Towards a Fossil-Free Internet:
The Fog of Enactment
Priority areas for research in digital sustainability
By Gauthier Roussilhe
Commissioned by the Green Web Foundation
Edited by Chris Adams with additional comments by Michelle Thorne
A decade ago, the Green Web Foundation set out to increase the internet’s energy efficiency and speed up its transition away from fossil fuels. As a Dutch non-profit, we steward the largest open dataset that tracks websites running on renewables—with an open tool suite used over 2 billion times.
We are now focusing on a fossil-free internet by 2030. We commissioned this report from one of the leading independent researchers in digital sustainability, Gauthier Roussilhe, as the opening chapter in a forthcoming series, Towards a Fossil Free Internet, which will explore paths to decarbonize the internet while working towards digital ecosystems that are open, diverse and in service to people’s needs.
The Fog of Enactment
More people are paying attention to the sustainability of the internet and the digital sector. Governments are preparing laws that refer to the environmental impact of digital services. Tech companies are making public commitments to reach goals like Net Zero, carbon neutrality or even becoming carbon positive. However, we see a gap between these announcements and the delivery of their promises.
Today, digital sustainability is in a fog of enactment. First coined by Professor Leah Stokes1 in her research about a transition in the energy sector, the fog of enactment describes the uncertainty that comes after significant public attention and announcements—where the path between pledges and the actual outcome is unclear and being negotiated.
In domains with much uncertainty, where innovation is common and the expertise is highly technical such as the digital sector, you can expect groups to use this ambiguity to shape the field to their advantage. So while we welcome the growing interest in digital sustainability, we risk losing trust in its progress without accessible and repeatable ways to independently assess companies’ claims and without clear steps to accelerate a transition from fossil fuels.
Why we published this report
We need ways to navigate this fog and to ensure climate action is meaningful and impactful. This wayfinding needs to be available not just to the most powerful tech companies, but to small organizations and open source communities who want to green their technology, as well as citizens, scientists, governments and journalists who want to verify claims and keep power in check.
In this report, Gauthier Roussilhe helps to describe this fog of enactment in digital sustainability and propose paths through it. His research builds upon his long-standing expertise especially among Francophone practitioners Through our collaboration, we hope to bring his insights to the English-speaking discourse and to complement it with our own desk research and conversations with policy makers, social movement organizers and industry players.
This report is dedicated to the prerequisites for tracking meaningful progress in the internet’s transition from fossil fuels. In future publications, we will outline possible paths to a fossil free internet by 2030 and where lines of conflict may emerge. We hope you enjoy this first chapter in the series Towards a Fossil Free Internet.
In 2021 digital sustainability became a fast-growing field with an uptick in interest and commitments from the public sector, the private sector and civil society.
We are now in a period after many public commitments have been made and where it is unclear how to move from those high expectations to meaningful policy and action. We call this the fog of enactment in digital sustainability. In this fog, different groups will use the situational ambiguity to shape markets or policy landscapes to their advantage. We need ways to navigate this fog and ensure climate action is meaningful and impactful. Furthermore, this wayfinding needs to be available not just to the most powerful tech companies, but to small organizations and open source communities who want to green their technology, as well as citizens, scientists, governments and journalists who want to verify claims and keep power in check.
In this report we identify three priority challenges to clearing this fog of enactment. These areas are prerequisites for achieving a fossil-free internet by 2030—a target that we will explore further in future reports. While the first two challenge areas are starting to gain some attention, in terms of urgency and the scale of environmental impact, we draw the readers’ attention to the third challenge: the technology sector’s role in accelerating fossil fuel extraction.
Gaps in how environmental impact is modeled
While carbon emissions are important, focussing exclusively on carbon can lead to impact transfer, where harm is caused along other critical dimensions, like water availability.
A lack of reliable, timely, and open environmental impact data is hindering accurate modelling. This has a knock-on effect on policy making, it distorts how costs are allocated in markets, and it restricts innovation.
Critical assessment of industry sustainability claims
Many approaches used to calculate impact are uncritically accepted and focus only on positive environmental effects, without discussing the tradeoffs made. These assessments are not a sound basis for policy decisions, and instead we need assessments that take potential harms into account as well.
Digitization should be seen as an accelerant of existing trends. While policy has to be informed by what is technically possible, technological interventions for sustainability are dependent on local context and existing policy landscapes to have any impact at scale. Policy makers should be prepared to look at the defaults set by existing policies when thinking about how they invest to support players in the digital sector.
Extractive industries and the tech sector: the case of fossil fuels
There is a big difference between measuring the operational emissions of a company and the emissions from the use of its services. For example, there is a very real impact from the induced refresh cycles for hardware products that is beyond what is measured when only looking at a company’s operations.
