This chapter examines the intersection of artificial intelligence (AI) and environmental sustainability and outlines how the German AI ecosystem can strengthen its leading position to leverage AI for rapid decarbonisation and other sustainability goals. It offers an overview of initiatives across federal ministries, states, academia, industry, and civil society. It describes how AI can enable climate action and crucial uses cases in strategic sectors such as energy, transport, industry, and agriculture. The chapter looks at ways Germany can strengthen its AI-sustainability ecosystem through inter-ministerial and interdisciplinary co-operation, knowledge-sharing and AI education. Germany could apply a whole-of-government approach to AI and environmental sustainability. Finally, the chapter covers approaches to measuring and mitigating the environmental impacts of AI compute infrastructure, identifies critical gaps and makes four recommendations for future action.
OECD Artificial Intelligence Review of Germany
9. Spotlight: AI and environmental sustainability
Abstract
The green and digital “twin transitions” promise to leverage digital technologies such as AI for a sustainable future. As a general-purpose technology, AI has the potential to decrease negative environmental impacts and lower emissions by accelerating progress in domains such as smart energy systems and interconnected transportation networks (OECD, 2022[1]). Germany is well-placed to build on its existing research base and initiatives across the AI ecosystem to become a global leader in AI for climate action and environmental sustainability.
Germany was among the first countries to recognise the potential of AI for environmental sustainability in its 2018 national AI strategy and in its 2020 strategy update, which put environmental and climate protection at the heart of new initiatives by “systematically identifying the potential harboured by AI [...] by funding and promoting AI-based instruments to solve specific challenges for sustainable development” (German Federal Government, 2020[2]). The 2020 update focuses on “making AI environmentally sound” by advancing green information and communication technology (ICT) methods and on “AI research to protect the environment and climate” with the stated goal of “funding and promoting AI-based instruments to solve specific challenges for sustainable development” (German Federal Government, 2020[2]). The strategy update also identified specific application areas such as “renewable energies and energy systems, energy efficiency, resource conservation and recycling, water protection and water management, emission control and health, nature conservation and mobility” (German Federal Government, 2020[2]).
Box 9.1. AI and environmental sustainability: Findings and recommendations
Findings
Germany benefits from a range of AI and environmental sustainability initiatives across federal ministries, states, academia, industry, and civil society. However, initiatives are not typically interconnected, and long-term financing remains a challenge.
AI has significant enabling potential for environmental sustainability and can be leveraged by the German government and industry for rapid decarbonisation across sectors.
Home to leading researchers, practitioners, and pioneers in the field, Germany’s AI and environmental sustainability ecosystem is ahead of most countries.
As part of the larger digital compute infrastructure, AI has considerable environmental impacts, such as energy consumption and water use that are not yet systematically measured in Germany.
Recommendations
Increase intergovernmental and interdisciplinary co-operation on AI and environmental sustainability to promote transfer and synergies between initiatives.
Define strategic focus areas and break down silos in sectors such as energy, transport, industry, or agriculture to maximise AI’s enabling effects for environmental sustainability and rapid decarbonisation.
Extend Germany’s leadership position in AI and environmental sustainability through knowledge-sharing and education programmes, the promotion of start-ups and small and medium-sized enterprises (SMEs) and the widening of the focus to include the circular economy, biodiversity and other planetary boundaries.
Expand measurement efforts by the government and compute providers to assess and mitigate the energy, water and resource impacts of AI compute infrastructure.
The AI and environmental sustainability ecosystem
Germany is recognised for its significant AI and environmental sustainability initiatives, involving collaboration across federal ministries, states, academia, industry and civil society. These efforts are underpinned by substantial funding and focused on creating AI applications that support environmental and climate goals. Crucial work areas include leveraging AI for resource efficiency, promoting sustainable mobility, biodiversity preservation, and fostering sustainable agriculture. The country's approach also emphasises inter-ministerial co-operation and the active engagement of civil society to ensure that AI developments align with ecological and social sustainability principles.
Germany benefits from substantial, appropriately funded AI and sustainability initiatives across federal ministries, states, academia, industry, and civil society
Federal governments and ministries
Several federal ministries have ongoing activities and initiatives at the intersection of AI and environmental sustainability. Based on self-reported input from national governments, the OECD.AI Policy Observatory lists Germany amongst the top countries in the number of AI and environment initiatives, together with countries like Norway, Spain and the United Kingdom (UK) (OECD.AI, 2023[3]). The Federal Ministry for the Environment, Nature Conservation, Nuclear Safety and Consumer Protection (Bundesministerium für Umwelt, Naturschutz, nukleare Sicherheit und Verbraucherschutz, BMUV) has published a “five-point programme for AI in support of the environment and climate” with a planned investment of EUR 150 million over five years. This includes a funding initiative for several AI lighthouse projects for climate innovation and resource-efficient AI, a Green AI Hub for German SMEs, and the platform AI Idea Workshop for Environmental Protection that brings together non-governmental organisations, civil society, academia, and start-ups to develop pilot AI projects for a more sustainable society (BMUV, 2023[4]). The German Environmental Agency has recently opened a new Application Lab for AI and Big Data, financed until 2025 through the Federal Government’s economic stimulus and future technologies package. The lab will focus on leveraging AI and big data methods for environmental research and the sustainable use and operation of AI applications. The insights gained from data-based findings can serve as a basis for political decisions and promote a deeper understanding of complex environmental processes in the public (UBA, 2022[5]).
The BMUV pioneered inter-ministerial co-operation with the Federal Ministry of Labour and Social Affairs (Bundesministerium für Arbeit und Soziales, BMAS) and the Federal Ministry for Family Affairs, Senior Citizens, Women and Youth (Bundesministerium für Familie, Senioren, Frauen und Jugend, BMFSFJ) through its Civic Coding initiative. This innovation network aims to design AI applications that are social, sustainable and participative, such as the AI Idea Workshop for Environmental Protection, a Civic Data Lab for data collection based on the common good and the Civic Innovation Platform (Civic Coding, 2023[6]).
