Address correspondence to: Åsa Moberg Division of Environmental Strategies Research–fms Department of Urban Planning and Environment KTH, Royal Institute of Technology Stockholm, 100 44 Sweden email@example.com
The positive and negative environmental impacts of information and communication technology (ICT) are widely debated. This study assesses the electricity use and greenhouse gas (GHG) emissions related to the ICT and entertainment & media (E&M) sectors at sector level, including end users, and thus complements information on the product level. GHGs are studied in a life cycle perspective, but for electricity use, only the operational use is considered. The study also considers which product groups or processes are major contributors. Using available data and extrapolating existing figures to the global scale for 2007 reveals that the ICT sector produced 1.3% of global GHG emissions in 2007 and the E&M sector 1.7%. The corresponding figures for global electricity use were 3.9% and 3.2%, respectively. The results indicate that for the ICT sector, operation leads to more GHG emissions than manufacture, although impacts from the manufacture of some products are significant. For the E&M sector, operation of TVs and production of printed media are the main reasons for overall GHG emissions. TVs as well as printed media, with the estimations made here, led to more GHG emissions on a global level in 2007 than PCs (manufacture and operation). A sector study of this type provides information on a macro scale, a perspective easily lost when considering, for example, the product-related results of life cycle assessments. The macro scale is essential to capture changes in total consumption and use. However, the potential of the ICT sector to help decrease environmental impacts from other sectors was not included in the assessment.
The concentration of carbon dioxide (CO2) in the atmosphere is increasing and the climate is changing (IPCC 2007). In order to reach the two-degree target (i.e., limit the temperature increase to 2 degrees Celsius), considerable and fairly distributed reductions are needed on a global level. For a country like Sweden, reductions on the order of 85% or more may be necessary by the year 2050 (Åkerman et al. 2007). These are drastic reductions, truly challenging, and will involve all sectors in society. In order to provide a basis for suggestions for improvement and to monitor consequences of actions and everyday practices, measurements and estimates of emissions from different sources need to be made. On a global level, figures on greenhouse gas (GHG) emissions are often aggregated; large and/or energy-intensive industry sectors such as production of different metals, pulp and paper, and food are specifically reported on (IPCC 2007).
The information and communication technology (ICT) sector is currently growing and developing fast. This sector is of interest because it is growing and because it makes it possible to achieve reductions in GHG emissions in other sectors by providing solutions that may substitute travel, transport, physical products, and so forth (e.g., Berkhout and Hertin 2004; Hilty et al. 2006). However, it is not always clear if substitution actually takes place or if new products are added on. Environmental assessments on a product level indicate that user practices may influence the possibility for ICT solutions to achieve environmental improvements (for examples related to media, see, e.g., Moberg 2010). In order to learn more about the environmental impact related to the ICT sector globally, product-related studies should therefore be complemented with macroscale studies considering global environmental impact related to the sector. In this way the environmental impacts of the whole sector can be monitored.
In order to reduce emissions of gases contributing to climate change, it is important to know in which parts of society and connected to what activities the emissions occur. For strategic decision making, it is important for the sector to recognize the contributions from different product groups and from different life cycle phases. A sector analysis can provide information on the total emissions from the sector compared with the total global emissions. It can also be used to identify which product groups and which life cycle phases are most important from an environmental perspective. Policy makers and managers can use this information in order to focus their efforts on the most important sectors, products, and life cycle phases. If the analysis is repeated, the sector analyses can also be used to monitor the development over time. Other forms of analysis could provide further learning, complementing product and sector studies.
There are a number of estimations of the CO2 emissions and energy use of the ICT sector globally (Gartner 2007; Global eSustainability Initiative [GeSI] 2008; Buttazzoni 2008). However, these are often based on rough, unspecified, obsolete data and on extrapolations, without taking into account increasing functionality and efficiency in new technology. The transparency of these reports is also low in some cases (e.g., Gartner 2007). The manufacturing phase of ICT-related products is generally not included or is included in a rough way. There are several studies regarding environmental impacts of ICT-related goods and services specifically (see, e.g., Scharnhorst 2008 for a review of life cycle assessments [LCAs], LCAs in the field of telecommunication). There are also several product-focused studies on electronic and printed media (e.g., Gard and Keoleian 2003; Reichart and Hischier 2003; Toffel and Horvath 2004; Enroth 2009; Moberg et al. 2010). In the scientific literature we know of few studies on the sector level and none that cover the entire sector in detail. There is thus a need for a comprehensive and updated study with greater transparency.
When assessing the environmental impact of a sector, a first important step is to define the sector. Different and incomplete definitions have previously been used for the ICT sector. There is an increasing overlap between the ICT and the entertainment and media (E&M) sectors. The media sector has become a pioneer regarding use of ICT (Lindqvist et al. 2003), and media is indeed a means of communication. Media can be described as content, that is, some kind of message, that is created, distributed, and accessed. To a large extent, access was previously enabled by platforms such as TV, radio, and paper. These platforms are now complemented by personal computers (PCs), mobile phones, and other devices, and distribution is to an increasing degree performed using the Internet. It can be argued that a large proportion of the active use of PCs and the Internet (e.g., gaming, videos, etc.) would be more accurately labeled E&M rather than ICT. The use of the Internet for video and film distribution as well as E&M purposes is increasing rapidly and will increase further. The same trend is noticed for mobile phones and other mobile applications (Anonymous 2009a, 2009b). Conversely, some areas of E&M could be labeled ICT. Still, although there is a clear overlap, there are also distinctive features. There are ICT solutions that are used in other sectors, and there are media platforms that are not primarily based on ICT. It can therefore be useful to study both sectors in parallel.
The aim of this study is to assess global operational electricity use and GHG emissions in the ICT and E&M sectors and compare these with the corresponding total global data. The aim is also to identify the main product groups or processes, as well as life cycle phases, giving rise to major operational electricity use and GHG emissions.
