This report synthesizes key findings from a diverse range of sources, including academic literature, corporate sustainability initiatives, and emerging environmental tracking tools. Collectively, these documents provide a thorough overview of current methodologies for evaluating the environmental im
...
pacts of artificial intelligence (AI) systems. While several advances in methodology and tooling are evident, the review highlights substantial inconsistencies in how different lifecycle stages of AI are measured, analysed, and reported.
[.] One of the most pressing issues uncovered is the widespread reliance on indirect estimates when assessing energy consumption during the training phase of AI models. These estimates often lack real-time, empirical measurement. Furthermore, equally important lifecycle stages — such as inference (the operational use of models), Scope 3 emissions (from supply chains and hardware manufacturing), and infrastructure-level impacts (such as water consumption and cooling) — remain significantly underexplored. This reliance on proxies introduces substantial data gaps, impedes accountability, and restricts consumers’ ability to make informed, sustainable choices about AI.
To address these issues, the report uses a lifecycle-based approach, dividing the AI system's environmental impact into three stages: 1. Training, 2. Inference, 3. Supply Chain. For each stage, we examine measurement methodologies, identify current limitations, and offer recommendations for key stakeholder groups: developers (producers), users (consumers), and policy-makers. The overarching aim is to ensure that sustainability becomes a foundational element — embedded from the earliest stages of AI design to its deployment and continued use — rather than an afterthought." (Executive summary, pages v-vi)
more
"The development and deployment of large language models like ChatGPT across the world requires expanding data centers that consume vast amounts of electricity. Using descriptive statistics and a multi-country computable general equilibrium model (IMF-ENV), we examine how AI-driven data center growt
...
h affects electricity consumption, electricity prices, and carbon emissions. Our analysis of national accounts reveals AI-producing sectors in the U.S. have grown nearly triple the rate of the private non-farm business sector, with firm-level evidence showing electricity costs for vertically integrated AI companies nearly doubled between 2019-2023. Simulating AI scenarios in the IMF-ENV model based on projected data center power consumption up to 2030, we find the AI boom will cause manageable but varying increases in energy prices and emissions depending on policies and infrastructure constraints. Under scenarios with constrained growth in renewable energy capacity and limited expansion of transmission infrastructure, U.S. electricity prices could increase by 8.6%, while U.S. and global carbon emissions would rise by 5.5% and 1.2% respectively under current policies. Our findings highlight the importance of aligning energy policies with AI development to support this technological revolution, while mitigating environmental impacts." (Abstract)
more
"The fourth edition of the Greening Digital Companies: Monitoring Emissions and Climate Commitments report continues to track greenhouse gas (GHG) emissions, energy use and climate commitments of 200 digital companies. It provides a critical foundation for setting science-based targets (SBTs), measu
...
ring progress over time and addressing the International Telecommunication Union (ITU) role in tracking the environmental impact of the digital sector. As a driver of global innovation and economic growth, the digital sector is increasingly stepping up to its responsibilities in the low-carbon transition. Climate ambition is on the rise, with more companies setting emission reduction targets, increasing use of renewable energy and aligning with SBT frameworks. However, to fully realize this potential, persistent gaps in transparency, reporting and accountability must still be addressed as outlined in this report. The Greening Digital Companies report is based on the most recent full fiscal year for which consistent data can be obtained across all companies—currently 2023." (Executive summary)
more
"Das Versprechen der Digitalen Revolution ist die Heilserzählung unsererZeit. Dieses Buch erzählt eine andere Geschichte: Die des digitalen Kolonialismus. Statt physisches Land einzunehmen, erobern die heutigen Kolonialherren den digitalen Raum. Statt nach Gold und Diamanten lassen sie unter mensc
...