The Federal Ministry of Transport and Digital Infrastructure (Bundesministerium für Digitales und Verkehr, BMDV) promotes the use of AI for earth observation and funds model projects for sustainable mobility through its Artificial Intelligence and Mobility (AIAMO) project in partnership with local municipalities (BMDV, 2023[7]). The Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung, BMBF) has financed an application hub for a circular economy for plastic packaging through AI methods with EUR 30 million until 2025 (BMBF, 2021[8]) and has initiated a funding programme for research projects on AI as a tool for biodiversity preservation with a funding volume of up to EUR 20 million (BMBF, 2023[9]). The Federal Ministry for Economic Cooperation and Development (Bundesministerium für wirtschaftliche Zusammenarbeit und Entwicklung, BMZ) addresses environmental sustainability and AI through its FAIR Forward project for sustainable development, which provides partner countries in the Global South with access to “climate-smart” agricultural advice and has developed a practitioner’s guide to green data centres in co-operation with the World Bank, the International Telecommunication Union (ITU), and the German Agency for International Cooperation (Deutsche Gesellschaft für Internationale Zusammenarbeit, GIZ) (ITU/World Bank, 2023[10]). The Federal Ministry of Economic Affairs and Climate Action (Bundesministerium für Wirtschaft und Klimaschutz, BMWK) funded projects like AI for advanced material science, predictive maintenance, and advanced groundwater analysis, while the Federal Ministry of Food and Agriculture (Bundesministerium für Ernährung und Landwirtschaft, BMEL) promotes the uptake of AI in smart and sustainable agriculture and rural areas with 36 co-operative projects with a budget allocation of EUR 44 million (BMEL, 2021[11]).
Federal states (Länder)
Various federal states (Länder) also run initiatives at the intersection of AI and environmental sustainability. For example, the Bavarian AI Agency co-ordinates projects on sustainable mobility and addresses environmental challenges through the Hub on Intelligent Robotics and the Hightech Agenda Bavaria (Baiosphere, 2023[12]). The Hessian Centre for AI is pursuing the build-up of sustainable AI compute infrastructure, hosting its supercomputer in the Green IT Cube of GSI Helmholzzentrum, one of Europe’s most efficient and sustainable data centres (hessian.AI, 2023[13]). The competence platform KI.NRW in North Rhine-Westphalia supports lighthouse projects such as AI for flood protection and control, AI for combatting plastic litter in oceans and rivers, and AI for earth system data and environmental prediction (KI.NRW, 2023[14]). These examples showcase the substantial innovation ecosystems for environmentally beneficial AI systems and point to potential benefits and synergies from increased co-operation, both within federal states and between federal, state, and local authorities. The permanent conference of federal state digital ministers announced in November 2023 could serve as a potential forum for such co-operation efforts on AI and environmental sustainability (STMD, 2023[15]).
Academia and research institutes
Germany is home to some of the world’s leading researchers and practitioners in the AI and environmental sustainability field. The Sustainable AI Lab at Bonn University aims to measure and assess the environmental impacts of AI, including in the context of the United Nations’ Sustainable Development Goals (SDGs) (Sustainable AI Lab, 2023[16]). The German Research Centre for Artificial Intelligence (Deutsches Forschungszentrum für Künstliche Intelligenz, DFKI), one of the German Centres of Excellence for AI Research, bundles expert knowledge on the topic through its Competence Centre AI for Environment and Sustainability, DFKI4planet. The competence centre focuses on knowledge transfer and AI development for diverse environmental applications such as pollution detection, green mobility, circular economy, and resource conservation (DFKI, 2023[17]). At Technical University Munich and University of Applied Sciences of Munich, the sustAInability project educates interdisciplinary master students on social and environmental challenges at the intersection of AI and sustainability. Students also research and develop prototypes in various environmental application areas (sustAInability, 2023[18]). The BMUV’s AI lighthouse initiative has also funded two projects on a green consumption assistant and circular textile intelligence at Technical University Berlin in co-operation with the Einstein Centre Digital Future (ECDF) (TU Berlin, 2020[19]). The University of Tübingen and the Hasso Plattner Institute (Hasso-Plattner-Institut, HPI) are part of ELIAS AI consortium, which aims to establish Europe as a leader in AI research for sustainable development (ELIAS, 2023[20]).
Civil Society
Another strength of the German AI and environmental sustainability ecosystem is the active involvement of civil society organisations that provide crucial perspectives on the ecological and social sustainability of AI systems. For example, AlgorithmWatch and the Institute for Ecological Economic Research (Institut für ökologische Wirtschaftsforschung, IÖW) created SustAIn, a sustainability index for AI systems funded by the BMUV. Based on the 17 SDGs, SustAIn developed a set of indicators and metrics for measuring the sustainability of AI along its lifecycle (Rohde et al., 2021[21]). Other civil society organisations like the Green Web Foundation, Wikimedia Foundation, Germanwatch, and the Institute for Applied Ecology (Öko-Institut) have all called for a more sustainable approach to digitalisation and AI development (Bits & Bäume, 2021[22]).
Industry and start-ups
AI plays an important role for large German industrial companies and their environmental sustainability goals, both for improving energy efficiency and reducing a company’s own environmental footprint and for providing innovative product and service solutions for customers. For example, the manufacturing company Siemens employs AI in a variety of industrial environmental sustainability solutions and regards the technology as a key enabler for environmentally sustainable infrastructure. Co-ordinated by a large core AI technology team and the Siemens AI Lab, it leverages AI for predictive maintenance, sustainable construction and building management, digital twins, and intelligent transport networks (Siemens, 2023[23]). The chemical company BASF uses AI for targeted prototyping and digital farming solutions, improving agricultural yields and developing new crops that are environmentally robust in the face of rapid climate change (BASF, 2023[24]). Automotive industry companies like Porsche, Audi, and Volkswagen use AI to identify environmental sustainability risks in their supply chain through a monitoring system that produces automatic warnings about environmental risks the entire procurement system and low-level supply chains (Porsche, 2021[25]).
