Artificial intelligence and sustainable consumption in Europe

Linking artificial intelligence to sustainable consumption in Europe

This briefing explores the potential impact of artificial intelligence (AI) on sustainable consumption in Europe. As the fastest-spreading technology in human history, AI is at the centre of the European Commission’s strategy to foster competitiveness and strengthen technological sovereignty, as established in the AI Continent Action Plan. AI has become a transformative force that is restructuring industries and individual behaviours. As such, if well managed, it has the potential to create new avenues, if well managed, for a transformation towards a more sustainable future.

AI has the potential to improve systems the environmental sustainability of systems by increasing efficiency, effectiveness and diffusion rate and speed of sustainable innovations. Yet the environmental trade-offs of this 'blitz scaling up' of AI remain difficult to assess, let alone mitigate.

Transformations in lifestyles and consumption behaviour have the potential to lower greenhouse gas emissions by 40–70% by 2050 (IPCC, 2023). However, the consumption footprint of EU citizens is projected to rise again by 2030, following recent trends that show a 6.3% increase between 2020 and 2022 (EEA, 2024). At the same time, online consumption is becoming increasingly dominant, with 75% of Europeans purchasing or ordering goods or services for personal use online in 2024 (Eurostat, 2025a).

As AI emerges as a key driver of online behaviour, understanding its potential to foster sustainable consumption becomes increasingly important to inform the possible adaptation of current and future regulatory frameworks. A central component here is the availability of supportive technologies and infrastructure promoting shifts in behaviour toward more sustainable choices (OECD and UNDP, 2025).

This also affects private and public procurement since AI has the potential for fostering higher sustainability as corporations and governments integrate AI to direct their suppliers’ selection, contracting and monitoring (OECD, 2022).

While AI offers significant potential to support more sustainable purchasing practices, inadequate governance or uneven deployment risks reinforcing resource-intensive business models, market concentration, and inequalities. Hence, a precautionary approach is needed to identify priority areas for policy action and ensure that decisions taken today enable AI to contribute to a sustainable, competitive, and secure Europe.

Building upon previous work on sustainable and circular consumption (EEA, 2023) and the role of digital technologies in Europe’s circular economy (ETC CE, 2025), this briefing discusses key links between AI and consumption relevant for environmental policy making.

Why is AI important for sustainable production and consumption patterns and competitiveness?

The EU Single Market shows the magnitude of the potential contribution from directing EU consumption patterns towards higher sustainability. With a GDP of €17.1 trillion and representing 15% of global trade, the Single Market makes Europe a major economic player at global scale (EC, 2025h). However, trade between EU member states has fallen for the first time in almost a decade - except for the Covid years - from 23.5% of EU GDP in 2023 to 22% in 2024 (Hancock and Moens, 2026). A main reason lies on the hidden costs involved in trading goods within the EU which are estimated at 65% tariff for goods and 100% for services (ECB, 2025).

The integration of AI across the EU’s domestic market is expected to offer new avenues for harnessing the efforts of simplification and optimisation of trade and financial operations as consistently targeted by a range of policy packages across the different economic sectors. For example, the Data Union Strategy (EC, 2025e) is opening up new sources of high-quality, large-scale data and is enabling businesses and public administrations to share data seamlessly and at scale; it relies on AI for data quality, access and sovereignty.

• The European Open Science Cloud (EOSC) is a federation of data repositories and services that provides scientific services to support the uptake of AI.
• The strategic roadmap for digitalisation and AI in the energy sector is seeking to speed up the deployment of digital technologies, including European AI solutions, in areas that are key to the decarbonisation transition (e.g. optimising electricity grids, improving energy efficiency in buildings and industry, and enhancing demand-side flexibility) (EC, 2025d).
• The Destination Earth initiative is using AI to create a digital replica of key aspects of the Earth’s systems, supporting climate change adaptation; it harnesses the most powerful supercomputer systems in Europe (EC, 2025c).

One of the major strengths of AI lies in its ability to identify patterns in data and to analyse past information to support accurate predictions. As such, pervasive implementation of AI across the EU domestic market could help to align competitiveness and sustainability targets. This makes AI highly valuable not only for environmental monitoring, but also for supporting governments, businesses and individuals to make more environmentally sustainable decisions. If deployed appropriately, AI could enhance the EU’s capacity to address the triple planetary crisis: climate change; the loss of nature and biodiversity; and pollution and waste (JRC, 2025b; OECD, 2022; UNEP, 2024).