Technology is enabling fossil fuel extraction. A single contract from a tech company to accelerate fossil fuels extraction can result in carbon emissions that are many times the operational emissions of the company itself. Policy needs to account for this “enablement” of digital products and services.
In the face of the current environmental crisis, the Paris Agreement is a legally binding international treaty on climate change to limit global warming to well below 2, preferably to 1.5 degrees Celsius, compared to pre-industrial levels. The Paris Agreement commits its signatories to transform human activities towards sustainability.
Like all other sectors of the economy, the digital sector is impacted by the Agreement because it will have to (1) reduce its own carbon footprint and (2) adapt to the new global conditions: climate change, pollution, resource availability, trade, and transition policies amongst many others.
- How big is the digital sector’s carbon footprint? It is estimated that the digital sector accounted for between 2.1 and 3.9% of global emissions in 20202, and its emissions are likely to rise. According to the International Telecommunication Union (ITU) the digital sector must reduce its emissions by at least 2.5% per year to fit with the +2°C pathway (or 4.2% per year for the +1.5°C pathway)3. However, this reduction does not take into account the digital sector’s impact from the consumption of mineral resources and water, the production of waste, or the pollution of air, soil and water. Because of this, many estimates about the digital sector’s environmental impact are incomplete, as we’ll unpack more in this report.
- Regarding adaptation, the digital sector’s resource extraction as well as supply and production chains will be affected by the environmental crisis and by the increase in intensity of extreme weather events. Therefore, no matter how much the digital sector reduces its emissions, it will have to adapt its material conditions of production due to a changing climate.
Adapting the digital sector is all the more important in societies that rely heavily on digital systems to run their activities. If the digital sector does not integrate the increase in risks and the need for adaptation, then it is only amplifying its fragility. This is becoming increasingly visible as we can see today that a digital system failure of a few hours, or a semiconductor production crisis can have serious consequences in all sectors. For example, the 2017 British Airways server failure cost the company a reported 80 million pounds or, Renault’s car output decreased by 300,000 this year due to car chip shortage.
The digital sector does not operate in a vacuum, and it is instead impacted by changes on the Earth . Since our planet is transforming its cycles and several planetary limits have been or are being crossed due to human activities, the digital sector will also have to be transformed.
This report outlines some of the gaps and considerations of this transformation.
Gaps in how we model impact
The first environmental assessments of the digital sector began in the early 2000s4. Over the last 10 years, discussion of this topic has increased and, recently, has been more and more covered by mainstream media.
Assessments in this sector are difficult due to the complexity and lack of transparency of companies all along supply chains. Also, there is a lack of training and education regarding the environmental footprint of digital technologies which results in very few people understanding the complexity of the topic.
Therefore, we would like to review the different tools at our disposal and evaluate their use and their limits.
Estimating footprint of a digital service or product
The choice of a method depends on the results we want to achieve and the means at our disposal. Life cycle assessment (LCA), as defined by ISO 14040-445, provides a clearer picture of the manufacture, use and sometimes end-of-life of digital products and services. Furthermore, LCAs give a more detailed understanding of the environmental footprint of a product or service over several criteria: GHG emissions, primary energy consumption, abiotic resource consumption (non-organic, such as mineral resources) and water consumption.
It is very important to use several factors to observe impact transfers (the shift of the environmental burden from one factor to another). For example, some data centres reduce their electricity consumption by cooling with water. In this case the energy consumption is reduced at the expense of an increased water consumption. In another example, a device with more refined materials and components might achieve better energy efficiency during the use phase but at the cost of an increased material footprint during the manufacturing phase.
Doing an LCA requires available and verified data, and this makes the assessment very difficult because quality data about the digital sector is rarely open and often requires the purchase of specialised software. It also needs people who are trained in LCA because the work of inventory, allocation and interpretation is not straightforward. Also, it takes more time and is therefore more expensive.
To conclude, LCAs provide the best understanding of the environmental footprint of a digital service or a product if done by an expert with access to quality data and an allocated budget.
Electricity/carbon-focused models focus on the usage phase of a digital service or system. These models often look at the electricity consumption related to data transfer, computing time and the use of digital equipment6. The electricity consumption is then translated into GHG emissions according to the energy mix involved.
These models are easier to produce because it is more manageable to obtain aggregate data on the processes studied and on the origin of the electricity. However, focusing on a single electricity/carbon factor does not account for impact transfers and does not take into account the impacts of manufacturing, which are generally very high for user equipment. Sometimes, models include manufacturing impacts by assuming that the production of equipment represents around 20% of the overall energy consumption for an average lifespan of IT equipment around 4 to 5 years (most hyperscalers are closer to 3 years, and in some cases even lower).