Some of Germany’s most valuable and innovative start-ups and scale-ups put AI and environmental sustainability at the core of their business models. For instance, Enpal provides solar energy solutions for customers and uses AI for installation services and home energy management (Enpal, 2023[26]). Software company TWAICE offers an AI-supported battery analytics platform to simulate battery behaviour and improve its lifetime while also building solutions for battery development, energy storage systems, and electric vehicle operations (TWAICE, 2023[27]). Celonis, a global pioneer in process mining, increasingly leverages process-specific machine learning models for environmental sustainability transformation solutions including emission reduction and order management (Celonis, 2023[28]). The Greentech Alliance brings together many of these start-ups to support them with venture capital (VC) and entrepreneurial advice (Greentech Alliance, 2023[29]), while the German AI Association has established a working group on climate and environmental sustainability to disseminate knowledge amongst its members (German AI Association, 2023[30]). Entrepreneurship innovation centres like the TUM Venture Lab have recognised the importance of environmental sustainability and have introduced dedicated labs that “boost the translation of deep tech into scalable, circular businesses” for “sustainable environmental impact” (UnternehmerTUM, 2023[31]). An example of a well-functioning start-up ecosystem is OroraTech, a provider of space-based and AI-driven thermal intelligence for wildfire prediction and mitigation that spun out of TUM and was supported by grants from the Bavarian and German government and a consortium of European VC investors (OroraTech, 2021[32]).
Use cases for environmental sustainability and rapid decarbonisation
Germany's AI strategy integrates environmental sustainability across multiple sectors. The BMUV's five‑point programme commits EUR 150 million to AI for the environment, supporting initiatives like AI lighthouse projects and a Green AI Hub. The collaboration spans federal ministries, Länder, academia, civil society and industry, aiming to utilise AI for sustainable solutions in areas like climate innovation, healthcare, and agriculture. The strategy encourages cross-sectoral co-operation and public-private partnerships to advance environmentally sustainable AI applications.
The German government and industry can leverage AI’s huge enabling potential for environmental sustainability and rapid decarbonisation across sectors
Decarbonisation across sectors
AI can play a key role in achieving the German government’s climate and environmental goals. AI can be leveraged for key transformational projects such as the clean energy transition, sustainable transport networks, or the rapid decarbonisation of Germany’s industrial base. The German Advisory Council on Global Change (Wissenschaftlicher Beirat der Bundesregierung Globale Umweltveränderungen, WBGU), a scientific advisory body that advises the German government on the environmental sustainability transformation, states in a 2019 report that “digital technologies such as AI play a key role in enabling a global transformation of energy systems” (WBGU, 2019[33]). AI could help to reduce emissions in all six sectors of the German climate protection law: energy, transport, industry, buildings, agriculture, and waste (Rolnick et al., 2019[34]).
Many initiatives in the German AI and environmental sustainability ecosystem cover these critical sectors (Figure 9.1). According to several experts interviewed for this review, Germany could benefit from defining strategic focus areas and sectors and from bundling initiatives to create synergies, avoid the duplication of work, and share knowledge and best practices. While AI has various application areas in almost all sectors of the economy, four crucial sectors in Germany show particularly high potential: energy, transport, industry, and agriculture.
Figure 9.1. Several initiatives in the German AI ecosystem leverage AI for rapid decarbonisation across sectors
Smart energy systems and networks
According to the International Energy Agency, digital technologies like AI will play a fundamental role in enabling the transition to a resilient and clean energy grid by improving efficiency, reducing costs, and accelerating clean technologies and diffusion across supply chains (IEA, 2023[35]). Germany’s energy sector was responsible for 34% of greenhouse gas (GHG) emissions in 2022 and has ambitious targets to increase the share of renewable sources to 80% of electricity consumption by 2030, with “energy savings and energy efficiency as top priorities” (OECD, 2023[36]). Initiatives for energy efficiency, smart energy solutions and renewable energy can be found across federal ministries, states, academia, and in industry and start-ups. This includes AI initiatives from the BMWK for the clean energy transformation or several lighthouse projects from the BMUV that address renewable power generation from wind and waterpower. Already in 2019, the German Energy Agency (Deutsche Energie-Agentur, DENA) recognised that leveraging AI for the energy sector promises to accelerate Germany’s Energiewende (energy transition) and the decarbonisation of the power grid, which enables the ecological transformation of other sectors such as transport and industry (dena, 2019[37]). While Germany’s energy sector recently met its climate targets, “decoupled energy demand and carbon dioxide”, and is “one of the G20 and EU27 countries with the highest levels of energy efficiency”, challenges in the energy sector persist, especially in the wake of the global energy crisis (OECD, 2023[36]). Industry associations like the German Data Associations report on impaired competitiveness by Germany’s high electricity costs and call for an acceleration of the availability of electricity from renewable energy (bitkom/eco/German Datacentre Association, 2022[38]). AI can be a key acceleration enabler, especially in energy bottlenecks like the digitalisation and expansion of electricity grids and energy infrastructure (OECD, 2023[36]).
Interconnected transport networks
One of the most promising application areas for AI is the transport sector. AI can help to reduce the overall demand for travel and transportation, for example through videoconferencing and teleworking. It can also increase fuel efficiency and infrastructure longevity through AI-enabled digital twins and reduce overall passenger transport activity through on-demand ride services or vehicle sharing (EEA, 2023[39]). Initiatives such as the BMDV’s AIAMO project for sustainable mobility, start-ups such as TWAICE work on battery efficiency improvements, and academic excellence clusters focused on digital mobility, such as in Bavaria, are examples of German activities and expertise in this field. Another example is project RASMUS of German start-up north.io, which combines AI with oceanographic models to calculate shipping routes that leverage small dynamic ocean currents. The optimised routes could result in GHG emission savings of up to 10% for shipping operations (Christian-Albrechts-Universität zu Kiel, 2023[40]). Researchers have often observed that a narrow focus on increasing the efficiency of the transport sector might not be enough to meet the sector’s climate goals, as an increase in overall transportation demand offset efficiency gains (Creutzig et al., 2015[41]). The International Transport Forum encourages policy makers to enable “modal shift” and demand management for urban environments and short-distance travel, nudging passengers towards low-carbon transport options, another significant application area for AI systems (ITF, 2023[42]). Harnessing these opportunities for decarbonisation could be especially relevant for Germany as the transport sector accounted for 18% of Germany’s GHG emissions in 2022 and has been the slowest sector to cut emissions (OECD, 2023[36]).