Different approaches can be chosen for assessing the environmental impact of a sector in a life cycle perspective. A distinction can be made between top-down and bottom-up approaches. In top-down approaches, the starting point is generally data from official statistics and the system of environmental and economic accounts, which are allocated to different sectors through the use of input-output (I-O) analysis (Engström et al. 2007). Bottom-up approaches typically start with products and services of the sector, and information from LCAs is aggregated to the sector as a whole (cf. Seppälä et al. 2002). Top-down approaches require statistical data that typically are available nationally. For global assessments, such data are typically not available, and bottom-up studies may be more feasible. Depending on the level of aggregation, top-down studies may also have a drawback in that different product groups are not identifiable. In this study, a bottom-up perspective was mainly used. The starting point was the products and services that are in operation in the sectors. Data were gathered from available studies on parts of the ICT and E&M sectors, including LCAs, reports from companies and organizations, and other studies based on current practices.
Another aspect of sector definition is whether only impact within the sector is considered, or whether a life cycle perspective is used. In the latter case, impact from production and use of products within the sector should be considered, as well as impacts from extraction of raw materials, production of energy carriers, waste management, and so forth. In this study, GHG emissions in the life cycle phases raw material acquisition, manufacturing, distribution, and use were considered. Emissions relating to final disposal of products were not covered. For the ICT sector, business activities (offices, stores), research and development (R&D), employee travel, and so forth were included in the results, but this was not possible for the whole E&M sector. We have not included potential reduction of heating requirements in buildings as a result of the heat generated by computers and other equipment. GHG emissions were covered as far as possible, although in some cases only CO2 emissions were available and therefore used. The emissions from electricity generation, which constitutes a major part of total emissions, include all GHGs.
For estimations of operational electricity, only operation of equipment, including networks, was included. Thus, any electricity use associated with manufacturing and network operator activities was not considered.
This study presents estimations for the year 2007. Since data were not always available for 2007, extrapolations have been made. Most of the data used were from 2005. By choosing 2005 as a baseline year, it was possible to get published data from the same year for the majority of the subsectors investigated, making extrapolations to 2007 more consistent.
We define standby as all modes that are not active (including off mode).
Key parameters for the two sectors are presented in Table 1. More details are given in the next section. Total operational electricity and GHG emissions during use were based on average energy use and emissions per device and year, and number of devices used mid-year 2007 (installed base). For manufacturing, emissions per device produced and total number of devices sold in 2007 were taken into account. The emissions from manufacturing of cables and other equipment with a long lifetime were related to that lifetime. For printed media, the total amount of graphic paper produced globally in 2007 was used as a basis for calculations. It was assumed that no electricity was needed in the use of printed media.
Table 1. Key Parameters for Products Manufactured in 2007 and Total Products in Use Mid-2007
Note: In this table and throughout this article, $= US dollars. PCs = personal computers; LCD = liquid crystal display; CRT = cathode-ray tube; PDP = plasma display panel; GPSs = global positioning systems.
Mobile subscriptions and mobile phones in use
2.7 billion subscriptions in use end of 2006, 3.2 billion subscriptions in use end of 2007. In this study we assumed 2.95 billion subscriptions (assumed equal to mobile phones in use) in mid-2007 (Eskilsen et al. 2009).
250 million subscriptions end of 2006, 350 million subscriptions end of 2007 (Point-topic 2008) Also includes cable-TV broadband. In the study, we assumed 300 million subscriptions and the same number of broadband modems in use in mid-2007.
150 million units (Research and Markets 2008), 100 million modems assumed, based on broadband subscription growth, gives 50 million home routers.
Cordless phones shipped and in use
180 million shipped in 2007 (assumed to be equal to number produced) (MZA Consultants 2008). 800 million assumed in use based on 179 million in use in United States in 2005 (Roth and McKenney 2007), scaled up with U.S./global PC ratio and extrapolated to 2007.
PBXs, faxes, and various business systems in use
No new market research has been carried out. U.S. figures for equipment in use in 2002 (Roth et al. 2002) have been scaled up using U.S./global PC ratio in 2002, no increase assumed.
270 million in 2007, of which: 150 million were desktops, 120 million were laptops, 60% commercial (IDC 2008).
PC monitors shipped
180 million in 2007, of which: 160 million were LCD, 20 million were CRT (Display Search 2008).
200 million divided into: 110 million CRT, 80 million LCD, 10 million PDP (Display Search 2008).
TV peripherals shipped
140 million STBs (set-top-boxes) (Research in China 2009). $10 billion market value assumed, based on figures for 2006 from Britton and McGonegal (2006). 120 million DVD players, $13 billion market value (Herbert 2009). Game consoles, $13 billion market value (IFA 2008). Home theater market value $7 billion. A total of 300 million units, about $50 billion total market value assumed in the study.
Consumer electronic data 2007 based on IFA (2008): Digital cameras & camcorders: 150 million, $37 billion audio devices: $14 billion (excluding home theater systems included as TV peripherals above) Data storage (flash): 650 million, $16 billion Media/mp3 players: 200 million, $15 billion Portable game devices: $10 billion GPSs: $10 billion Other consumer electronics (not identified but assumed to be mainly PC peripherals): $50 billion Data from other sources:Car infotainment: $16 billion (Research in China 2009) Other computer HW: $150 billion ($420 billion for total computer HW minus $270 billion for PCs and servers, based on Eskilsen et al. (2009). The total other E&M HW is estimated to be $260 billion based on all the data above, PC peripherals with the largest uncertainty.
Optical discs shipped
30 billion discs produced in 2007 (with blank CDs) including about 5 billion DVDs (A faltering format 2009). Assumption based on 30 billion CDs in 2004 (BBC 2007), but music CDs declined since then.
Not included in the study
IT services, software, and Internet services/content
Not included in study due to its general nature (can belong to any sector).
Concerts, cinemas, radio broadcasting, and events (sports, theater, etc.)
Not included. Radio devices are included as audio devices in other E&M HW above.
TV broadcasting and satellites
Not included in study, assumed to be small.