henunwürdigen Bedingungen nach Rohstoffen graben, die wir für unsere Smartphones benötigen. Statt Sklaven beschäftigen sie Heere von Klickarbeiter:innen, die zu Niedriglöhnen in digitalen Sweatshops arbeiten, um soziale Netzwerke zu säubern oder vermeintlich Künstliche Intelligenz am Laufen zu halten. Der Kolonialismus von heute mag sich sauber und smart geben, doch eines ist gleich geblieben: Er beutet Mensch und Natur aus und kümmert sich nicht um gesellschaftliche Folgen vor Ort. Im Wettkampf der neuen Kolonialmächte ist Digitalpolitik längst zum Instrument geopolitischer Konflikte geworden - der Globale Süden gerät zwischen die Fronten." (Verlagsbeschreibung)
more
"Ist Künstliche Intelligenz nur eine weitere Entwicklung der Digitalisierung des Alltags, eine effiziente Software in einer „Cloud“? Die KI-Forscherin Kate Crawford entlarvt diese verharmlosende Vorstellung und beleuchtet die konkreten Auswirkungen der Technologie auf die physische Welt. Ihre R
...
echerche bietet Einblicke in den Bergbau, wo für die KI reale Ressourcen wie Gestein, Lithium-Sole und Erdöl abgebaut werden, und in Logistik-Zentren, wo menschliche Arbeitskraft für Unternehmensgewinne ausgebeutet werde. Künstliche Intelligenz gehe, so Crawford, mit Umweltzerstörung und Raubbau an der Natur sowie sozialer Ungerechtigkeit einher, etwa der Ausbeutung von Clickworkern. Zudem basiere Künstliche Intelligenz auf einer Klassifizierung von Daten, die keineswegs einer objektiven Logik folge, sondern Hierarchien perpetuiere und Ungleichheiten verstärke. Angesichts dieser Verzerrungen könne KI nicht als neutrale Technologie angesehen werden, deren Probleme wiederum durch weitere technische Innovationen gelöst werden können. Vielmehr seien politische Regulierung, demokratische Kontrolle und die Einbeziehung der von Vorurteilen, Diskriminierung und Ausbeutung Betroffenen erforderlich. Die bereits existierenden Gefahren der Technologien für Privatsphäre, Menschenrechte, Natur und Arbeitswelt seien weitaus bedrohlicher als die von manchen befürchtete Entwicklung einer künstlichen Superintelligenz." (Back cover)
more
"Climate action continues to be a key priority for the mobile industry. In 2019, the mobile industry set a goal to reach net zero by 2050, becoming one of the first sectors in the world to set such an ambitious target. This report is the fourth annual assessment of the industry’s progress towards
...
this goal and it provides key recommendations for how mobile network operators, suppliers and governments can work together to accelerate progress across the sector." (Executive summary)
more
"In dieser Studie werden auf Basis der Literaturanalyse eine Vielzahl an Anwendungsfällen mit besonders hohen Potenzialen für positive Umwelteffekte identifiziert. Insbesondere im Energiebereich gibt es eine Vielzahl von vergleichsweise gut erforschten Anwendungsfällen. Hierzu zählt beispielswei
...
se der Einsatz von Smart Metering und preisdynamischen Tarifen. Auch Automatisierung und Monitoring im Gebäudebereich können den Energieverbrauch deutlich verringern. Durch Digitalisierung im Energiesektor kann außerdem das Abregeln erneuerbarer Energien vermieden werden. Im Bereich Verkehr können Technologien wie Internet of Things (IoT) und 5G z. B. die THG-Emissionen im Güterverkehr deutlich senken. Die positiven Umweltpotenziale der Digitalisierung sind jedoch nicht auf das Einsparen von CO2-Emissionen beschränkt. In der Landwirtschaft können z. B. durch den Einsatz von Precision Farming durch digital gestütztes Monitoring der Einsatz von Pflanzenschutzmitteln und Bewässerung reduziert werden. Trotz einer Vielzahl von Studien, die sich mit Umwelteffekten von Digitalisierung beschäftigen, ist der Wissensstand über das Ausmaß der Potenziale in vielen Bereichen jedoch immer noch lückenhaft. In den für diese Studie analysierten Veröffentlichungen wird eine Vielzahl von digitalen Technologien und Anwendungsfällen qualitativ bzw. anekdotisch als sehr relevant beschrieben, es fehlen jedoch in der Regel Quantifizierungen. Obwohl z. B. für den Bereich KI viele Veröffentlichungen identifiziert wurden, enthielten nur wenige quantifizierte Analysen, auch Bilanzierungen fehlten zum Teil. Diese Studien wurden daher nur begrenzt in die Metastudie miteinbezogen, deuten aber auf ein potenziell disruptives Potenzial von KI in manchen