Industry 4.0
AI can support the rapid decarbonisation of Germany’s industrial sector, which accounted for 23% of Germany’s total GHG emissions in 2022 (OECD, 2023[36]). This opportunity is reflected in several initiatives from federal ministries and the many applications and industrial solutions by some of Germany’s largest companies. Companies like Siemens or Bosch deploy AI for predictive maintenance, digital twins, and the overall digital and ecological transformation of German industry – a cornerstone of the Industry 4.0 vision of interconnected machines and processes through digital technologies. AI can also be used for advanced material discovery and scientific innovation, which will be crucial for industrial companies to decarbonise their operations and reach environmental goals (IEA, 2023[35]). Platform Industry 4.0, a network platform for the digital transformation of the manufacturing sector led by the BMWK and BMBF, recognised environmental sustainability as a key aspect of the Industry 4.0 vision and introduced a task force on sustainability (BMWK, 2022[43]). The task force regards digital technologies such as AI as a key enabler of the sustainable transformation of German industry, which aligns with positions from industry associations such as Bitkom. Bitkom calculated that the GHG emission reduction potential of accelerated digitalisation for German industry could be up to 34% of the required emission cuts until 2030 (bitkom, 2023[44]). AI is also regarded as a key enabler for future circular economic business models of industry 4.0, characterised by the “connectivity and flow of information and data across value chains and processes” (One Planet Network, 2023[45]).
Smart agriculture
The OECD-FAO Agricultural Outlook 2021-2030 highlights that the necessary improvements in productivity to feed the global population sustainably will not happen “without an important acceleration in digitalisation, technology, better data, and human capital” (OECD/FAO, 2021[46]). This presents an opportunity for Germany to leverage AI to improve agricultural yields and environmental sustainability in Germany’s agricultural sector and to export AI-based technological solutions to support the sustainable development of farmers across the globe. As climate change accelerates rapidly, a strong increase in demand for advanced technology and climate-resilient crops is expected. Germany’s agricultural sector could also benefit from more efficient and sustainable agricultural practices enabled by AI as the sector was responsible for around 9% of GHG emissions in 2022 and has not decreased significantly in the past decade (OECD, 2023[36]). Initiatives such as the promotion of AI for smart agriculture and rural areas by the BMBF or the FAIR Forward project from the BMZ that provides partner countries with access to climate-smart agricultural services could be cornerstones of an AI strategy for the agricultural sector. This builds on the BMEL programme on AI for sustainable agriculture (BMEL, 2021[11]). Industry partners like BASF that work on AI-based digital farming solutions and research institutes such as Fraunhofer and the DFKI could also co-operate with federal and state initiatives on the smart AI-enabled agriculture of the future.
Strengthening Germany’s leadership role in AI and environmental sustainability
The German government actively endorses AI applications for the common good, illustrated by the Civic Coding Initiative, which fosters social, sustainable, and participative AI development. Key German initiatives include the Civic Innovation Platform, which supports human-centric AI ideas and projects; the AI Ideas Workshop for Environmental Protection, which fosters eco-friendly AI solutions; and the Civic Data Lab, which enhances data-driven efforts for social good. These programmes exemplify Germany's commitment to leveraging AI for societal and environmental advancement.
The German AI and environmental sustainability ecosystem is ahead of many countries, being home to leading researchers, practitioners, and pioneers in the field
The first important step in strengthening Germany’s AI and environmental sustainability ecosystem is to significantly increase co-operation within the federal government, between federal and state governments, and on in knowledge sharing between government, universities, civil society, and industry. The intersection of two highly complex topics like AI and environmental sustainability requires interdisciplinary knowledge and skills that can be leveraged through widespread co-operation. The analysis of current initiatives in the ecosystem suggests that the ample opportunities in sectors such as energy, transport, and agriculture could be better explored by strategic clusters where specific domain knowledge is consistently shared. Examples of such co-operation are initiatives like the Community Sustainable Digitalisation of the BMUV that connects academia, start-ups, industry, municipalities, and the federal government (BMUV, 2023[47]). Germany could also mirror initiatives like the UK’s Artificial Intelligence for Decarbonisation’s Virtual Centre of Excellence (ADViCE) which serves as a central hub for AI and decarbonisation projects and aims to “foster cross-sector collaboration” and “disseminate information to relevant stakeholders” (Alan Turing Institute, 2023[48]). Another example is the whole-of-government initiative Clean Growth Hub in Canada, an inter-departmental “coordination centre for federal clean tech initiatives and a one-stop shop for information about funding and services” (Government of Canada, 2023[49]). Germany could replicate such an approach for its AI activities that target decarbonisation and other environmental sustainability initiatives.
A second crucial area is awareness of opportunities and challenges for AI and environmental sustainability. In the interviews conducted for the review, experts highlighted a lack of knowledge and cross-disciplinary environmental and technical knowledge in both the public and private sectors, which calls for implementing AI literacy and upskilling programmes for German policy makers, higher education, research, and industry. Examples such as the sustAInability project at TUM Munich, which builds capacity amongst interdisciplinary students from both technical and non-technical backgrounds, could be expanded to more universities and research institutes and replicated in the context of government training programmes. Policy makers involved in climate-relevant legislation and AI policy should also be offered training programmes, as should leaders in civil society and high-emission sectors. Germany could follow examples such as the Stanford Institute for Human-Centred AI (HAI) Congressional Boot Camp on AI and introduce a similar programme dedicated to environmental sustainability (HAI, 2023[50]). Another important field of action is support for start-ups and German SMEs to leverage AI for environmental sustainability applications and business models. The BMUV’s Green AI Hub Mittelstand shows that this is already on the federal government's agenda and could be further expanded to the start-up ecosystem nationwide.