A global electricity mix was applied for the use phase, but also in the manufacturing and distribution phases where possible and relevant (Tables 2 and 3). In secondary data sets (data not collected for this study specifically) the original electricity mix was kept. The GHG emissions of the global mix were estimated at 0.6 kilograms carbon dioxide equivalent per kilowatt-hour (kg CO2-eq/kWh) (see appendix S1 in the supporting information on the Journal's Web site).1 This figure was based on 0.5 kg CO2-eq/kWh arising directly from the production of electricity (International Energy Agency [IEA] 2008). The additional 0.1 kg CO2-eq/kWh comes from the fuel supply chain (Herzog 2009), construction work and land use for energy distribution, losses in distribution, and waste management (Frischknecht et al. 1996; Vattenfall 2005).
Table 2. Calculation Overview, Information and Communication Technology (ICT) Sector
Note: Based on key parameters in Table 1 and data in the supporting information on the Journal's Web site. Global electricity emission factor is 0.6 kg CO2-eq/kWh; see appendix S1 in the supporting information. PSTN = public switched telephone network; LAN = local area network; WAN = wide area network.
Mobile networks, operation
16 kWh/subscriber, see appendix S2 in the supporting information on the Journal's Web site. 2.95 billion subscribers in mid-2007. = 50 TWh 0.6 kg CO2-eq/kWh = 30 Mt CO2-eq + 20% CO2-eq from operator activities (offices, vehicles, stores etc.), see appendix S2 in the supporting information on the Journal's Web site. + 14.5 Mt CO2-eq from diesel consumption, and − 4.5 Mt from reduced use of grid electricity when using diesel, based on Ericsson internal information. = 46 Mt CO2-eq
Mobile networks, manufacturing
3 kg CO2-eq/subscriber. Based on Ericsson mobile network LCAs, summarized in Malmodin (2007). Includes (besides network equipment): Steel antenna towers and shelters, concrete foundations, batteries, cooling and power systems, etc.
Mobile phones, operation
3 kWh/average phone, appendix S3 in the supporting information on the Journal's Web site. (40% of battery capacity charged/day + standby scenario: 50% of time). 2.95 billion phones in active use in mid-2007. = 9 TWh * 0.6 kg CO2-eq/kWh = 5 Mt CO2-eq
Mobile phones, manufacturing
18 kg CO2-eq/phone, see appendix S3 in the supporting information on the Journal's Web site. Includes the whole delivered package. 1.15 billion new phones in 2007. = 21 Mt CO2-eq
Fixed networks, operation
45 kWh/subscriber, see appendix S2 in the supporting information on the Journal's Web site. 1.3 billion PSTN lines/subscribers in mid-2007. 300 million broadband lines/subscribers in mid 2007 (cable-TV broadband also included). = 72 TWh * 0.6 kg CO2-eq/kWh = 43 Mt CO2-eq Transport network excluded. + 25% CO2-eq per subscriber from operator activities (offices, vehicles, stores, etc.), see appendix S2 in the supporting information on the Journal's Web site. = 54 Mt CO2-eq
Fixed networks, manufacturing
6 kg CO2-eq/subscriber. Based on Ericsson and TeliaSonera internal LCA studies summarized in Malmodin (2007), results in:5 Mt CO2-eq from cable production and deployment (20% of all cables, the rest are allocated to data centers, enterprise networks, and transport networks), 5 Mt CO2-eq from sites and equipment manufacturing.
Cordless phones, operation
27 kWh/average phone. 800 million in use in mid-2007. = 22 TWh * 0.6 kg CO2-eq/kWh = 13 Mt CO2-eq
Broadband modems & routers, operation
80 kWh/modem/router, see appendix S5 in the supporting information on the Journal's Web site. 300 million modems (assumed equal to broadband lines) and 150 million routers in use 2007. Routers in use assumption based on 200 million in use at end of 2008 (Research in China 2008). = 35 TWh * 0.6 kg CO2-eq/kWh = 21 Mt CO2-eq
PBXs, faxes, and various business systems, operation
See appendix S6 in the supporting information on the Journal's Web site. 35 TWh 0.6 kg CO2-eq/kWh = 20 Mt CO2-eq
End-user telecom equipment, manufacturing
180 million cordless phones 10 kg CO2-eq and 150 million modems and routers 15 kg CO2-eq, see appendix S3 in the supporting information on the Journal's Web site. + 10% CO2-eq from manufacturing assumed for PBXs, faxes and business systems, based on Malmodin (2007). = 6 Mt CO2-eq
250 kWh/average PC, see appendix S5 in the supporting information on the Journal's Web site. 1.05 billion in use mid-2007. 35% laptops, 60% in homes. Operation of extra PC monitors and docking stations/AC-DC adapters standby included. 260 TWh* 0.6 kg CO2-eq/kWh = 158 Mt CO2-eq
360 kg CO2-eq/PC, see appendix S4 in the supporting information on the Journal's Web site. 270 million new PCs in 2007, of which 120 million laptops, also including 160 million LCD monitors and 20 million CRTs. The commercial sector buys 60% of new PCs. = 97 Mt CO2-eq
70 kWh per office PC, appendix S6 in the supporting information on the Journal's Web site. 420 million office PCs, includes all LAN and WAN equipment (hubs, switches, routers). = 29 TWh * 0.6 kg CO2-eq/kWh = 18 Mt CO2-eq
Transport networks, operation
17 TWh, see appendix S2 in the supporting information on the Journal's Web site. Includes all transport networks. 17 TWh * 0.6 kg CO2-eq/kWh = 11 Mt CO2-eq
Data hardware, manufacturing
Estimations based on Ericsson and TeliaSonera internal LCA studies for which a summary is given in Malmodin (2007). 10 Mt from data center equipment and buildings manufacturing. 20 Mt from cable manufacturing and deployment including submarine optical cables (also maintenance). 80% of all cables are accounted for here, the rest are allocated to telecom. The manufacturing of data network equipment (cables are included though) is assumed to be small compared to operation and has not been included.
Table 3. Calculation Overview, Entertainment and Media (E&M) Sector
Note: Based on key parameters in Table 1 and data in appendix. The electricity emission factor is 0.6 kg CO2-eq/kWh; see appendix S1 in the supporting information on the Journal's Web site. CO2-eq: Carbon dioxide equivalent is a measure for describing the climate-forcing strength of a quantity of greenhouse gases using the functionally equivalent amount of carbon dioxide (CO2) as the reference.