These models have a higher level of uncertainty compared with LCAs and are not meant to answer the same questions about overall environmental footprint. These models also don’t provide a measurement but give orders of magnitude. This is due to the fact that they mostly use aggregate data and a top-down approach in a sector with complex infrastructures and systems that’s rapidly changing.
The ITU recommendations imply access to a higher number of primary data (from direct measurement at machine-level) in order to comply with their method7. However, very little primary data is openly available, even for LCA experts, and what is available remains with the Original Equipment Manufacturers (OEM). LCAs of digital services tend to move in the direction of the ITU but, given the data available, they can only be partially compliant and focus mostly on ISO 14040-44.
What about global estimates?
Today, three global estimates of the environmental footprint of the digital sector are discussed: Andrae & Enders (2015, Andrae 2020)8, Belkhir & Elmeligi (2018)9, Malmodin & Lundén (2018)10.
All these studies use a mix of LCAs, models, and global sales figures (a mix of bottom-up and top-down). Each estimate has its limits and history.
Andrae’s study from 2015 is known to be too high due to old reference data and extrapolation methods. The author published significantly lower figures in 2020, therefore Andrae & Enders 2015 shouldn’t be used in upcoming scientific literature as Andrae 2020 replaced it. However, the older study provided full access to its data and calculation methods, which explains why it has been widely used in the past.
Belkhir & Elmeligi 2018 also opened their data and calculation methods, but they are known to have overestimated datacenters footprint due to old reference data.
Malmodin & Lundén 2018 used more primary data than others due to authors’ position in Ericsson and Telia and due to partnerships between Ericsson and other major companies. This study provided the lowest estimate and projected that the digital sector’s footprint will lower in the next ten years. However, most of the primary data cannot be reviewed as it was placed under non-disclosure agreements (NDA).
Paradoxically, Malmodin & Lundén’s study is considered as a high-quality one due to the quality of its primary data; yet, the impossibility of reviewing the data and replicating its results limits its reach within the scientific community.
Global estimates are always a trade-off between transparency and quality. Andrae & Enders and Belkhir & Elmeligi use public data and make their calculation transparent; however, their results might be overestimated and lead to uncertainty due to the lack of access to newer data. Malmodin & Lundén had access to high-quality data that couldn’t be disclosed; therefore, their results lead to another kind of uncertainty due to blackboxed data and unreplicable estimates. Also, Malmodin and Lundén work respectively at Ericsson and Telia and the influence of corporate ties has to be assessed.
It has been twenty years since researchers started to assess the environmental footprint of the digital sector. This research has mostly focused on energy consumption and carbon footprint, and yet uncertainty is still high due to the lack of open data and replicable results.
One of the key points to understand this uncertainty is that supply chains for the digital sector are wide and material-intensive. It does not require a high volume of metals to make digital equipment, but it requires a large variety of it. The more metals you use, even in small quantities, the more complex your supply chains get (mines, traders, smelters, refiners, component makers, assembly, etc.). The manufacturing phase of most digital equipment remains opaque to researchers and companies alike.
Finally, the material footprint, water consumption, different types of pollution and End of Life (EoL) are not studied enough. Our current view of the environmental footprint of the digital sector is incomplete and limited to a small perimeter, mainly carbon emissions from electricity consumption. The direction in which the digital sector’s footprint is heading is still unclear. Going from a global point of view to a territorialised one is an encouraging perspective, meaning to position the environmental assessment at a “local” scale (country, city, etc.). This allows us to reduce uncertainty and to assess where we have agency in global supply chains and places of consumption.
Critical assessment of industry sustainability claims
Financialization, the logistics revolution (such as containerization) and digitization have profoundly changed most sectors since the 80s. Financialization made it easier to invest capital in foreign countries. The logistics revolution made it easier to maintain on-time supply chains from distant places of production. Digitization made it easier to communicate with remote teams and coordinate in almost real-time production and supplies. The impact of digitization is therefore widespread, and it’s becoming increasingly hard to assess its direct and indirect effects on human activities.
Looking at direct and indirect effects
In 2016, Nathaniel Horner, a researcher working for the US department of defense known for his work on data centers, proposed a new taxonomy for digitization’s effects15 encompassing previous models16 : Direct effects are the energy use linked to manufacturing, use phase and disposal, and everything else is considered as indirect effects.
At a service level, indirect effects are efficiency gains, substitution (a smartphone replaced many other devices) and a direct rebound effect (efficiency gains can lead to a higher use/consumption, increasing the overall footprint). Indirect rebound effects are also accounted for, both in a positive and negative way: a GPS can both increase traffic in specific areas (direct rebound effect) but also reduce fuel consumption due to more efficient routing (indirect rebound effect). Finally, other indirect effects are economy and society-wide, it relates to the way digitization can change markets and can also change the way “people choose to live and work”.