Anwendungsbereichen hin (z. B. im Bereich Klimaanpassung kann KI durch die Verbesserung von Prognosen die Reaktionsfähigkeit auf Umweltereignisse erhöhen). Mit der Digitalisierung gehen neben Umweltchancen auch negative Umwelteffekte einher. Zu den negativen Effekten der Digitalisierung gehören direkte Effekte, die durch den Energie- und Ressourcenverbrauch der Produktion und den Betrieb digitaler Infrastruktur entstehen. Ein weiteres prominentes Beispiel sind negative systemische Effekte wie Rebound-Effekte. Daher führt Digitalisierung nicht zwangsläufig zu einer Verringerung des Ressourcenverbrauchs. Positive Umwelteffekte der Digitalisierung gehen oft auf positive Enabling-Effekte wie Optimierungs- und Substitutionseffekte zurück, oder ergeben sich durch den Wandel zu nachhaltigen Verhaltens- und Konsummustern. Im Fokus der Literatur stehen meist positive Enabling-Effekte. Die vorhandenen Quantifizierungen konzentrieren sich meist auf die Potenziale digitaler Technologien (die positiven Enabling-Effekte). Nur in wenigen Studien werden vor- und nachgelagerte Umwelteffekte der Produktion der digitalen Technologien sowie weitere systemische Effekte wie Rebound-Effekte in die Umweltbewertung mit einbezogen. Eine übergeordnete Bewertung der Gesamtbilanz digitaler Technologien im Rahmen wissenschaftlicher Analysen wird somit erschwert. Bei vielen Studien handelt es sich zudem um Fallstudien, die Umwelteffekte unter sehr spezifischen Rahmenbedingungen ermitteln. Es ist daher häufig nicht klar, inwieweit sich diese Potenziale skalieren bzw. auf andere Kontexte übertragen lassen. Schließlich werden Umweltauswirkungen häufig auf CO2-Äq. verkürzt und auf eine breitere Betrachtung von Umwelteffekten im Sinne von Ressourcenverbrauch wird oft verzichtet." (Zusammenfassung, Seite 7-8)
more
"The past several years have seen increased scholarly attention to the concept of ‘extraction’ and ‘extractivism’ as critical frameworks in the humanities and social sciences. These are not only concepts and processes through which to understand material extraction but also expanded formatio
...
ns of extraction as requiring an assemblage of interlocking activities united under an umbrella of exploitative, material economic practices. This article demonstrates that digital data infrastructures, especially data centres, are acting as tools in which to manage the compounding contradictions of paired ‘green,’ digital growth in constrained electricity systems like Ireland. Building on theories of ‘green extractivism’ in the digital sphere and drawing upon fieldwork and policy analysis in data centre industry settings, this article argues that in the form of interconnected climate and digital infrastructures, tech capital is shoring up its role in green transformations, including in the grid systems that will need to adapt away from fossil fuel supply to intermittent energy sources and increasing demand from data centres. In this way, data centres are becoming technologies of green extractivism for overlapping projects of digital grid transformations and climatefriendly capitalism." (Abstract)
more
"The objective of this report is two-fold. First, the report breaks down the energy and emissions profile of the sector and assesses the 30 highest emitting countries for telecommunications while providing global estimates for other ICT sector segments. The report uses a key framework for categorizi
...
ng energy use and emissions, the Greenhouse Gas Protocol Corporate Standard. Scope 1: Emissions are direct emissions from owned or controlled sources; Scope 2: Emissions are indirect emissions from the generation of purchased electricity, steam, heating, and cooling consumed by the firm; Scope 3: Emissions are all indirect emissions, upstream or downstream, (not included in Scope 2) that occur in the firm’s value chain. Second, the report addresses the policy and regulatory implications inferred from this data and the examination of these issues through several country case studies." (Executive summary, page 2)
more
"International standards provide the guidelines and benchmarks needed to measure and improve the environmental impact of AI. Codifying established best practices, standards help mitigate risks such as high energy consumption and lifecycle emissions. They also provide measurement methodologies to ass
...