The 2020 strategy update mentions diverse application areas for AI, such as “renewable energies and energy systems, energy efficiency, resource conservation and recycling, water protection and water management, emission control and health, nature conservation and mobility” (German Federal Government, 2020[2]). The expert interviews conducted for the review and the analysis of the ecosystem suggest that, in practice, many initiatives in Germany currently focus on energy and resource efficiency, which already bring significant energy and raw material savings in various sectors. However, researchers have consistently argued that a general focus on efficiency can result in rebound effects, and efficiency savings do not always translate into overall emission reductions (Creutzig et al., 2015[41]). The focus of the AI and environmental sustainability ecosystem could therefore be widened to include other environmental dimensions such as circular economic models, biodiversity preservation, or water consumption, which the 2020 strategy update already identified as key focus areas. Comparable initiatives exist in countries like France, where the National Agency for Research (Agence nationale de la recherche, ANR) has launched a research challenge to converge the French National Artificial Intelligence Research Strategy and the France’s National Biodiversity Plan (ANR, 2021[51]). Such initiatives could further strengthen Germany’s leadership in the area as they represent a more holistic view on the sustainable use of AI that could serve as a role-model for other countries.
Measuring and mitigating the environmental impacts of AI compute infrastructure
Germany is recognised for its significant AI and environmental sustainability initiatives, involving collaboration across federal ministries, states, academia, industry, and civil society. These efforts are underpinned by substantial funding and are focused on creating AI applications that support environmental and climate goals. Key work areas include leveraging AI for resource efficiency, promoting sustainable mobility, biodiversity preservation, and fostering sustainable agriculture. The country's approach also emphasises inter-ministerial co-operation and the active engagement of civil society to ensure that AI developments align with ecological and social sustainability principles.
The environmental impacts of AI’s digital compute infrastructure such as energy consumption and water use are not systematically measured
As the computational needs for advanced AI systems grow, so are sustainability concerns regarding the environmental impacts of AI compute infrastructure (OECD, 2023[52]). Across its lifecycle, AI compute has direct impacts from production, transport, operations, and end-of life stages such as energy consumption, water use, and resource consumption. Digital technologies like AI also have further indirect impacts such as enabling effects (through applications in specific sectors) and systemic effects (by changing social or cultural behaviour). While AI only represents a fraction of overall impacts from digital technologies, the proliferation of AI applications and the exponential dynamic of AI compute requirements call for implementing measurement standards and expanding data collection on the environmental impacts of AI compute infrastructure and applications (OECD, 2022[1]).
With the Energy Efficiency Act (Energieeffizienzgesetz, EnEfG), the German Federal Government has proposed one of the first laws of its kind regarding reporting requirements for data centre operators. It is the first European country to implement the European Union (EU) Energy Efficiency Directive. Operators will be obliged to report on various environmental indicators, procure renewable energy for their operations, meet energy efficiency targets, make economical use of cooling system power, and utilise waste heat (German Federal Government, 2020[2]). The utilisation of waste heat from data centres is regarded as an often-untapped opportunity by the International Energy Agency (IEA), which recommends governments and policy makers work together with operators and local communities to provide district heating and supply industrial heat users wherever possible (IEA, 2023[53]). Many German data centre operators already capture waste heat, although bureaucratic hurdles often limit the effective distribution.
The German government recognises the need for environmentally specific standards and measurements in its 2020 strategy update. It aims to make AI environmentally sound by “systematically expanding its funding and research linking digitalisation and ecological sustainability goals” and to “advance energy- and resource-saving information and communication technology (green ICT)” (German Federal Government, 2020[2]). It proposes to develop an advanced “concept for environmental impact assessment of AI and step up its funding for research on the environmental impacts of AI, in particular commissioning the collection of empirical data and a systematic analysis of the carbon dioxide (CO2)-saving potential of AI, duly taking into account possible negative effects (such as rebound effects)” (German Federal Government, 2020[2]). Recognising the high complexity of such a concept, other countries have not implemented an environmental impact assessment of AI that includes rebound effects and would represent a trailblazing contribution to global efforts on measuring the environmental impact of AI.
Regarding best practices and knowledge dissemination, the strategic goal of advancing energy- and resource-saving ICT is already being implemented across ministries and the private sector. For instance, the BMBF and the Research Fab Microelectronics Germany (Forschungsfabrik Mikroelektronik Deutschland, FMD) sponsor the Green ICT initiative, a competence centre for sustainable information and communication technology. Green ICT provides resources and expertise to project partners from industry and science to develop sustainable microelectronics and energy-efficient technology infrastructure. It also provides educational offerings for students, industry specialists, start-ups and SMEs (Green ICT, 2023[54]).
Public compute providers in Germany have pioneered green ICT methods. For example, the Leibniz Supercomputing Centre (LRZ) of the Bavarian Academy of Science and Humanities (Bayerische Akademie der Wissenschaften, BAdW) has worked on and implemented energy-efficient computing infrastructure for over a decade, for instance through implementing a warm water-cooling loop that results in significant energy savings. The LRZ researches green information technology (IT) methods under a methodology examining the environmental impacts of building infrastructure, hardware, management software, and sustainability applications (LRZ, 2023[55]). Amongst the world’s 50 most energy-efficient supercomputers, Germany has 11 supercomputers and is second only to the United States (US) (14), ahead of France (6), Japan (5) and Australia (2) (TOP500, 2023[56]). The BMZ and the GIZ have developed a practitioners’ guide to Green Data Centers: Towards a Sustainable Digital Transformation together with the ITU and the World Bank. The guide encourages public and private investment in green data centre infrastructure through public procurement strategies and wider policies and regulations (ITU/World Bank, 2023[10]).
The energy consumption of Germany’s servers and data centres has increased since 2015 (Figure 9.2). Estimated at 17 billion kilowatt-hour (kWh) in 2021, data centres consumed 6.5% more energy than in 2020 and 14% more than in 2019 (14 billion kWh). This represented around 3.3% of the German national electricity supply in 2021, compared with 2.7% in the Netherlands and 2.5% in the UK (Statistics Netherland, 2021[57]); (nationalgridESO, 2022[58]). Extrapolating current trends forward could mean a consumption of around 28 billion kWh by 2030 (Borderstep Institute, 2022[59]). The Office for Technology at the German Bundestag (Büro für Technikfolgen-Abschätzung beim Deutschen Bundestag, TAB) arrives at similar numbers, estimating German data centre energy consumption at 14.9 billion kWh in 2019 and at a projected 30.6 billion kWh by 2030 (including telecommunication networks) if current trends continue. However, the authors of the TAB note that “the state of knowledge on energy requirements of ICT infrastructure is incomplete and often contradictory”, stating a “considerable need for more research” and for “regular data collection and reporting including real data from companies” (Grünwald and Caviezel, 2022[60]).