This framework is useful when it comes to understanding effects of digitization regarding energy use; however, it could be less well fitted for other indicators such as carbon or material footprint. For instance, increasing efficiency on an offshore oil platform’s digital system can lead to an increase of the barrel production, thus increasing carbon emissions when these barrels would be used as fuel. Increasing the production of barrels through efficiency is the intended effect of the digitization process and, in this case, is a direct effect. Carbon emissions due to the use of supplementary barrels are an intended and indirect effect. Both direct and indirect effects are linked to the emission of GHGs, so how does it fit Horner’s taxonomy? Ultimately, the activities for which digitization happens are to be carefully assessed to set up a direct and indirect effects model.
Rebound effects in the digital sector, both positive and negative, should not be taken as a universal mechanism. Context matters and limits the scalability of any given solution. Efficiency gains from a specific technology don’t only depend on the technology itself but on the organization culture, other infrastructures, local policies, local expertise and workforce, etc.
Vlad Coroama and Friedemann Mattern use the term “digital rebound” to define the increase of the environmental footprint of a service or system due to its digitization17. However, they also point out that digitization can happen without rebound for many reasons:
- the digital rebound has a smaller footprint than the non-digitized service;
- the rebound is constrained by available space; the market in which the efficiency happens is already saturated;
- digitization can push activities with a lesser footprint.
“The mechanisms behind rebound effects in general, and thus of digital rebound as well, are essentially non-technical in nature. Their roots reside in economics and in human behavior. It is thus highly unlikely that digital rebound can be addressed solely through technological means. While digitalization does often wait on the side-line, ready to provide efficient substitutes for existing technologies and processes, the avoidance of digital rebound effects needs to be enforced differently, possibly by policy measures.”18
Vlad Coroama and Friedemann Mattern
Knowing that direct and indirect effects of digitization are context-based and policy-dependent, how can we assess its effects for the ecological transition? Facing the urgency of the ongoing crisis, it will be more than necessary to understand where digitization walks the path of adaptation and transition.
Claims of positive impacts
The global positive impacts of digitization for climate change have been mostly linked to two reports from industry representatives : « SMARTer 2030 »19 published in 2015 by the Global e-Sustainability Initiative (GeSI), « The Enablement Effect »20 published in 2019 by Global System for Mobile Communications Association (GSMA) and Carbon Trust. The report from GeSI projects that digitization could reduce global GHG emissions up to 20% by 2030 and the GSMA report assumes that 1 ton of CO2e emitted by the digital sector leads to 10 tons of CO2e avoided in other sectors.
The methods for estimating the positive impacts of the digital sector have already been widely questioned by the scientific community21. The representativeness of the samples used and their extrapolation is generally problematic. A technical report (from the author) reviewed the assumptions, calculation methods and results of these reports and concluded that it is very unlikely the claims in the GeSI and GSMA are correct. The level of uncertainty in such reports would normally preclude their use to guide public policy, but unfortunately they are still widely used by industry. At the request of the European Green Digital Coalition22 (a working group of 26 tech CEOs) a 1.2m euro contract has been awarded by the European Union to produce a methodology to assess the positive impact of digitization on the environment and climate23. We will see whether this report makes an interesting methodological contribution or whether it is an attempt to prove the merits of digitization for the climate.
If digitization effects are context-based and policy-dependent, you can not assume that a successful implementation can be repeated globally in a large variety of situations and legal and political landscapes. This also works the other way around: a failure somewhere doesn’t imply a failure everywhere. However, the extrapolation of positive results of a case study to a global estimate should be questioned.
When it comes to transition and sustainability, digitization can lead to more efficiency, but it doesn’t seem appropriate to reverse a global trend. If a trend goes in the wrong direction (or the right one), digitization effects might emphasize it, not change it. For instance, car sharing apps have been successful in large Chinese urban areas because local governments made procedures for number plates long and costly, thus reducing car ownership and vehicle miles travelled (VMT). In the same fashion, bike sharing apps worked better due to investments in cycling and public transport infrastructures. Once again, policies are a context-setter and digitization acts more like a minor factor increasing efficiency in a given direction, or, as Coroama said, waits on the sideline. Obviously policymaking needs to take into account what is technically possible to make the right decisions (what we can technically do). But this also means policymakers need to think critically about digitization to make the right decisions (what we ought or oughtn’t do) and this will not be provided by companies.
The positive impacts of digitalisation in the context of the ecological transition have yet to be fully explored. It seems more rigorous to study impacts on a case-by-case basis but also to integrate other positive and negative impacts factors beyond avoided emissions. More research is also needed against the claims of the positive impacts of digital technologies on climate.
Extractive industries and the tech sector: the case of fossil fuels
Transition policies prioritize specific actions such as quickly reducing our carbon emissions linked to the use of fossil fuels (energy production, transport, etc.). Keeping that in mind, global warming is only one key issue among many others (loss of biodiversity, acidification of oceans, pollution, etc.). Furthermore, fossil fuel energy is just one part of global extractives industries (mineral extraction, etc.)