However, such enhanced capabilities, largely driven by stronger reasoning and greater autonomy, increase the challenges related to managing the risks associated with AI. Broader geographical coverage, longer operational timeframes and the use of more advanced tools complicate oversight. This is especially the case given that AI systems function with reduced human supervision in high-stakes contexts and evidence about their real-world risks remains incomplete (Bengio et al., 2025).

2025 reports show AI is already widely used in the EU as by 30% of workers (JRC, 2025a), and an average of 20% of EU companies or 50% for large ones (Eurostat, 2025b). Nevertheless, using AI represents both a major opportunity to foster more sustainable consumption and production systems but also a complex challenge for policymakers. Balancing the advantages of a competitive AI system with environmental sustainability requires coordination across funding systems, governance structures, infrastructure development, and skill building strategies (Bianchini et al., 2024; EP, 2021).

AI impacts on sustainable consumption and procurement

AI is reshaping patterns of consumption at the household, corporate and governmental levels — from AI companions that influence everyday choices and data-driven business models that optimise production and supply chains, to smart public procurement policies aiming to increase the security of supply for critical raw materials.

Consumer behaviour

Sustainable consumption is a cornerstone of Europe’s circular economy strategy thereby contributing towards tackling natural resources scarcity, biodiversity loss, achieving climate neutrality by 2050, and increasing resilience and competitiveness (EC, 2024a). Consumer behavioural changes towards more sustainable practices have the potential to contribute to could reduce EU green investment needs by about 8% (EC, 2024c). This involves ensuring that products are designed for optimal lifespans, high durability, and easy repair, upgrading, disassembly, and recycling. This also includes extending product use through second-hand markets and encouraging higher rates of utilisation via business models based on renting and leasing.

Equally important is providing consumers with clear, reliable, relevant and timely information so that their decisions are well informed and they can actively support the transition towards a circular economy. For example, the Digital Product Passport aims to enable consumers and businesses to make informed, sustainable choices by providing detailed information on products (e.g. material composition, carbon footprint, environmental impact, durability, repairability and end-of-life).

AI increasingly shapes consumer behaviour and hence how households engage with sustainability — from personalised recommendations to automated decision-support systems that drive choices. AI can promote, but also undermine, sustainable consumption depending on how it is designed and deployed.

Sustainable consumption is a complex topic that encompasses the use of goods and services that meet basic needs and improve quality of life while minimising ecological footprints and resource use (UNEP, 2017). As such, AI’s capacity to process data on a very large scale and to direct behaviour may have a significant impact.

For example, retail companies are already using in-app shopping agents (e.g. AI chatbots) capable of understanding context and intent, such as a person’s time constraints and meal plans, and of combining that with customer data previously gathered, including price sensitivity, flavour and brand affinity. For some companies, engagement with a shopping agent leads to doubling of conversion rates (Lin, 2026).

Figure 1. Examples of links between AI and sustainable consumer behaviour

Corporate procurement

Corporate procurement accounts for 92-96% of an organisation’s total emissions (CDP and BCG, 2024). As such, fostering sustainability in this area is a critical factor in addressing climate change.

AI tools are allowing corporate procurement to shift from a transactional to a strategic and data-driven function, improving speed, accuracy and sustainability outcomes. Across industries, integrating AI into sustainable procurement is expected to bring benefits in:

• cost savings through dynamic pricing and reduced maverick spend;
• risk mitigation via real-time supplier analytics and early-warning indicators;
• mapping of low-carbon and circular supply options (McKinsey, 2025; Menache et al., 2025; Paul et al., 2024).

In addition, AI has the potential to enable organisations to increase their supply chain resilience by:

• identifying and anticipating the impacts of external shocks due to climate change and geopolitical uncertainty;
• enhancing collective investment in climate adaptation by enabling standardised communication, risk measurement and data sharing (WEF, 2025).

Figure 2. Examples of links between AI and sustainable corporate procurement

Public procurement

The AI Continent Action Plan (EC, 2025g) and the Competitiveness Compass (EC, 2025b) set out public procurement as a core element of EU strategies for competitiveness and AI. With annual public procurement spending of nearly EUR 2 trillion — roughly 15% of EU GDP (EC, 2025g) and 10% of the EU’s GHG emissions (Carbone4, 2024) — EU governments have significant purchasing power. This means they are capable of fostering smart AI adoption and also lowering the carbon footprint of goods and services.