ess GHG emissions and energy consumption, and thereby identify the actions needed to improve. Achieving this vision of sustainable AI that offers powerful tools for climate action will demand close collaboration among a diverse array of stakeholders from government, industry, academia and civil society. The International Telecommunication Union (ITU) stimulates this collaboration as the United Nations specialized agency for information and communication technologies. This report explores the environmental implications of AI and presents a summary of relevant standards available and under development. It highlights the importance of a coordinated, international approach to standardization and the need for continued engagement and cooperation across all sectors." (Foreword)
more
In 2022, a record 62 billion kg of e-waste was generated globally (equivalent to an average of 7.8 kg per capita per year); 22.3 per cent of this e-waste mass was documented as formally collected and recycled in an environmentally sound manner. In 2010, the world generated 34 billion kg of e-waste,
...
an amount that has since increased annually by an average of 2.3 billion kg. The documented formal collection and recycling rate has increased as well, growing from 8 billion kg in 2010 at an average rate of 0.5 billion kg per year to 13.8 billion kg in 2022. The rise in e-waste generation is therefore outpacing the rise in formal recycling by a factor of almost 5 - driven by technological progress, higher consumption, limited repair options, short product lifecycles, growing electronification and inadequate e-waste management infrastructure - and has thus outstripped the rise in formal and environmentally sound collection and recycling. The e-waste generated in 2022 contained 31 billion kg of metals, 17 billion kg of plastics and 14 billion kg of other materials (minerals, glass, composite materials, etc.) An estimated 19 billion kg of e-waste, mainly from metals like iron which is present in high quantities and has high recycling rates in almost all e-waste management routes, were turned into secondary resources. Platinum-group metals and precious metals were among the most valuable metals but present in much lower quantities; nonetheless, an estimated 300 thousand kg were turned into secondary resources through formal and informal recycling practices.
The share of patent applications for e-waste management rose from 148 per million in 2010 to 787 per million in 2022. Most of those applications were related to technologies for cable recycling, with hardly any signs of an increase in the number of patents filed for technologies related to critical raw materials recovery. Although rare earth elements have unique properties that are crucial for future technologies, including renewable energy generation and e-mobility, the world remains stunningly dependent on the production chains of a few countries. The recycling of such elements remains economically challenging, even in the case of devices with a higher content. Consequently, recycling activities are taking only around 1 per cent of the current demand for the recycling of rare earth elements. The market price for rare earth elements is still too low to support larger-scale commercial recycling operations." (Executive summary, pages 12-13)
more
"This study enabled countries in East Africa to produce national statistics on e-waste generated that are harmonized and comparable across countries because the same methods and tools have been used in the assessment. In addition, households and businesses surveys were piloted in Kenya and Burundi t
...
o find out more on the possession rates of EEE and disposal routes of the e-waste generated within the countries. E-waste generated was calculated separately, using the e-waste generated tools. The two tools, the E-waste generated Tool on one end and the survey on the other, are complementary and provide useful insights on the e-waste generation and management status in the countries and can be used as a basis for policy making purposes, such as designing e-waste collection and management systems. The households and businesses surveys conducted in Kenya and in Burundi revealed that the higher possession rates in Kenya apply to mobile phones, flat display panel televisions and laptops. While in Burundi the products with the highest possession rates are mobile phones and personal care equipment. This indicates that high numbers of these items can be found in the e-waste stream and will be disposed of by the user. Overall, this study highlights that the amount of e-waste in the region is increasing rapidly while there is a lack of e-waste recycling infrastructure in the region. The rapid increase of e-waste generated represents a challenge where the only available formal e-waste recycling facility is located in Rwanda with an annual capacity of approximately 7 kt." (Conclusion)
more
"This report examines the overall statistics of electronic and electrical equipment (EEE) placed on the market (POM) and e-waste generation in Botswana. The main findings indicate that 21 097 tonnes of EEE were placed on the market in Botswana in 2020 generated by demand from households and business
...