We will provide here a clearer picture of the role of the digital sector in this regard.
Reaching carbon neutrality?
Many tech companies are claiming to be or about to be carbon neutral24. In practice, it means that their emissions on their scope 1 and 2 (and business travels/commute) will be reduced to 0 due to cleaner energy and carbon compensation. According to the EPA, “Scope 1 emissions are direct greenhouse (GHG) emissions that occur from sources that are controlled or owned by an organization (e.g., emissions associated with fuel combustion in boilers, furnaces, vehicles)”25 and “Scope 2 emissions are indirect GHG emissions associated with the purchase of electricity, steam, heat, or cooling.”26 The missing scope, the scope 3, represented all “emissions that are the result of activities from assets not owned or controlled by the reporting organization, but that the organization indirectly impacts in its value chain.”27 To say it clearly, most tech giants claim to be carbon neutral on their operations but don’t include manufacturing of all the equipment and their use. Only Apple and Google have declared to work on their scope 3 (suppliers, manufacturers, etc) to achieve carbon neutrality. But it is a long road ahead and they will only achieve carbon neutrality through hypothetical carbon removal projects. To give an example, scopes 1 & 2 ( and business travel/commute) represents 2% of all 2020 Apple’s carbon footprint, and 11% of 2019 Google’s carbon footprint.
Furthermore, can a company claim to be carbon neutral? Carbon neutrality is a matter of sequestering as much carbon as we emit in order to stabilise its concentration in the atmosphere and thus limit the increase in the global temperature of the planet. The French Agency for Ecological Transition reminds that: “The objective of carbon neutrality therefore only really makes sense on a global scale. […] Carbon neutrality – as a balance between GHG emissions and sequestration – cannot be applied at any other scale (sub-national territory, organisation (companies, associations, local authorities, etc.), product or service, etc.) than the planet or the States coordinated through the Paris Agreement.”28 A company can claim to participate in carbon neutrality in the states where it has activities; however, a company cannot claim to be carbon neutral on its own.
Having that in mind, can tech giants largely reduce their GHG emissions? There is no doubt that they will achieve this within the scope they have given themselves (scope 1 & 2). Big tech companies are some of the biggest investors in financial mechanisms to fund renewable energy projects. These mechanisms are called Renewable Energy Certificates (RECs) or Power Purchase Agreements (PPA). While RECs have been considered low-quality contracts, PPAs provide a more trustworthy way to generate renewable energy. According to the International Energy Agency (IEA), the technology sector is the largest investor in PPAs with Google, Facebook and Amazon leading the way in 201929. With that amount of investing tech companies will surely reduce the carbon intensity of the energy they use for their operations. Beyond the market effect, it is interesting to consider whether these investments will only benefit the tech giants or whether these new energies will also be available to local communities. It should also be noted that the tech giants will have significant power generation capacity in the future, and could become energy suppliers.
Digitization for fossil fuel industry?
We need to consider the areas where digital technologies are applied. It is well known that many technology companies provide a digital software and service base to oil, gas and coal companies (machine learning, artificial intelligence, cloud solutions, etc.). So what is the responsibility of these technology companies and how much of the emissions from the fossil fuel industry can be allocated to them?
On the one hand, technology companies provide services to make the operation more efficient: less energy consumed per barrel, less carbon to produce a barrel, maintenance, easy remote monitoring, international collaboration, etc. On the other hand, efficiency gains allow for a production increase. However, fossil energy production is dependent on market movements and production orders (OPEC, etc.) and the efficiency gains enabled by digitization can only have a limited direct rebound effect due to these market specificities. New technologies also make it possible to discover and better quantify new deposits, just as they make it possible to exploit reserves that would otherwise be inaccessible or unexploitable.
Would it be possible to estimate the added emissions due to the digitization of the fossil fuel sector? Using a similar approach to the GSMA and GeSI reports we can do this exercise. McKinsey reports that a major oil and gas producer has increased its production volume by 2% at its offshore oil facilities by using real-time digital analysis of its production equipment30. So what are the potential emissions added by the extra 2% of crude oil extraction? According to the International Energy Agency, 9.64 billion barrels of oil equivalent were extracted from offshore wells in 201631. According to Jing et al., a barrel of crude oil would have an average carbon footprint of 40.7 kgCO2e32 (low estimate for offshore). If digitization can effectively increase offshore oil production by 2% then this adds 192,800,000 barrels per year. We can then estimate that the additional offshore oil production enabled by digital technologies would add emissions equivalent to 7.84 MtCO2e. This is only for producing a barrel of crude oil, not burning it. According to the Petroleum Service Company 81% of a barrel is used to produce gas, diesel, jet fuel and other fuels.33 The EPA assumes that burning an oil barrel emits on average 430,80 kgCO234. Then the previous 192,800,000 extra barrels, once burned, could lead to 67.28 MtCO2.