200 kWh per TV, see appendix S7 in the supporting information on the Journal's Web site. 1.7 billion TVs in use per mid 2007. Majority still CRT (90%), sales of LCD/PDP overtook CRTs globally in later half of 2007. = 340 TWh 0.6 kg CO2-eq/kWh = 204 Mt CO2-eq
300 kg CO2-eq/TV, see appendix S6 in the supporting information on the Journal's Web site 200 million new TVs in 2007 (110 million CRT, 90 million LCD/PDP). = 60 Mt CO2-eq
TV peripherals, operation
80 kWh per TV, see appendix S7 in the supporting information on the Journal's Web site. Includes set top boxes, DVDs, game consoles, home theater systems and VCRs. Slightly more than one peripheral/TV. U.S. installed base would scale up to about 3.5 billion devices but this approach is not used. Sales ratio TV/devices indicates 2.5 billion devices, but 2 billion devices are assumed in use in 2007. = 82 Mt CO2-eq
TV peripherals, manufacturing
$50 billion, 0.45 kg CO2-eq/$, same emission factor as the average PC assumed. Approx. 300 million devices. = 22 Mt CO2-eq
TV networks, operation
Telecom network results scaled with market value down to $375 billion market value for TV services. = 22 Mt CO2-eq (equals 30 TWh in operation)
Pulp and paper production
0.97 tonne CO2-eq/tonne paper (CEPI 2008) and average 1.2 tonne CO2-eq/tonne paper (Hischier 2007). In this study we assumed 1.1 tonne CO2-eq/tonne paper. 1.1 tonne CO2-eq/tonne paper * 157 Mt paper = 170 Mt CO2-eq Transportation to regional storage, 0.12 tonne CO2-eq/tonne paper (Hischier 2007). 0.12 tonne CO2-eq/tonne paper * 157 Mt paper = 19 Mt CO2-eq
0.74 MWh/tonne paper, including compensation for waste paper 1.17 tonne paper/tonne printed product (Enroth 2006; Larsen et al. 2006). 0.74 MWh/tonne paper * 0.6 tonne CO2-eq/MWh * 157 Mt paper = 69 Mt CO2-eq
Based on the figure for home and office printers from Ecoinvent, 68 kg CO2-eq/printer (Hischier et al. 2007), and the shipped numbers we estimated:10 Mt CO2-eq
Other E&M devices, operation
Includes audio products, cameras, mp3/media players, portable gaming devices, GPS, etc. Based on U.S. consumer electronics (Roth et al. 2007) scaled by number of global PCs/U.S. PCs; see appendix S7 in the supporting information on the Journal's Web site. = 70 TWh * 0.6 kg CO2-eq/kWh = 42 Mt CO2-eq
Other E&M devices, manufacturing
$260 billion, 0.25 kg CO2-eq/$, average emission factor of mobile phones and laptops assumed. = 65 Mt CO2-eq
Optical disc manufacturing
20 billion packed 0.8 kg CO2-eq and 10 billion blank 0.25 kg CO2-eq, emission factors based on Weber et al. (2009), stores and logistics not fully included. = 18.5 Mt CO2-eq
Finally, the operational electricity and emissions of the sectors were related to total global electricity use, CO2 emissions, and CO2-eq emissions to illustrate the respective contribution of these sectors. The global figures used for electricity consumption were 2006 figures from the IEA (2008), extrapolated to approximate 2007 figures. The total global CO2 emissions and CO2-eq emissions were extrapolated to 2007, based on 2005 figures from work by Herzog (2009).
There is no standard definition for what is included in the ICT sector. In this article, ICT is defined as mobile and fixed telecommunication networks (including broadband), data centers, enterprise networks, transport networks, and end user equipment such as phones, PCs, and modems. This corresponds to what is covered by GeSI (2008) and Gartner (2007), with the exception that they included printers, which in this article are included within the E&M sector. What is generally considered to be ICT is included either in the ICT or the E&M sector as defined here. Equipment integrated within other products (e.g., smart grid meters) or used in industrial processes and military equipment, however, is not covered here. Roughly 20% of processors were used for such purposes in 2007. Technological development is fast and ICT is integrated in new products and processes continuously.
The ICT sector as assessed here is divided into four subsectors: fixed telecom, mobile telecom, PCs, and finally data centers, enterprise networks, and transport networks. The division could have been done differently, but based on available data and knowledge we chose a split based on usage by an end user for the first three groups. The fourth, data centers, enterprise networks, and transport networks, was used to gather services that are not easily divided by consumer usage but facilitate the other three groups.
Mobile and Fixed Telecom
Telecom network operation has been considered as a difficult “black box” by previous studies (GeSI 2008). It is difficult to use a bottom-up approach due to the number of different product types and configurations, different traffic loads, cooling and power systems, and so forth. A telecom operator investigation was carried out for the purpose of this study (see appendix S2 in the supporting information on the Journal's Web site). This investigation included operators in North and South America, Europe, and Asia, corresponding to nearly 45% of telecom services revenue globally, 40% of mobile subscribers, and nearly 30% of fixed subscribers, including broadband. The results were then scaled up to global number of subscribers (Eskilsen et al. 2009). The telecom operator investigation covered the electricity used to run the network as well as the operator's overall energy use for buildings (offices, stores, service sites, etc.), their vehicle fleet, and several other activities.
Diesel generators are important, especially for mobile networks, at off-grid sites and as back-up power. The telecom operator investigation underestimated the usage of diesel generators since it is much higher in developing countries. Another estimate based on economic data (Ericsson, internal economic analysis data 2008) was therefore used to complement the investigation.
It should be noted that part of the fixed network also carries a mobile traffic, but it was not possible to specify the amount.