Figure 9.2. Data centres’ share of total German electricity consumption has steadily increased in recent years
German data centre electricity consumption
Note: Estimating data centre electricity consumption dedicated to AI is technically and methodologically challenging.
Source: Borderstep Institute (2022[59]), “Cloud computing drives the growth of the data centre industry and its energy consumption”, https://www.borderstep.de/wp-content/uploads/2022/08/Borderstep_Rechenzentren_2021_eng.pdf.
Recommendations
Increase inter-government and interdisciplinary co-operation on AI and environmental sustainability to promote transfer and synergies between initiatives
The German AI and environmental sustainability ecosystem benefits from a range of initiatives, funding, and strong public and private institutions that put AI and environmental sustainability on their agenda. At the same time, initiatives are often isolated and do not share knowledge and best practices or create synergies through co-operation. This is a critical issue because working at the intersection of AI and environmental sustainability requires technical and environmental expertise. Germany’s 2020 AI strategy update aims to “bolster the links between SMEs, start-ups and public interest actors and research so as to promote the transfer and application of research findings across the breadth of the economy and society” (German Federal Government, 2020[2]). This goal should be expanded to create an authentic multistakeholder and multidisciplinary approach that benefits from the resources and knowledge of Germany’s AI experts, environmental groups, and academic thought leaders.
Define strategic focus areas to maximise AI’s enabling effects for environmental sustainability and rapid decarbonisation
Germany has strong potential to cluster AI and environmental sustainability initiatives and knowledge in strategic focus areas. While initiatives exist in sectors like energy, transport, industry, and agriculture, AI can be used in nearly every sector of the economy, such as buildings and cities, green financing, or green consumption. Germany could prioritise areas to align research efforts, streamline funding, and define objectives, for example, in inter-ministerial working groups or whole-of-government initiatives.
Extend the German leadership position in AI and environmental sustainability through knowledge-sharing, education programmes, promoting start-ups and SMEs; widen the focus to include the circular economy, biodiversity, and other planetary boundaries
Germany is positioned to become a leader in the field of AI and environmental sustainability based on its strong strategic and political mandate, diverse and appropriately funded initiatives, leading academics and researchers in the field, and innovative industrial companies and start-ups that export sustainable AI solutions. This could be strengthened by encouraging and promoting knowledge-sharing and transfer, educating policy makers and students, supporting German start-ups and SMEs, and widening the focus of what constitutes sustainability beyond energy and resource efficiency to include circular economic models, biodiversity preservation, land-system change, freshwater depletion, and other planetary boundaries, to make AI work for the good of the planet.
Expand measurement efforts to assess and mitigate the energy, water, and resource impacts of AI compute infrastructure
Germany recognises the need for systematic and standardised measurement of the environmental impacts of AI and ICT infrastructure, including direct impacts, such as energy and water consumption, and indirect enabling and systemic effects from application across sectors. Germany took important first steps in collecting data on environmental impacts, such as the reporting requirements for data centre operators in the upcoming EnEfG. Such efforts should be expanded, for instance, by implementing the environmental impact assessment of AI with the Federal Statistical Office and in partnership with research institutes that track indicators like data-centre energy consumption. The resulting indicators could be included in public databases such as the high-performance computing map of the Gauss Alliance to increase transparency (GCS, 2023[61]). Germany should widen the focus from operational energy consumption and GHG emissions to wider environmental impacts such as biodiversity degradation, the lifecycle impacts of producing compute equipment, impacts from water consumption and rare-earth mining (OECD, 2022[1]). An environmental impact assessment of AI as envisioned in the national AI strategy update would benefit from intra-ministerial co-operation. It would pioneer a comprehensive ecological review of national AI compute resources and could set international standards to shape the development of metrics such as those proposed in the European Union Regulation on Artificial Intelligence (the “EU AI Act”) (European Union, 2024[62]).
References
[48] Alan Turing Institute (2023), ADViCE Introduction, https://www.turing.ac.uk/research/research-projects/advice#:~:text=Artificial%20Intelligence%20for%20Decarbonisation%27s%20Virtual,the%20transition%20to%20Net%20Zero (accessed on 11 December 2023).
[51] ANR (2021), Challenge IA-Biodiv: Research in Artificial Intelligence in the Field of Biodiversity, Agence nationale de la recherche, https://anr.fr/en/call-for-proposals-details/call/challenge-ia-biodiv-research-in-artificial-intelligence-in-the-field-of-biodiversity/ (accessed on 11 December 2023).
[12] Baiosphere (2023), Über die baiosphere, https://baiosphere.org/baiosphere/ (accessed on 11 December 2023).
[24] BASF (2023), Innovative and Disruptive Technologies for Sustainable Future Success, https://www.basf.com/global/en/who-we-are/digitalization/artificial-intelligence.html (accessed on 11 December 2023).
[44] bitkom (2023), Klimaeffekte der Digitalisierung, https://www.bitkom.org/sites/main/files/2021-10/20211010_bitkom_studie_klimaeffekte_der_digitalisierung.pdf.
[38] bitkom/eco/German Datacentre Association (2022), Perspectives for a Sustainable Data Centre Industry by 2030, https://www.germandatacenters.com/fileadmin/documents/publications/Perspectives-for-a-sustainable-data-center-industry-by-2030.pdf.
[22] Bits & Bäume (2021), Forderungen, https://bits-und-baeume.org/konferenz-2022/forderungen/ (accessed on 11 December 2023).
[9] BMBF (2023), “Bekanntmachung Methoden der Künstlichen Intelligenz als Instrument der Biodiversitätsforschung”, Bundesministerium für Bildung und Forschung, https://www.bmbf.de/bmbf/shareddocs/bekanntmachungen/de/2023/01/2023-01-04-Bekanntmachung-Biodiversit%C3%A4tsforschung.html (accessed on 11 December 2023).
