate. We make recommendations to 3 groups of stakeholders on how to improve and better understand the literature—end users who make use of data center energy estimates (e.g., journalists), the research community (e.g., academics), and policy makers or regulators within the energy sector (e.g., grid operators).
... Hintemann argued credibly against too pessimistic (e.g. expected and worst case in [3]) and optimistic scenarios for global data center power by listing indisputable global trends such as cryptocurrency mining, relentless speed of data center construction and cloud to hybrid cloud [4]. Moreover, for 2018 Hintemann estimated as much as 400 TWh for global data center electricity use [4]. ...
... expected and worst case in [3]) and optimistic scenarios for global data center power by listing indisputable global trends such as cryptocurrency mining, relentless speed of data center construction and cloud to hybrid cloud [4]. Moreover, for 2018 Hintemann estimated as much as 400 TWh for global data center electricity use [4]. Then it has been argued that the efficiency gains will continue unhindered between 2022 and 2030 thanks to Artificial Intelligence (AI) [ [5]. ...
Article
Full-text available
The main problems with several existing Information and Communication Technology (ICT) power footprint investigations are: too limited (geographical and temporal) system boundary, overestimation of power saving potential in the next decade, assume that historical power use can predict future global power use in the next decade despite unprecedented data traffic growth, assume that Moore's law relation to digital circuitry can continue "forever" and that no problems with extra cooling power will occur for several decades. The highly variable outlooks for the future power consumptions depend on "starting values", disruptions, regional differences and perceptual estimations of electricity intensity reductions and data traffic increase. A hugely optimistic scenario-which takes into account 20% annual improvement of the J/bit in data centers and networks until 2030 is presented. However, the electric power consumption of the present ICT scope will be significant unless great efforts are put into power saving features enabling such improvements of J/bit. Despite evident risks, it seems though that planned power saving measures and innovation will be able to keep the electricity consumption of ICT and the World under some kind of control. The major conclusion is based on several simulations in the present study-that future consumer ICT infrastructure cannot slow its overall electricity use until 2030 and it will use more than today. Data traffic may not be the best proxy metric for estimating computing electricity. Operations and J/operation seem more promising for forecasting and scaling of bottom-up models.
... Globally, data centers were estimated to use between 196 terawatt hours (TWh) (Masanet et al, 2020) [54] and 400 TWh (Hintemann, 2020) [55] in 2020. This would mean data centers consume between 1-2% of global electricity demand. ...
Chapter
Full-text available
In the past few decades, thanks to the continuous development of machine learning and deep learning algorithms, as well as the continuous improvement of computing power and databases, facial recognition technology (FRT) has developed rapidly. Widespread use of this technology can be seen in all fields of life, such as facepay, individual identification, smart-city surveillance, e-passport or even face to DNA identification. However, some experts believe that certain errors that commonly creep into its functionality and a few ethical considerations need to be addressed before its most elaborate applications can be realized. This article explores the ethical issues of FRT used in different scenarios, tries to examine some legal and regulatory issues that may be encountered during the use of FRT, and technically analyze how to build a trustworthy intelligent system.
Article
Full-text available
The role of the digital technologies are becoming increasingly important in our day-to-day life. Digital technologies have become part of our social life as well as business operations across different industrial and public sectors. Because of digitalization the internet protocol and data traffic have been on the rise for several years. A few studies and estimates have been presented about the increasing future use of energy and electricity. Occasionally consumer behaviors have been claimed to be behind of these increases in electricity use. This report makes an effort to understand the energy and electricity consumption of the Finnish Information sector (ICT-sector) from 2011 until 2018. Additionally, we compare the Finnish electricity use to other European countries with similar data available. Our report shows that energy efficiencies of the information sector have not been retained while the internet protocol and data traffic have been increasing in both fixed and mobile networks.
Article
Full-text available
Informaatiosektorin ja ohjelmistopohjaisten sovellus- ten rooli jokapäiväisessä toiminnassa on yhä merkittävämpi. Ohjelmistot ovat läsnä niin ihmisten välisessä sosiaalisessa yhteydenpidossa kuin eri alojen yritysten liiketoiminnassakin. Viime vuosina on esitetty useita arvioita tietoverkkojen ja informaatiosektorin energiankäytön kasvusta. Datan määrän ja informaatiosektorin lisääntyneen energiankäytön taustalla ovat muuttuneet kuluttajatottumuksemme. Informaatiosektorilla hyödynnettävien teknologioiden energia- tehokkuus ei ole pysynyt kiinteän verkon datan ja mobiiliverkkodatan kasvun tahdissa. Tässä tutkimuksessa käydään läpi, kuinka energian- ja sähkönkäyttö on informaatiosektorilla kasvanut vuosina 2011–2017. Lisäksi kuvataan kuluttajien sähkönkäytön kasvua ja kuluttajatottumusten muutosta. Saadut tulokset ovat osittain ristiriidassa aiemmin esitettyjen arvioiden kanssa.