In 2019, Microsoft announced a new contract with ExxonMobil to extract oil in the Permian Basin in Texas, anticipating that they could increase barrels production up to 50,000 barrels a day, or 18,250,000 a year.35 Using the same numbers as above, we can assume that this contract alone could lead to 0,74 MtCO2 from extraction to 6,36 MtCO2 once barrels are burned. This link between big tech and big oil has been extensively explored by Greenpeace in 2020.36 The enablement effect of digitization doesn’t always go in the right direction for sustainability but these kinds of enablement are never included in industry reports.
It would seem that digitization maintains a strange status quo, allowing this sector to increase its efficiency with a limited effect on production, but also allowing the exploitation of new resources otherwise unreachable. Given the need to organise a rapid exit from fossil fuels, this potential status quo seems unacceptable and the fact that technology companies continue to provide services to this sector appears dangerous. We will still need to produce some fossil energy to enable a transition and to maintain some key activities, and efficient extraction processes will obviously be needed. But allowing the fossil fuel sector to continue to produce at equivalent levels is not the right solution. The Fossil Fuel Non-Proliferation Treaty37 provides the basis for this commitment, and it would be appropriate for technology companies to sign the treaty. Such a commitment would be potentially more useful and effective than talk of carbon neutrality on scope 1 & 2.
Reducing embodied emissions
Given that most of the emissions of large technology companies are in the manufacture of their equipment (embodied emissions), it would seem that the most effective way to actually reduce their emissions would be either to produce less equipment or to consume less. Raw mineral extraction and manufacturing are processes that will be very difficult to “decarbonize” as mining sites will have almost no clean energy to power generators, trucks, milling operations etc. The same will be true for smelters, refiners and manufacturing operations in countries that cannot afford to build new energy infrastructure quickly enough. While some solutions might arise to reduce the carbon footprint of such sectors (fossil-free steel38, power-to-heat generation39), their capacity to scale worldwide seems fairly limited. This means that there is an incompressible part of our means of production that will be powered by fossil fuels. So, we might as well reduce this incompressible part.
For example, Apple, which is one of the tech giants with the most advanced environmental strategy, commits to carbon neutrality or the use of recycled materials but never questions its business model of releasing a new range of products every year. In 2019, 1.36 billion smartphones were produced for an estimated user population of just over 4 billion. We can surely reduce the current production by favoring durable, repairable equipment with a more spaced-out release rate.
It is important to emphasize how crucial repairability of equipment is in slowing down the pace of production and consumption. Fairphone, for smartphones, and Framework, for laptops, show that it is largely possible to move in this direction while still being commercially active.
The development of some equipment needs to be questioned. While displays have greatly improved their energy efficiency, the increase in screen size thwarts any gains. A large increase in the production of connected objects is anticipated. However, the lifespan of connected objects is likely to become problematic due to low production quality and software obsolescence.
In the end, reducing emissions from the operations of technology companies is necessary but largely insufficient. Carbon neutrality, from the point of view of a state or at the global level, can only be achieved by reducing the production of equipment. Similarly, technology companies need to consider their involvement in the activities of fossil fuel companies if they are to be able to reduce their emissions.
Firstly, it is necessary to bring the digital sector within planetary limits. In this respect, GHG emissions cannot be taken as the only indicator; digitisation must be analysed in terms of its life cycle, including the consumption of resources, water and energy, as well as pollution and waste production. However, performing life cycle assessment is constrained by the limited data available. Global models based on energy/carbon have been used so far to bypass this issue but have a bigger uncertainty and can only provide orders of magnitude.
The opacity of companies and also the complexity of supply chains are some of the factors explaining the variability of current numbers. If estimates are inaccurate because of insufficient data, then openness and transparency are cornerstones to get a better picture of the environmental footprint of the digital sector. For now, we only have a limited understanding of what’s at stake, especially in the context of the ongoing environmental crisis.
Rather than solely focusing on energy consumption and GHG emissions during the use phase of digital services and products, it’s highly recommended to increase research regarding the material footprint of ICT and manufacturing phase. From an environmental perspective these topics might need to be a priority. Furthermore, the digital sector relies on a particularly extensive and intense material base of strategic importance. There are many doubts about the production capacity of mines in the years to come and shortages of certain metals can be expected, affecting the whole industrial sector of electronic components and devices production.40 Furthermore, the ability to supply energy and water to critical digital factories may be undermined by extreme weather events (drought, floods, etc.).