[8] BMBF (2021), “Bekanntmachung”, Bundesministerium für Bildung und Forschung, https://www.bmbf.de/bmbf/shareddocs/bekanntmachungen/de/2021/09/2021-09-17-Bekanntmachung-KI.html (accessed on 11 December 2023).
[7] BMDV (2023), “Artificial Intelligence and Mobility - AIAMO”, Bundesministerium für Digitales und Verkehr, https://bmdv.bund.de/SharedDocs/DE/Artikel/DG/KI-Projekte/aiamo.html (accessed on 11 December 2023).
[11] BMEL (2021), Künstliche Intelligenz (KI) für eine nachhaltigere Landwirtschaft, Bundesministerium für Ernährung und Landwirtschaft, https://www.bmel.de/SharedDocs/Downloads/DE/Broschueren/k-i-fuer-nachhaltige-landwirtschaft.pdf?__blob=publicationFile&v=7 (accessed on 11 December 2023).
[47] BMUV (2023), Community “Nachhaltige Digitalisierung”, Bundesministerium für Umwelt, Naturschutz, nukleare Sicherheit und Verbraucherschutz, https://www.bmuv.de/themen/digitalisierung/community-nachhaltige-digitalisierung (accessed on 11 December 2023).
[4] BMUV (2023), Künstliche Intelligenz für Umwelt und Klima, Bundesministerium für Umwelt, Naturschutz, nukleare Sicherheit und Verbraucherschutz, https://www.bmuv.de/themen/digitalisierung/kuenstliche-intelligenz-fuer-umwelt-und-klima (accessed on 11 December 2023).
[43] BMWK (2022), Industrie 4.0 gestalten. Resilient, nachhaltig, wettbewerbsstark, Bundesministerium für Wirtschaft und Klimaschutz, https://www.bmwk.de/Redaktion/DE/Publikationen/Industrie/industrie-40-gestalten-resilient-nachhaltig-wettbewerbsstark.pdf?__blob=publicationFile&v=1 (accessed on 11 December 2023).
[59] Borderstep Institute (2022), “Cloud computing drives the growth of the data center industry and its energy consumption”, Borderstep Institute for Innovation and Sustainability, https://www.borderstep.de/wp-content/uploads/2022/08/Borderstep_Rechenzentren_2021_eng.pdf.
[28] Celonis (2023), Sustainability Transformation, https://www.celonis.com/solutions/sustainability-transformation/ (accessed on 11 December 2023).
[40] Christian-Albrechts-Universität zu Kiel (2023), Rasmus: Cost-efficient Ship Routing Utilizing Ocean Currents, https://www.ai.informatik.uni-kiel.de/de/projects/current-projects-2/copy_of_rasmus-ship-routing (accessed on 11 December 2023).
[6] Civic Coding (2023), Die Initiative, https://www.civic-coding.de/ueber-civic-coding/die-initiative (accessed on 2 November 2023).
[41] Creutzig, F. et al. (2015), “Transport: A roadblock to climate change mitigation?”, Science, Vol. 350/6263, pp. 911-912, https://www.science.org/doi/10.1126/science.aac8033.
[37] dena (2019), Künstliche Intelligenz für die integrierte Energiewende, Deutsche Energie-Agentur GmbH, https://www.dena.de/fileadmin/dena/Publikationen/PDFs/2019/dena-ANALYSE_Kuenstliche_Intelligenz_fuer_die_integrierte_Energiewende.pdf.
[17] DFKI (2023), Artificial Intelligence for Environment and Sustainability, Deutsches Forschungszentrum für Künstliche Intelligenz GmbH, https://www.dfki.de/en/web/applications-industry/competence-centers/ai-for-environment-and-sustainability (accessed on 11 December 2023).
[39] EEA (2023), Transport and Environment Report 2022, European Environment Agency, https://www.eea.europa.eu/publications/transport-and-environment-report-2022/transport-and-environment-report/view.
[20] ELIAS (2023), European Lighthouse of AI for Sustainability, https://elias-ai.eu (accessed on 11 December 2023).
[26] Enpal (2023), Über Enpal, https://www.enpal.de/ueber-uns (accessed on 11 December 2023).
[62] European Union (2024), Regulation (EU) 2024/ ...... of the European Parliament and of the Council laying down harmonised rules on artificial intelligence (Artificial Intelligence Act), https://data.consilium.europa.eu/doc/document/PE-24-2024-INIT/en/pdf.
[61] GCS (2023), Our Centres, Gauss Centre for Supercomputing, https://www.gauss-centre.eu/about-us/our-centres (accessed on 11 December 2023).
[30] German AI Association (2023), Working Groups, https://ki-verband.de/en/working-groups/ (accessed on 11 December 2023).
[2] German Federal Government (2020), Artificial Intelligence Strategy of the German Federal Government: 2020 Update, https://www.ki-strategie-deutschland.de/files/downloads/Fortschreibung_KI-Strategie_engl.pdf.
[49] Government of Canada (2023), About the Clean Growth Hub, https://ised-isde.canada.ca/site/clean-growth-hub/en/about-clean-growth-hub (accessed on 11 December 2023).
[54] Green ICT (2023), About Green ICT, https://greenict.de/en/ (accessed on 11 December 2023).
[29] Greentech Alliance (2023), What We Do, https://www.greentech.earth/what-we-do (accessed on 11 December 2023).
[60] Grünwald, R. and C. Caviezel (2022), “Energieverbrauch der IKT-Infrastruktur. TAB-Fokus”, Institut für Technikfolgenabschätzung und Systemanalyse (ITAS), Karlsruher Institut für Technologie (KIT), https://publikationen.bibliothek.kit.edu/1000151166.
[50] HAI (2023), Congressional Boot Camp on AI, Stanford University, https://hai.stanford.edu/congressional-boot-camp-ai (accessed on 11 December 2023).
[13] hessian.AI (2023), “Supercomputer for cutting-edge AI research in Hesse”, https://hessian.ai/supercomputer-for-cutting-edge-ai-research-in-hesse/ (accessed on 11 December 2023).