The manufacturing and deployment data for telecom network equipment were based on Ericsson's and TeliaSonera's long experience with LCA in the telecom sector. The LCA data for fixed and mobile networks included, for instance, concrete foundations for antenna towers, optical fiber installation and dismantling, and cooling and power systems, as well as production and operation of the equipment itself. Figures were scaled up to total global emissions based on the total numbers of subscribers (Eskilsen et al. 2009). Most cable manufacturing and deployment is included in data transport related to enterprise networks and transport networks (see Table 2).
End user modems and routers represent a small but important part of the electricity use of telecom end user equipment, partly due to rather high standby consumption (e.g., Zimmermann 2009). The figures for modems and routers used in this study were based on our own measurements of devices, scaled up to global figures using data on shipments (see Table 1).
The estimates of electricity used for operation of PBXs (private exchanges), faxes, and various business systems were based on U.S. values from Roth and colleagues (2002) (see appendix S8 in the supporting information on the Journal's Web site), scaled up to global level based on the U.S./world PC ratio (Roth et al. 2002; Britton and McGonegal 2006).
The manufacture of end user equipment was also covered. A review of mobile phone LCAs has been carried out by Ericsson/Sony Ericsson (see appendix S3 in the supporting information on the Journal's Web site). In the current study, emissions of 18 kg CO2-eq for manufacturing of a mobile phone were assumed (Bergelin 2008). This figure includes all transport, R&D, and other business overhead activities. The manufacture of cordless phones and modems/routers with a short lifetime are based on the same mobile phone manufacturing model, resulting in emissions of 10–15 kg CO2-eq per device. The numbers of mobile and cordless phone devices sold during 2007 were derived from Eskilsen and colleagues (2009) and MZA Consultants (2008), respectively (see Table 1).
For faxes and various business systems, CO2-eq emissions for manufacturing were assumed to be 10% of emissions for operation, based on LCAs of similar energy-intensive products with a longer lifetime (Malmodin 2007), as no specific information was available.
PCs represent the largest share of product value and weight or volume within the ICT sector. A number of studies, both process and I-O LCAs on PCs, were reviewed for this study (see appendix S4 in the supporting information on the Journal's Web site). The review gave information on manufacturing of an average PC in 2005 (an average PC with respect to the ratios stationary/laptop 70/30, CRT/LCD monitor 30/70 and use of extra monitors) (Britton and McGonegal 2006). Projections for 2007 were made, taking into account production improvements (−5%/year) and revised ratios for stationary/laptop (55/45) and CRT/LCD monitors (10/90) (Display Search 2008).
Estimates for PC operation were based on a number of studies (see appendix S5 in the supporting information on the Journal's Web site), primarily field studies and measurements of PCs in operation at offices and in homes (IVF 2007; Roth and McKenney 2007; Zimmermann 2009). Special consideration was given to the relative proportions of stationary/laptop PCs, office/home use, LCD/CRT monitors, operating system power management settings, and use as primary/secondary PC (see Table 1 and appendix S5 in the supporting information on the Journal's Web site). PCs operated by ICT equipment manufacturers and telecom operators are counted twice, as they are included both in business overhead activities in the manufacturing phase of ICT products and in network operation by operators. However, this error is judged to be small.
Data Centers, Enterprise Networks, and Transport Networks
Data centers are large energy users that have seen rapid growth in numbers and size in recent years. Servers are located together with other data network equipment (data storage, routers, and transmission equipment) in data centers, data rooms, or data closets, depending on application and enterprise/organization size. The term data center is used hereafter for all different deployments of servers and data network components; that is, all servers are covered within this subsector. U.S. data on energy use (Koomey 2007a; U.S. Environmental Protection Agency [EPA] 2007) were scaled up to global level based on volume of servers (Koomey 2007b). The United States has nearly 40% of all servers in the world.
Hubs, switches, and routers used in office local area networks (LANs), which provide the network connection for all commercial PCs, were not included in the estimates for data centers described above. An estimated average power consumption of 8 W per connected PC, including cooling and power systems and standby consumption of unconnected ports, was used for enterprise networks.2 Wide area network (WAN) equipment was also included in the estimate for enterprise networks (Roth et al. 2002), although it represents a small share of the total.
Transport network data for transport over longer distances was based on estimated data transport within the United States (Roth et al. 2002). These figures were extrapolated to global level based on number of PCs (Britton and McGonegal 2006; Eskilsen et al. 2009). All data transport, except in access networks, is included here.
The manufacturing of servers, disk storage, routers, other network equipment, required infrastructure such as cooling, power, and back-up systems, and the building itself was estimated based on information collected from central office sites, transport equipment, and cables included in different LCA studies of telecom networks. However, the estimate is crude, and data center infrastructure is an area that needs further investigation. All network cable manufacture (including deployment) was based on U.S. figures on cable per PC scaled up using U.S./global PC-ratio in 2007.
Entertainment and Media
In this paper, the E&M sector is defined as consisting of the subsectors TV (including TV peripherals), printed media, and a range of consumer electronic products (hereafter called other E&M hardware). TV peripherals include set-top boxes, DVD players, game consoles, VCRs, and home theater systems. Other E&M hardware includes audio devices, mp3/media players, digital cameras and camcorders, portable gaming devices, and various computer hardware peripherals. This is comparable with other definitions of media for statistics in Sweden (Nordicom-Sverige 2009) with some exceptions. In the Swedish statistics, the Internet, and in some cases PC and mobile phones, are included as media technology, but in this article they are included in the ICT sector. Cinema, which is also included in the Nordicom statistics, is not covered here. In general, different definitions are used for E&M depending on the question and purpose of study; there is no single standard definition available.
The E&M sector was not fully studied since it is so wide and covers many distribution solutions. Content production and business overhead for music, theater, and sports were excluded, despite the fact that today such events are accessed using TV or PC solutions. In contrast, software production and business overhead activities including logistics, warehouses, and stores were included for the PC and TV industries, based on I-O LCA data.
A review of LCAs for TV was made in order to estimate electricity use and emissions for manufacturing and operation of TVs (see appendix S6 in the supporting information on the Journal's Web site). Emissions from manufacturing of TV peripherals were calculated based on market value of peripherals (see Table 1) and emissions per monetary value for PCs.