The direct and indirect impacts of digitization are still unclear. Successful, or failed, deployments cannot be replicated somewhere else because the relevance of digital systems is context-based and policy dependent. This does not mean that the digitalisation of certain uses, services and systems does not have a positive effect. It is necessary to understand, however, that the effects of a given technology invariably depend on the context in which it is applied.
For example, remote working has a very different effect if you live in the heart of London than if you live in the suburbs, and this effect will vary from city to city depending on the quality of public transport, traffic, etc. It cannot be assumed by default that digitization produces an overall positive and replicable effect. In this regard, the positive claims coming from the industry are based on flawed methodology and overly optimistic scenarios. Therefore, they shouldn’t be taken into account by policymakers and very cautiously by the research community.
In many cases, claims of carbon neutrality by these companies are based on the less impactful scopes of their activities, leaving aside manufacturing and product use. Given the limited scope of their transition, there is little doubt that the tech giants can become “carbon neutral”, at least from an accounting perspective. Reducing the carbon footprint of tech companies in concrete terms will consist of reducing activities that will be incompressibly fossil-fuel intensive, such as the manufacture of hardware. This means producing more durable and repairable equipment and thus reducing the rate of production of such equipment.
There is still a lot of uncertainty about the environmental footprint of the digital sector, and when focusing only on carbon we miss out other impactful factors of the footprint (material footprint, water, manufacturing, end of life). In any case, the digital sector should transform to match with new conditions of Earth due to the environmental crisis. There is a long road ahead for this sector.
This report is made available under a Creative Commons Attribution 4.0 International (CC BY SA 4.0) licence. For media or other inquiries: [email protected].
The Green Web Foundation
The Green Web Foundation is a non-profit based in the Netherlands and Berlin working towards a fossil-free internet by 2030.
We aim to shift narratives to address root causes and imagine more sustainable and just futures for the internet. We support internet practitioners in greening their organizations and technology while learning to center on climate justice. We also build open decision-making infrastructure to track and enable the transition to a fossil-free internet, including stewarding the largest open dataset about websites using renewables, which has been accessed over 2 billion times.
About the author
Gauthier Roussilhe has specialized in researching environmental impacts of digital technologies for 4 years. He is now a PhD candidate at the Royal Melbourne Institute of Technology (RMIT) focusing on digital systems compatible with a world stabilized at +2°C. Based in France, he regularly advises French public institutions such as the National Agency for Ecological Transition (ADEME), the French regulator for Electronic Communications (ARCEP), or the French government digital service.
Rousillhe, Gauthier (2021). Towards a Fossil-Free Internet: The Fog of Enactment. Priority areas for research in digital sustainability. The Green Web Foundation.
Scope 1+2+3 of Big tech companies, market-based & location-based ; data for CSR reports of Google, Apple, Facebook, Amazon and Microsoft
Leah Stokes, Short Circuiting Policy –Interest Groups and the Battle Over Clean Energy and Climate Policy in the American States, (Oxford: Oxford University Press). ↩
Charlotte Freitag et al, “The real climate and transformative impact of ICT: A critique
of estimates, trends, and regulations,” Patterns 2, 2021. ↩
“Recommendation ITU-T L.1470: Greenhouse gas emissions trajectories for the information and communication technology sector compatible with the UNFCCC Paris Agreement,” ITU, 2020. ↩
Franz Berkhout and Julia Bertin, “Impacts of information and communication technologies on environmental sustainability: Speculations and evidence – Report to the OECD”, 2001. ↩
“ISO 14040:2006 – Environmental management — Life cycle assessment — Principles and framework,” ISO, 2006 ; “ISO 14044:2006 – Environmental management — Life cycle assessment — Requirements and guidelines,” ISO, 2006. ↩
Google, “Carbon Footprint reporting methodology”, Google Cloud, 2021 [link]. ↩
“Recommendation ITU-T L.1410: Methodology for environmental life cycle assessments of information and communication technology goods, networks and services,” ITU, 2014. ↩