Conference Paper
Full-text available
The increasing digitalization of the economy and society is leading to a dynamic increase in the amount of data that is processed and stored in data centers. Some scientific studies assume that this growth will also lead to a significant increase in the energy consumption of data centers worldwide. This article deals with the questions of whether and to what extent the energy consumption of data centers have increased in the past and what future developments can be expected. In addition, future challenges to sustainable data center operation will be analyzed. The article deals in particular with the topics of improving the energy efficiency of IT hardware and software, the use of regenerative electricity for the operation of data centers, and the use of waste heat from data centers.
Article
Full-text available
Data centers are energy intensive buildings that have grown in size and number to meet the increasing demands of a digital economy. This paper presents a bottom-up model to estimate data center electricity demand in the United States over a 20 year period and examines observed and projected electricity use trends in the context of changing data center operations. Results indicate a rapidly increasing electricity demand at the turn of the century that has significantly subsided to a nearly steady annual electricity use of about 70 billion kWh in recent years. While data center workloads continue to grow exponentially, comparable increases in electricity demand have been avoided through the adoption of key energy efficiency measures and a shift towards large cloud-based service providers. Alternative projections from the model illustrate the wide range in potential electricity that could be consumed to support data centers, with the US data center workload demand estimated for 2020 requiring a total electricity use that varies by about 135 billion kWh, depending on the adoption rate of efficiency measures during this decade. While recent improvements in data center energy efficiency have been a success, the growth of data center electricity use beyond 2020 is uncertain, as modeled trends indicate that the efficiency measures of the past may not be enough for the data center workloads of the future. The results show that successful stabilization of data center electricity will require new innovations in data center efficiency to further decouple electricity demand from the ever-growing demand for data center services.
Technical Report
Full-text available
Together with a powerful broadband infrastructure, data centers form the backbone of digitization and have a very great influence on current and future economic development. Digitalization may enable an additional potential for value creation in Germany of around €500 billion per year. This is expected as a result of high-performance digital infrastructures and an expanded eco-system of IT service providers, software providers, system houses, digital platforms, content providers, etc. Borderstep investigated the importance of digital infrastructures on behalf of the eco-Association of the Internet Industry Germany (eco-Verband der Internetwirtschaft e.V.). The study concludes that with the digitization of business enterprises, more and more technical knowledge as well as data on customers, suppliers, etc. is being stored in data centers. Data centers have a high sustainability potential and can also make a significant contribution to the transformation of energy systems. To successfully exploit the opportunities of digitalization and to ensure Germany's digital sovereignty, it is necessary to have a strong data center infrastructure in Germany. Like the Scandinavian countries, the Netherlands and Great Britain, the Federal Government should therefore develop an active strategy for securing and expanding data center infrastructures in Germany.
Technical Report
Full-text available
The growth of the data center market is currently mainly driven by cloud computing and colocation. With a focus on the federal state of Hesse, this report has the following tasks: The role of colocation data centers in the data center market, their economic importance and their energy consumption must be determined with a focus on Hesse. Energy saving potentials for colocation data centers have to be identified and estimated. Hindering and promoting factors of the spread of the essential efficiency technologies must be identified and evaluated specifically for data centres in Germany in general and with a focus on colocation providers. The potential of individual energy efficiency technologies (air cooling, water cooling, fuel cells, waste heat utilisation, etc.) for colocation data centres should be evaluated and measures should be proposed on how the potential can be realised in practice. Hesse is very well placed to remain one of the central European locations for high-performance data centers in the future. On behalf of Hessen Trade & Invest GmbH, Borderstep has investigated whether the growth of colocation space in Frankfurt can lead to a strengthening of the location in an international comparison.
Article
Full-text available
This work presents an estimation of the global electricity usage that can be ascribed to Communication Technology (CT) between 2010 and 2030. The scope is three scenarios for use and production of consumer devices, communication networks and data centers. Three different scenarios, best, expected, and worst, are set up, which include annual numbers of sold devices, data traffic and electricity intensities/efficiencies. The most significant trend, regardless of scenario, is that the proportion of use-stage electricity by consumer devices will decrease and will be transferred to the networks and data centers. Still, it seems like wireless access networks will not be the main driver for electricity use. The analysis shows that for the worst-case scenario, CT could use as much as 51% of global electricity in 2030. This will happen if not enough improvement in electricity efficiency of wireless access networks and fixed access networks/data centers is possible. However, until 2030, globally-generated renewable electricity is likely to exceed the electricity demand of all networks and data centers. Nevertheless, the present investigation suggests, for the worst-case scenario, that CT electricity usage could contribute up to 23% of the globally released greenhouse gas emissions in 2030.