Electricity use for TV operation, as for PC operation, was estimated based on previous field studies and measurements of TVs in use (Roth and McKenney 2007; Zimmermann 2009). New larger flat-screen TVs have a higher active power consumption but lower standby consumption. However, their impact was still small in mid-2007. The TV peripherals covered here include primarily set-top boxes, for which about two thirds of the total electricity use is standby, but also game consoles, VCRs, DVD players, and home theater systems (see Tables 1 and 3).
Paper and Printed Media
Calculations regarding printed media included paper production, transport, printing, and distribution. Based on different sources (Hischier 2007; Confederation of European Paper Industries [CEPI] 2008), global emissions of CO2-eq from production of graphic paper were estimated (see Tables 1 and 3 and appendix S9 in the supporting information on the Journal's Web site). The emissions of CO2-eq from graphic paper production can vary substantially depending on the energy source used and the electricity mix of the producing country. The figures used here mainly reflect European conditions and the period 2000–2002. Energy use for printing was roughly estimated based on previous studies regarding offset and gravure printing (Enroth 2006; Larsen et al. 2006) (see Tables 1 and 3 and appendix S9 in the supporting information on the Journal's Web site). The average energy use in industrial printing was used for the total amount of graphic paper produced during 2007 since there were no easily available data on the proportions of graphic paper printed with different techniques industrially, in homes and offices, or not printed at all. The energy use of small home and office printers is generally lower (Hischier et al. 2007). To calculate the GHG emissions, the total energy use for printing was assumed to be electricity and the global electricity mix was used.
The manufacture of industrial printing machinery was not covered in the study. The lifetime of printing presses is long, and the impact was assumed to be comparatively small. The manufacturing of office and home printers was included based on available data in Ecoinvent 2.0 (Hischier et al. 2007). Since the manufacturing figures represented smaller printers, figures were scaled up a little in the final estimation (see Table 3).
A large number of other items of E&M hardware, mainly small portable devices, were considered (examples given above). Roth and McKenney (2007) estimated the operational electricity use of many of these electronic devices in the United States. To scale their figures up to global operational electricity, the ratio of number of PCs in the United States to that in the world was used, as there was no information available on total amount of “other E&M hardware.”
The global manufacturing data concerning these products were estimated based on results from generic I-O LCAs on laptops and mobile phones, giving an average of 0.25 kg CO2-eq/$, and the total market value of $260 billion for “other electronic products” in 2007 (see Tables 1 and 3). In addition, the global production of about 30 billion optical discs (estimation based on British Broadcasting Company [BBC] 2007) was included based on emissions data provided by Weber and colleagues (2009).
An overview of calculations is given for the ICT sector in Table 2 and for the E&M sector in Table 3. Additional comments are presented in the text below. The results of the assessment of operational electricity and total GHG emissions (CO2-eq) for the ICT and E&M sectors are presented in Tables 4 and 5.
Table 4. Operational Electricity (TWh) and Total CO2-eq Emissions (Mt) Relating to Information and Communication Technology (ICT) and Its Subsectors in 2007
Operational electricity [TWh]
Total CO2-eq emissions [Mt]
Note: n.a. = not applicable. Operational electricity for manufacturing not assessed. BB = broadband; PBX = private branch exchange.
Mobile networks, operation
Mobile networks, manufacturing (incl. also sites, etc.)
Mobile phones, operation
Mobile phones, manufacturing
Fixed networks, operation
Fixed networks, manufacturing (incl. also sites, etc.)
Cordless phones, operation
BB modems and routers, operation
PBXs, faxes and various business systems, operation
End-user telecom equipment, manufacturing
Data centers, enterprise networks and transport networks
Data centers, operation
Enterprise networks, operation
Transport networks, operation
Data hardware, manufacturing
Table 5. Total Operational Electricity Use and CO2-eq Emissions for the Entertainment and Media (E&M) Sector
Total CO2–eq emissions [Mt]
Note: n.a. = not applicable. Operational electricity is not assessed for manufacturing.
aOperational electricity only includes the operation of equipment and excludes operator activities or any electricity consumption associated with manufacturing.
bTV peripherals are set-top boxes, DVD players, game consoles, VCRs, and home theater systems.
cOther E&M HW includes audio devices, mp3/media players, digital cameras & camcorders, portable gaming devices, and also various computer HW peripherals.
TV peripherals,b operation
TV peripherals,b manufacturing
TV networks, operation
Pulp and paper production
Printing and distribution
70 (+ 30 for distribution)
Home and office printers, manufacturing
Other E&M Hardwarec
Other E&M devices,c operation
Other E&M devicesc and optical discs, manufacturing
Mobile and Fixed Telecom
The global 2007 figures for electricity used for the operation of mobile and fixed networks are calculated to be 50 TWh and 72 TWh, respectively, and those for the emissions are calculated to be 46 megatons carbon dioxide equivalent (Mt CO2-eq) and 54 Mt CO2-eq, respectively (Table 4).3
The manufacture of mobile phones, operation of broadband modems and routers, as well as operation of faxes, PBXs, and various business systems give rise to emissions of roughly 20 Mt CO2-eq each. In total, end-user equipment is estimated to give rise to emissions of GHG of 59 Mt for operation and 27 Mt for manufacturing (see Table 4).
The total global CO2-eq emissions related to mobile and fixed telecom during 2007 are estimated at 80 and 120 Mt/year, respectively, and operational electricity use for mobile and fixed telecom is estimated at 60 and 160 terrawatt hours (TWh), respectively (see Table 4).4
Global CO2-eq emissions related to PCs during 2007 are estimated to be 250 Mt, and those for operational electricity are estimated to be 260 TWh (see Table 4).
Data Centers, Enterprise Networks, and Transport Networks
Electricity for running data centers is estimated to be 180 TWh, and emissions are estimated to be 108 Mt CO2-eq in 2007 (see Table 4). Operation of enterprise networks is estimated to use 29 TWh of electricity and to give rise to 17 Mt CO2-eq emissions during 2007. Operation of transport networks were estimated to use 17 TWh and give rise to 10 Mt CO2-eq emissions (see Tables 2 and 4). Altogether, the electricity used for operating data centers and enterprise and transport networks in 2007 is estimated to be 230 TWh, and global CO2-eq emissions are estimated to be 170 Mt (see Table 4).