In the area of sustainable public procurement, the expectation is that AI will inform how 2040 EU climate targets are translated into practical regulatory procurement requirements while also enabling better monitoring. Through integrated analysis of procurement data alongside environmental impact indicators, AI systems can inform supplier selection and contracting according to pre-established standards. AI could even establish dynamic standards tailored to external parameters (Sanchez-Graells, 2024).

This means that policy instruments such as the EU Public Procurement Directive may become more effective as AI algorithms enable public procurers to include specific environmental (e.g. design for repairability) and social-economic criteria (e.g. % of parts made in EU) in calls for competition with lower uncertainty about requirements suitability and the availability of compliant suppliers. In addition, by enhancing data gathering and analysis from policy tools such as ecolabels and the Digital Product Passport, AI can provide tailored direction throughout the procurement lifecycle thereby facilitating compliance monitoring and environmental impact assessments across time (from cradle-to-cradle) and scale (from individual use to national and European patterns).

Figure 3. Examples of AI links to more sustainable public procurement in Europe

Key aspects of AI deployment for sustainability and competitiveness

AI can influence consumption patterns in ways that both exacerbate and mitigate unsustainable practices. While AI offers significant opportunities to enhance sustainable behaviour among individuals, corporations and governments, poorly governed or uneven deployment risks reinforcing resource intensity, market concentration, and social inequalities. Hence, exacerbating rather than softening the very challenges it has the potential to help solving.

At the same time, there is limited evidence about the current and historical use of AI due to its relatively recent adoption and constraints on data availability and access. As a result, a precautionary lens is required to highlight the aspects of AI that warrant careful policy attention. This approach can inform how decisions made today will determine whether AI effectively contributes to a sustainable, competitive and secure Europe.

Climate neutrality and impacts of AI-related energy consumption

The rapid expansion of AI presents a growing challenge to achieving climate neutrality in Europe. The training of AI models is driving exponential growth in computational resources. Training-related computing power increased by a factor of one million between 1960 and 2010, and by approximately one trillion times between 2010 and 2025 (Mills, 2025).

Although data centres accounted for an estimated 1.5% of global electricity consumption in 2024 — and around 3% of Europe’s — their energy demand is increasing rapidly and is often highly concentrated at the local level. It exceeds 20% of total electricity consumption in countries such as Ireland (EP, 2025a). By 2026, the electricity consumption of data centres is expected to reach 1,050 terawatt hours. Comparing data centre consumption to that of individual countries, it is equivalent to fifth place in the global energy consumption country rankings, between Japan and Russia (MIT, 2025).

Largely driven by AI, the rapid growth of data centres in Europe is expected to result in a near doubling of the sector's electricity needs by 2030 (S&P Global, 2025). Germany hosts the largest number of facilities, while Ireland has the highest density of data centres per capita. Many AI-focused data centres are clustered around major urban hubs, particularly in the so-called FLAP-D cities - Frankfurt, London, Amsterdam, Paris and Dublin - while new markets are emerging in countries such as Spain, Italy and Poland. This spatial concentration places increasing pressure on local electricity grids, land availability, and emissions profiles, raising concerns about grid capacity, sustainability, and the overall carbon footprint of Europe’s digital infrastructure.

Europe already responds for 15% of the global data centre electricity consumption, whereas the US accounts for 45% and China for 25% (IEA, 2025). The operational phase of AI, known as inference, now accounts for approximately 80 to 90% of total AI-related computing (MIT, 2025). This poses a regulatory challenge for the EU, as most inference is delivered through proprietary commercial models whose energy consumption data remains undisclosed. To address this, the EU Artificial Intelligence Act (EC, 2024b) introduced transparency obligations for general-purpose AI models, requiring providers to document and report energy usage. In addition, the forthcoming 'Cloud and AI Development Act' is expected to define the conditions and flexibility for rapid development of data centre capacity considering sustainability safeguards (EPRS, 2025). However, current provisions primarily focus on training-related compute, leaving inference-related energy consumption only partially regulated. This gap limits assess to data on AI’s environmental footprint, challenging the fit of AI governance to climate policy objectives.