Technical Report
Full-text available
While several studies in the recent past have highlighted the role of the Information and Communication technologies (ICT) in mitigating climate change, various studies have also indicated that ICT’s own footprint is expected to increase significantly over the next few years. Hence, the need of efforts from the industry as well as policy makers to implement measures to combat the increasing footprint of ICT becomes eminent. Over the last few years, numerous LCA based methodologies and initiatives have been developed in order to measure and quantify the ICT sector’s footprint and its enabling effect. Although these methodologies have been found to be largely compatible with each other when it comes to assessing ICT's own footprint [6], there is still a lack of transparency and consistency in measuring and reporting the environmental footprint of products, services, networks and companies, which in turn also hinders the objective judgment on the credibility of companies claiming to have done enough in terms of greening their products and corporate activities. As a result, results of environmental impacts of ICT have been seen to vary significantly from one study to another, which impedes the formulation of appropriate policy measures for ICT. Lastly, most of the existing methodologies have not yet been integrated into the ICT-related environmental policy measures. Against this background, this study seeks to perform a quantitative analysis of the current state and an assessment of the future development of the energy consumption of the ICT sector in Europe and identify the most promising options for integrating the methodologies that are applied for measuring GHG emissions and energy footprint of ICT, into concrete policy measures. As overall conclusions of the study, following points can be summarized: - The share of data centres and telecommunication networks in total ICT footprint is expected to increase significantly until 2020, while the total electricity consumption of end-user ICT products is forecasted to stay more or less stable (assuming that forecast growth in the number of devices would be compensated by expected reductions in electricity consumption of individual devices). - There is a lack of policy measures for regulating the energy consumption and GHG emissions of data centres and telecommunication networks, which can partially be attributed to a lack of publicly available data on the energy consumption and GHG emissions of these two sectors. - Existing policy measures do not yet capture all key elements (e.g. use of PCRs, critical reviews) of the methodological framework. Hence, policy measures would require a continuous adaptation of their methodological basis if, for instance the use of PCRs or other important methodological elements (e.g. use of common GHG emission factors, putting higher value to public data bases for generic emission data, GHG of IPCC instead of Kyoto Protocol) are to be made mandatory. Such a step might still not require a revision of framework legislation, but changes in the underlying methodo-logical framework or introduction of a new methodological framework could have a significant effect on the current product-related environment legislation. - Development of harmonized PCRs is more important than developing more detailed methodologies.
Article
In light of the concerted efforts to reduce global greenhouse gas emissions (GHGE) per the so-called Paris Agreement, the Information and Communication Industry (ICT) has received little attention as a significant contributor to GHGE and if anything is often highly praised for enabling efficiencies that help reduce other industry sectors footprint. In this paper, we aim at assessing the global carbon footprint of the overall ICT industry, including the contribution from the main consumer devices, the data centers and communication networks, and compare it with the to the total worldwide GHGE. We conduct a detailed and rigorous analysis of the ICT global carbon footprint, including both the production and the operational energy of ICT devices, as well as the operational energy for the supporting ICT infrastructure. We then compare this contribution to the global 2016-level GHGE. We have found that, if unchecked, ICT GHGE relative contribution could grow from roughly 1–1.6% in 2007 to exceed 14% of the 2016-level worldwide GHGE by 2040, accounting for more than half of the current relative contribution of the whole transportation sector. Our study also highlights the contribution of smart phones and shows that by 2020, the footprint of smart phones alone would surpass the individual contribution of desktops, laptops and displays. Finally, we offer some actionable recommendations on how to mitigate and curb the ICT explosive GHGE footprint, through a combination of renewable energy use, tax policies, managerial actions and alternative business models.
Article
The IT industry in general and data centers in particular are subject to a very dynamic development. Within a few years, the structure and components of data centers can change completely. This applies not only to individual data centers (see [27], in this volume), but also to the structure of the data center market at the national or international level. The sizes, types, and locations of data centers are changing significantly because of trends such as the consolidation of data centers, the increasing use of colocation data centers, virtualization, and cloud computing. The construction of large cloud data centers, for example Google in Finland, Facebook in Sweden, or Microsoft in Ireland, is an example of these developments. In consequence, there is an impact on the overall energy demand of data centers. This chapter discusses these developments and the impact on the overall energy consumption of data centers using the example of Germany.