Entertainment and Media
Global electricity use for TVs and peripherals is here calculated to be 510 TWh and CO2-eq emissions related to TVs, 390 Mt during 2007 (Table 5).
Overall, global CO2-eq emissions related to printed media during 2007 are estimated to be 300 Mt (see Table 5).
Other E&M Hardware
The global electricity use for operation of “other E&M hardware” is estimated to be 70 TWh, and the GHG emissions are estimated to be 130 Mt globally in 2007 (see Table 5).
Comparing the Subsectors
In order to get an indication on the main sources of GHG emissions related to the ICT and E&M sectors, the different subsectors are compared. As illustrated in Figure 1, the PC subsector contributes most to the CO2-eq emissions of the ICT sector, followed by telecom, with fixed telecom giving rise to more emissions than mobile telecom. It can be noted that this corresponds to about 1 billion PCs in use and about 3 billion mobile phone subscriptions. Altogether, network operation gave rise to more GHG emissions (235 Mt) than end user operation (210 Mt) and manufacturing (170 Mt). Total emissions related to end user equipment (340 Mt) are slightly higher than emissions related to other equipment (280 Mt). Operation of equipment (end user and network) is responsible for a larger part of the ICT CO2-eq emissions than manufacture (see Figure 1 and Table 4).
Within the E&M sector, the TV subsector gave rise to somewhat more CO2-eq emissions than printed media (Figure 2). Other E&M hardware also contributed to the total CO2-eq emissions of the sector.
In the case of TV, the majority of the CO2-eq emissions arise during end user operation, whereas for other E&M hardware, manufacturing is responsible for more emissions than use of the hardware, according to the calculations and estimations made. For printed media, all emissions calculated are from manufacturing and distribution.
Further, it is interesting to note that both TV and printed media lead to more CO2-eq emissions on a global level than PCs (manufacturing and operation).
This study indicates that the ICT and E&M sectors were responsible for about 1.3% and 1.7%, respectively, of global GHG emissions in 2007 (Table 6). Furthermore, about 3.9% and 3.2% of total global electricity during 2007 was used for operating equipment and networks in the ICT and E&M sectors, respectively.
Table 6. CO2 Emissions, CO2-eq Emissions, and Operational Electricity Use of the Information and Communication Technology (ICT) and Entertainment and Media (E&M) Sectors in Relation to World Figures, 2007
Note: One gigaton (Gt) = 109 tonnes (t) = one petagram (Pg, SI) ≈ 1.102 × 109 short tons. One petawatt-hour (PWh) = 1012 kilowatt-hours (kWh) ≈ 3.6 × 109 gigajoules (GJ).
aLand use for paper production and associated indirect CO2-eq emissions were not included due to high uncertainty.
bWhereof 17% end-user standby-related, 50% network-related and 32% active use (end-user).
cWhereof 36% end-user standby-related, 5% network-related and 59% active use (end-user).
This is the first study, to our knowledge, that presents operational electricity use for the ICT sector and the E&M sector and GHG emissions with a life cycle perspective for the E&M sector. As noted in the introduction, there are some earlier studies for GHG emissions for the ICT sector. However, this study is much more detailed and transparent. Nevertheless, the results are on the same order of magnitude as those of previous studies on emissions from the ICT sector. Gartner (2007) estimated ICT's own carbon footprint to be 2% of direct CO2-emissions (equal to about 1.5% of total CO2-eq). GeSI (2008) and Buttazzoni (2008) estimated ICT's own global GHG emissions to be 2% of total emissions.
The global anthropogenic GHG emissions in 2004, according to IPCC (2007), were caused by energy supply (26%), industry (19%), forestry (17%), agriculture (13%), transport (13%), buildings (8%), and waste and wastewater (3%). However, comparing results from different studies is difficult and should be done with care. The comparison between the ICT sector and the aviation sector made recently (Gartner 2007) is an example of a distorted comparison. The ICT figure (in that study 640 Mt CO2-eq) included the whole life cycle, including all air transport and travel by the ICT industry. The aviation figure, on the other hand, only included jet fuel combustion, no manufacturing or infrastructure, and no other emissions besides direct CO2 (a more correct figure would be 750 Mt for 2007, extrapolated from 733 Mt in 2005 [Gössling and Upham 2009, 30]). Other issues are NOx and high altitude emissions from airplanes; often a factor of 2–5 is used to multiply the CO2-emissions from aviation (Gössling and Upham 2009, 5).
Although the GHG emissions from the ICT and E&M sectors are small compared with total global emissions, climate change work in these sectors is important for several reasons:
• Given the seriousness of the climate change threat and the reductions required, all sectors need to contribute.
• Since these sectors have a tendency for rapid growth, there is a risk that the environmental impacts of the sectors will also grow rapidly.
The focus of this study is global impact of the ICT and E&M sectors. Since global data are not always available, extrapolations have been necessary. For example, data from different studies with different scope were used, and these mainly concerned industrial countries such as the United States, Sweden, and other European countries, and thus extrapolations to global data had to be used. Furthermore, product-specific studies were used as approximations of whole product groups. Data were extrapolated to reflect operational electricity and GHG emissions globally, and results of earlier studies were adjusted to correspond to 2007 values. Furthermore, the definitions of the ICT and E&M sectors are not clear-cut, and differences in delineation can make comparisons with other studies difficult. Thus, the uncertainty is considerable. The quantitative data used in this study do, however, give an indication of the order of magnitude that could be expected in a macro scale study. It can also be noted that some of the data that have a high uncertainty have a limited contribution to the overall result, thus adding only a limited uncertainty to the total result. One such example is where extrapolations are made from U.S. numbers to global data, based on the ratio of PCs in the United States and in the world that is used for calculating the number of cordless phones, PBXs, faxes, and various business systems, data transport in transport networks, and electricity used for operation of E&M hardware. The results are in these cases highly uncertain but have only a limited influence on the total results since the contribution to the total is limited.