es. In 2020, 13 494 tonnes of EEE became e-waste. The highest EEE POM recorded in the last 20 years rose to 24 742 tonnes in 2011. This may be attributed to a boom in construction and refurbishment of sporting facilities ignited by the 2010 World Cup hosted by neighbouring South Africa. In contrast, 2002 had the lowest recorded level of EEE POM. The main reason for the dip in this year is not yet known but it should be highlighted that the year 2002 saw the introduction of the use of the local Pula (BWP) currency from the previous South African Rand (ZAR) currency. The temperature exchange equipment category had the highest share for both POM and E-waste." (Executive summary)
more
"Namibia currently lacks the capacity to monitor e-waste, which makes e-waste hard to track and hampers the setting and assessment of targets that are necessary to move to a more sustainable and circular economy. In line with policy objectives, a draft national policy on management of waste electric
...
al electronic equipment (WEEE) was requested to help build a framework to assess baseline data on volumes of electrical and electronic equipment put on the market and e-waste generated, including types, routes and flows. Challenges encountered in Namibia including the lack of a quantification methodology, legislative processes, and the overall e-waste management are identified and a roadmap with recommendations on how to improve e-waste data quality and availability are presented in this report." (Executive summary)
more
"This is a guide for governments and other public sector organizations that procure information and communications technology (ICT) equipment, systems and services. It sets out the systems and process requirements for ICT procurement in a way that supports the transition to circular and sustainable
...
system solutions. It considers the need for policy and strategy, setting the conditions for and building circular and sustainable design into procurement processes. The guide draws on consultations with governments, suppliers and experts. A validation workshop, attended by 250 participants from 82 countries was held on 14 February, 2023, also contributed to the guide’s contents and themes." (About this guide)
more
"Wer KI als Lösung präsentiert, muss auch Beweise dafür liefern. Wir zeigen in der zweiten Ausgabe unseres SustAIn-Magazins, dass KI-Systeme in der Energieversorgung durchaus den Einsatz von erneuerbaren Energien verbessern können. Aber ihr Potenzial kann nur ausgeschöpft werden, wenn eine dafÃ
...
¼r geeignete Infrastruktur existiert. Auch ist immer wieder zu hören, dass autonom fahrende Kleinbusse gut fürs Klima sind. Meistens wird dann aber nicht erwähnt, dass für den Betrieb dieser Kleinbusse viele Ressourcen benötigt werden. KI-Hoffnungsfantasien bringen uns nicht weiter. Wir müssen uns den ganzen Lebenszyklus von KI-Systemen anschauen, wenn wir ihre Nachhaltigkeit bewerten. Wir müssen aufhören, nur nach den CO2-Emissionen zu fragen, wenn wir die Auswirkungen der Systeme auf die Umwelt untersuchen. Und wir müssen KI-Systeme genau, umfassend und unvoreingenommen analysieren, wenn wir ernsthaft versuchen wollen, KI nachhaltiger zu gestalten." (Editorial)
more
"The amount of Electrical and Electronic Equipment (EEE) Placed on Market (POM) in the Western Balkans region increased from 0.16 Mt (9 kg/inhabitant) in 2010 to 0.21 Mt (13 kg/inhabitant) in 2021. Likewise, the regional e-waste generated nearly doubled from 0.09 Mt (5.3 kg/inhabitant) in 2010 to 0.
...
15 Mt (9.0 kg/inhabitant) in 2021. Large Equipment (Category IVa(1)) and Small Equipment (Category V) are the largest categories of both EEE POM and e-waste generated, at 65% (Category IVa 40%, Category V 25%) and 59% (Category IVa 29%, Category V 30%), respectively. The e-waste generation growth rates in all countries of the region are generally positive, with the exception of Lamps, Small IT, and Screens and Monitors. The Western Balkan countries collected and managed 38.4 kt (2.4 kg/inhabitant) of e-waste in an environmentally sound manner in 2021, which yields a total collection rate of 27%. Environmentally sound e-waste management may be defined as taking all possible steps to ensure that end-of-life products and waste are managed in a manner that will protect human health and the environment, and it involves the separate collection, dismantling, and pollution of hazardous substances and recycling of valuable materials, while other waste-related activities include waste dumping, waste-picking, disposal, etc. and may include the informal sector(2,3). All Western Balkan countries have specific legislation on this currently in force, supporting the environmentally sound management of e-waste. In practice, the environmental sound management of e-waste in the Western Balkans occurs mainly in Serbia, North Macedonia, and Bosnia and Herzegovina." (Executive summary, pages 14-15)
more