Anders S. G. Andrae and Tomas Edler, « On Global Electricity Usage of Communication Technology: Trends to 2030 », Challenges 6, 2020, pp. 117-157 ; Anders S. G. Andrae, « New perspectives on internet electricity use in 2030 », Eng. Appl. Sci. Lett. 3, 2020, pp. 19-31. ↩
Lofti Belkhir and Ahmed Elmeligi, « Assessing ICT global emissions footprint: Trends to 2040 & recommendations », Journal of Cleaner Production 177, 2018, pp. 448-463. ↩
Jens Malmodin and Dag Lundén, « The Energy and Carbon Footprint of the Global ICT and E&M Sectors 2010-2015”, Sustainability 10, n° 9, 2018, p. 3027. ↩
Charlotte Freitag et al, “The real climate and transformative impact of ICT: A critique
of estimates, trends, and regulations,” Patterns 2, 2021. ↩
Anders S. G. Andrae, « New perspectives on internet electricity use in 2030 », Eng. Appl. Sci. Lett. 3, 2020, pp. 19-31. ↩
EDNA, “Addendum Report for the Total Energy Model V2.0 for Connected Devices,” IEA 4E EDNA, February 2020, p.19. ↩
Ibid., p.3. ↩
Nathaniel C. Horner et al, “Known unknowns: indirect energy effects of information and communication technology,” Environ. Res. Lett. 11, 2016. ↩
Franz Berkhout and Julia Hertin, “De-materialising and re- materialising: digital technologies and the environment,” Futures 36, 2004, pp. 903–920 ; Lorenz M. Hilty et al, “The relevance of information and communication technologies for environmental sustainability—a prospective simulation study,” Environ. Model. Softw. 21, 2006, pp. 1618–1629 ; Robert Rattle, Computing Our Way to Paradise?: The Role of Internet and Communication Technologies in Sustainable Consumption and Globalization, (Lanham, MD: Rowman & Littlefield) ; Eric Williams, “Environmental effects of information and communications technologies,” Nature 479, 2011, pp. 354–358. ↩
Vlad C. Coroamă and Friedemann Mattern, “Digital Rebound – Why Digitalization Will Not Redeem Us Our Environmental Sins,” ICT4S 2019, 2019. ↩
Ibid., p.8. ↩
GeSI and Accenture Strategy, « SMARTer2030 – ICT Solutions for 21st Century Challenges », GeSI, 2015. ↩
Carbon Trust, « The Enablement Effect – The impact of mobile communications technologies on carbon emission reductions », GSMA, 2019. ↩
Jens Malmodin and Vlad Coroama, “Assessing ICT’s enabling effect through case study extrapolation – the example of smart metering,” Electronics Goes Green 2016+, 2016 ; Jan Bieser and Lorenz Hilty, “Assessing Indirect Environmental Effects of Information and Communication Technology (ICT): A Systematic Literature Review,” Sustainability 10, n°8, 2018. ↩
European Commission, “European Green Digital Coalition,” Website, October 12 2021. [link] ↩
“Belgium-Brussels: European Parliament Pilot Project — European Green Digital Coalition — CNECT/2021/OP/0005,” Supplement to the Official Journal of the EU, May 28 2021 [link]. ↩
Carbon neutrality is about sequestering as much carbon as we emit in order to stabilise its concentration level in the atmosphere and thus limit the increase in global temperature. ↩
EPA Center for Corporate Climate Leadership, “Scope 1 and Scope 2 Inventory Guidance,” EPA, Website, [link]. ↩
EPA Center for Corporate Climate Leadership, “Scope 3 Inventory Guidance,” EPA, Website, [link]. ↩
ADEME, “Les avis de l’ADEME : La neutralité carbone,” [ADEME’s opinions: Carbon neutrality], ADEME, March 202,1 [link]. ↩
George Kamiya, “Data Centres and Data Transmission Networks,” IEA, November 2021, Website [link]. ↩
Ferry Grijpink et al., « How tapping connectivity in oil and gas can fuel higher performance », McKinsey & Company, McKinsey & Company, November 2020 [link]. ↩
IEA, « Offshore Energy Outlook 2018 », IEA, May 2018. ↩
Liang Jing et al., « Carbon intensity of global crude oil refining and mitigation potential », Nature Climate Change 10, 2020. ↩
“What’s in a Barrel of Oil? The 42-Gallon Breakdown”, Petroleum Service Company, February 26 2018, Website, [link]. ↩
EPA, “Greenhouse Gases Equivalencies Calculator – Calculations and References,” EPA, Website [link]. ↩
Microsoft News Center, “ExxonMobil to increase Permian profitability through digital partnership with Microsoft,” Microsoft, February 22 2019, Website [link]. ↩
Greenpeace, “Oil in the Cloud: How Tech Companies are Helping Big Oil Profit from Climate Destruction,” Greenpeace, May 19 2020 [link]. ↩
Volvo, “Volvo launches world’s first vehicle using fossil-free steel,” Volvo, October 13 2021, Website [link]. ↩
Silvia Madeddu et al., “The CO2 reduction potential for the European industry via direct electrification of heat supply (power-to-heat),” Environ. Res. Lett. 15, n°12, 2020. ↩
Olivier Vidal, “Modeling the Long-Term Evolution of Primary Production Energy and Metal Prices,” Mineral Resources Economy 1: Context and Issues, 2021 ; Nedal T. Nassar, “By-product metals are technologically essential but have problematic supply,” Science Advances 1, n°3, 2015 ; Shahana Althaf and Callie W. Babbitta, “Disruption risks to material supply chains in the electronics sector,” Resources, Conservation and Recycling 167, 2021. ↩