GHG emissions from land use related to forestry and paper production were not included in the study due to high uncertainties.
The disposal of electronic and printed products was not handled. Disposal may result in environmental benefits through material recycling and energy recovery, but the disposal process will also lead to negative environmental impacts, and by no means are all products disposed of recycled. Generally, the handling of e-waste is an issue of great concern for both sectors studied. However, in terms of GHG emissions, the disposal phase is a smaller part of the total life cycle (e.g., Choi et al. 2006) and may have a positive effect (decreasing the emissions), for example, if metals are recycled and energy recovered.
GHG emissions are just one environmental impact, and other impacts, environmental as well as social, need further study on a global scale to get a better understanding of the sectors. Examples of environmental impacts that would be of great interest include impacts from the use of hazardous chemicals, from mining of minerals, and from forestry.
Future Trends in the ICT and E&M Sectors
There is rapid development and continuous change in the ICT and E&M sectors. In many cases these developments are crossing the boundaries between the two sectors.
End-user equipment is undergoing continuous development, leading to changing environmental performance per unit. The GHG emissions per TV produced are likely to increase as the equipment becomes more advanced and screens become wider. On the other hand, modems, routers, and wireless local area network (WLAN) equipment are becoming more integrated and energy efficient. GHG emissions due to PC manufacturing are decreasing as laptops are replacing desktop PCs and screens are becoming less energy demanding. New TVs tend to have LCD screens with lower energy use than older screen types, although people tend to buy larger TVs, which increases energy use.
Data traffic has grown rapidly in the past couple of years and is expected to continue to grow. Nearly all of the traffic is associated with PCs with broadband or LAN connections, but mobile broadband data traffic is also increasing.
Previously shared items, for example one TV, telephone, and newspaper per household, are being replaced by personal devices, such as mobile phones and laptops. This trend is counteracting the possible reduced environmental impact per device. In the future, paper media may become a more exclusive and high-status media (Teljas et al. 2007).
The effects of increasing use of ICT for different purposes are impossible to predict, but conscious practices and purposeful use are required to bring about reductions in global energy use and GHG emissions as well as other environmental impacts. The direct impacts can be minimized by using the equipment wisely. In addition, the industry needs to make this wise, energy-efficient behavior easy and the default option. Using sector studies, the environmental impacts of the development can be estimated and monitored.
Improvement Potential in Other Sectors
The ICT sector may be able to contribute significantly to reductions in GHG emissions from other sectors. Examples of applications that may help reduce the environmental impacts of other sectors include virtual presence, e-commerce, dematerialization, e-health, e-learning, e-banking, smart manufacturing and transport systems, smart buildings, smart electric grid, and so forth. ICT may also be applied to itself through server virtualization, thin clients, one laptop/one phone concepts, hardware convergence, and so forth (cf. Erdmann et al. 2004; GeSI 2008). On the other hand, the use of ICT applications may lead to rebound effects, and the environmental benefits may not be straightforward. Berkhout and Hertin (2004, 916) conclude that “ICTs do not necessarily lead to a more environmentally-sound future, but they offer new opportunities to develop more sustainable solutions.” However, this study is accounting and we have not considered the potential for reduction of emissions in other sectors.
This study estimates global electricity during use and GHG emissions in the ICT and E&M sectors. GHGs are calculated using a life cycle perspective, although the disposal phase is excluded. For electricity use, only operation is considered. Use of available data and extrapolation of existing figures to global scale for 2007 indicated that the ICT sector gave rise to 1.3% of 2007 global GHG emissions and the E&M sector 1.7%, while their contribution to global electricity use was estimated to be 3.9% and 3.2%, respectively. The results indicate that considering all operation (end user and network), this is clearly a major part of the ICT GHG emissions, as compared with the manufacturing part, although impacts from manufacturing of some products are significant. For the E&M sector, operation of TVs and printed media are the main reasons for the overall GHG emissions. TVs as well as printed media, with the estimations made here, led to more CO2-eq emissions on a global level in 2007 than PCs (manufacturing and operation).
A sector study such as this provides information on a macro scale, a perspective that is easily lost when considering product-related results of LCAs. The macro scale is essential to capture any increase in total consumption and the risk of nonsubstitution in practice.
This article is a further elaboration of a paper presented at the Electronics Goes Green conference 7–10 September, 2008 (Malmodin 2008). Financial support from Vinnova and other partners to the Centre for Sustainable Communications at KTH—Royal Institute of Technology is gratefully acknowledged. Relevant and valuable comments from anonymous reviewers are greatly appreciated.
One kilogram (kg, SI) ≈ 2.204 pounds (lb). One kilowatt-hour (kWh) ≈ 3.6 × 106 joules (J, SI) ≈ 3.412 × 103 British Thermal Units (BTU).
One watt (W, SI) ≈ 3.412 British Thermal Units (BTU)/hour ≈ 1.341 × 10−3 horsepower (HP).
One megaton (Mt) = 106 tonnes (t) = one teragram (Tg, SI) ≈ 1.102 × 106 short tons.
One terawatt-hour (TWh) ≈ 4.184 × 1012 joules (J, SI) ≈ 3.97 × 109 British Thermal Units (BTU).
About the Authors
Jens Malmodin is a senior research engineer at Ericsson Research in Stockholm, Sweden. Åsa Moberg is a PhD working as a researcher in the Division of Environmental Strategies Research–fms at the Royal Institute of Technology (KTH) in Stockholm, Sweden. Dag Lundén is environmental manager of BA Broadband Services, Networks at TeleSonaria AB in Stockholm, Sweden. Göran Finnveden is a professor in the Division of Environmental Strategies Research–fms at the Royal Institute of Technology (KTH) in Stockholm, Sweden. Nina Lövehagen is a senior research engineer at Ericsson Research in Stockholm, Sweden. All authors are involved in Centre for Sustainable Communications (CESC) at the Royal Institute of Technology (KTH) in Stockholm, Sweden.