[53] IEA (2023), Data Centres and Data Transmission Networks, International Energy Agency, https://www.iea.org/energy-system/buildings/data-centres-and-data-transmission-networks (accessed on 11 December 2023).
[35] IEA (2023), Energy Technology Perspectives 2023, International Energy Agency, https://www.iea.org/reports/energy-technology-perspectives-2023.
[42] ITF (2023), ITF Transport Outlook 2023, OECD Publishing, Paris, https://doi.org/10.1787/b6cc9ad5-en.
[10] ITU/World Bank (2023), Green Data Centers: Towards a Sustainable Digital Transformation - A Practioner’s Guide, International Telecommunication Union and the World Bank, https://www.bmz-digital.global/wp-content/uploads/2023/08/Guide-to-Green-Data-Centers-Final.pdf.
[14] KI.NRW (2023), Selected Examples of the North Rhine-Westphalian AI ecosystem, https://www.ki.nrw/en/ai-highlights/ (accessed on 11 December 2023).
[55] LRZ (2023), Green IT at Leibniz Supercomputing Centre, Leibniz-Rechenzentrum, https://www.lrz.de/wir/green-it_en/ (accessed on 11 December 2023).
[58] nationalgridESO (2022), “Data centres”, https://www.nationalgrideso.com/document/246446/download (accessed on 11 December 2023).
[52] OECD (2023), “A blueprint for building national compute capacity for artificial intelligence”, OECD Digital Economy Papers, No. 350, OECD Publishing, Paris, https://doi.org/10.1787/876367e3-en.
[36] OECD (2023), OECD Environmental Performance Reviews: Germany 2023, OECD Environmental Performance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/f26da7da-en.
[1] OECD (2022), “Measuring the environmental impacts of artificial intelligence compute and applications: The AI footprint”, OECD Digital Economy Papers, No. 341, OECD Publishing, Paris, https://doi.org/10.1787/7babf571-en.
[3] OECD.AI (2023), AI and the Environment, OECD, Paris, https://oecd.ai/en/dashboards/policy-areas/PA9.
[46] OECD/FAO (2021), OECD-FAO Agricultural Outlook 2021-2030, OECD Publishing, Paris, https://doi.org/10.1787/19428846-en.
[45] One Planet Network (2023), Rethinking, Extending, Re-using. Harnessing Digital Technologies for the Circular Economy, https://www.oneplanetnetwork.org/sites/default/files/2023-11/RETHINKING%2C%20EXTENDING%2C%20REUSING-%20HARNESSING%20DIGITAL%20TECHNOLOGIES%20FOR%20THE%20CIRCULAR%20ECONOMY%20-%205%20%281%29.pdf.
[32] OroraTech (2021), “Maintain control of wildfires: OroraTech launches global thermal intelligence services”, https://ororatech.com/ororatech-launches-global-thermal-intelligence-service/ (accessed on 11 December 2023).
[25] Porsche (2021), “Porsche, Audi and Volkswagen use artificial intelligence to minimise sustainability risks”, https://newsroom.porsche.com/en/2021/sustainability/porsche-audi-volkswagen-pilot-project-artificial-intelligence-minimisation-sustainability-risks-supply-chain-23801.html (accessed on 11 December 2023).
[21] Rohde, F. et al. (2021), Nachhaltigkeitskriterien für künstliche Intelligenz, https://www.ioew.de/fileadmin/user_upload/BILDER_und_Downloaddateien/Publikationen/2021/IOEW_SR_220_Nachhaltigkeitskriterien_fuer_Kuenstliche_Intelligenz.pdf.
[34] Rolnick, D. et al. (2019), “Tackling climate change with machine learning”, https://arxiv.org/abs/1906.05433.
[23] Siemens (2023), The Climate Crisis – Solutions in the Ecosystem, https://www.siemens.com/global/en/company/stories/research-technologies/energytransition/sustainability-solutions-in-the-ecosystem.html (accessed on 11 December 2023).
[57] Statistics Netherland (2021), “Electricity supplied to data centres, 2017-2019”, https://www.cbs.nl/en-gb/custom/2020/51/electricity-supplied-to-data-centres-2017-2019 (accessed on 11 December 2023).
[15] STMD (2023), “Bundesländer setzen ständige Digitalministerkonferenz ein / Digitalminister Mehring: „Meilenstein für die Digitalisierung in Deutschland!“”, Bayerisches Staatsministerium für Digitales, https://www.stmd.bayern.de/bundeslaender-setzen-staendige-digitalministerkonferenz-ein-digitalminister-mehring-meilenstein-fuer-die-digitalisierung-in-deutschland/ (accessed on 11 December 2023).
[18] sustAInability (2023), About sustAInability, https://sustainability-ai.de (accessed on 11 December 2023).
[16] Sustainable AI Lab (2023), About Us, https://sustainable-ai.eu/about-us/ (accessed on 11 December 2023).
[56] TOP500 (2023), Green500 - November 2023, https://www.top500.org/lists/green500/2023/11/ (accessed on 11 December 2023).
[19] TU Berlin (2020), “KI-Leuchttürme für die Umwelt”, https://www.tu.berlin/ueber-die-tu-berlin/profil/pressemitteilungen-nachrichten/ki-leuchttuerme-fuer-die-umwelt (accessed on 11 December 2023).
[27] TWAICE (2023), TWAICE Battery Analytics Platform, https://www.twaice.com/products/battery-analytics-platform (accessed on 11 December 2023).
[5] UBA (2022), UBA Application Lab for AI and Big Data, Umweltbundesamt, https://www.umweltbundesamt.de/en/topics/digitalisation/uba-application-lab-for-ai-big-data (accessed on 11 December 2023).
[31] UnternehmerTUM (2023), TUM Venture Labs Sustainability and Circular Economy, https://www.unternehmertum.de/en/about/tum-venture-labs (accessed on 11 December 2023).
[33] WBGU (2019), Flagship Report - Towards Our Common Digital Future, German Advisory Council on Global Change, https://www.wbgu.de/fileadmin/user_upload/wbgu/publikationen/hauptgutachten/hg2019/pdf/wbgu_hg2019_en.pdf.