Impacts of data centre cooling on water resources

AI generates water demand across multiple stages of its lifecycle, including on-site data centre cooling (Scope 1), electricity generation (Scope 2), and semiconductor manufacturing (Scope 3). While water intensity varies depending on technological design, energy mix, and climatic conditions, electricity production generally represents the largest share of AI-related water consumption, followed by cooling processes and semiconductor fabrication. This multi-scope water footprint presents a growing challenge for EU environmental governance, as it spans across sectors and countries with different policy frameworks and climatic conditions.

Current projections indicate that AI-driven data centres could increase annual water consumption for cooling and electricity generation to approximately 1,068 billion litres by 2028 under a baseline scenario - an elevenfold increase compared to 2024 estimates (Morgan & Stanley, 2025). This estimation is anchored to an expected expansion of generative AI correspondent to a power demand rise by a factor of 8.5 between 2024 and 2028.

Recent analysis of data from EU data centres from submissions under Article 12(5) of the Energy Efficiency Directive has shown that almost 20% are located in regions with at least 165 cooling degree days a year (EC, 2025a) which considerably adds to cooling requirements. Such trends raise concerns regarding consistency with EU water and climate objectives, particularly under the Water Framework Directive, which prioritises sustainable water use and ecosystem protection, and the European Green Deal, which seeks to decouple economic growth from resource depletion.

In addition, as data centre development accelerates across Europe, often in regions already experiencing water stress, the rising water footprint of AI risks intensifying competition between industrial, agricultural, and household water users. These developments underscore the importance of enhancing Europe’s water resilience, including the capacity of river basins and local supply systems to withstand increasing demand pressures from climate change and high‑consumption sectors such as AI and data infrastructure.

This highlights the need for stronger integration between digital policy and environmental regulation, including enhanced disclosure requirements, cross-sectoral impact assessments, and alignment with emerging sustainability obligations under EU digital and energy legislation.

Impacts of critical mineral extraction for AI infrastructure

Global AI computing capacity is rapidly scaling, with total capacity doubling roughly every seven months (Epoch AI, 2026). Since 2022, computing capacity from major AI chip designers has grown by approximately 3.3 times annually, facilitating larger models and expanded consumer use. Currently, NVIDIA provides over 60% of global AI compute, while Google and Amazon supply much of the remaining capacity (You, 2026).

This AI growth is dependent on a narrow set of critical raw materials that are essential for advanced semiconductors, high-performance computing and large-scale data centre infrastructure. While policy attention has largely focused on battery materials for the green transition, AI systems depend heavily on less visible but strategically critical minerals such as gallium, germanium, indium, palladium and tantalum. These are indispensable for graphics processing units, microelectronics and fibre-optic communications.

Europe’s high import dependence and the global concentration of mining and refining capacity for several of these materials expose AI infrastructure to significant supply-chain and geopolitical risks. This reinforces concerns addressed under the EU Critical Raw Materials Act.

At the same time, the extraction and processing of these minerals entails substantial environmental and social impacts, including high energy use, pollution, biodiversity loss and risks related to human rights. These issues have received comparatively limited scrutiny in the context of AI deployment.

As demand for AI-related infrastructure accelerates, competition for critical minerals is intensifying, potentially constraining the parallel expansion of renewable energy technologies essential for achieving EU climate targets. While AI may contribute to mitigation efforts — for example by improving resource efficiency, supporting mineral exploration and advancing recycling and substitution technologies — these benefits depend on governance frameworks that ensure AI deployment aligns with Europe’s sustainability objectives, environmental standards and strategic autonomy goals.

Impacts on well-being

The diffusion of AI algorithms across consumer markets and beyond has given rise to a growing range of potential harms with direct implications for individual and societal well-being. The impact of AI algorithms on purchasing behaviour is typically designed to maximise profits, raising regulatory concerns about its effects on societal welfare.

The increasing use of AI algorithms carries risks relating to the identification of specific information and rationality deficits that affect the demand of individual consumers (Bar-Gill and Sunstein, 2025). Key areas affecting purchasing decisions are:

• pricing — affecting affordability and consumer surplus;
• targeting — driving which products are offered to a particular consumer or group of consumers;
• algorithmically enhanced misperceptions — reinforcing and even creating misperceptions of present bias or unrealistic optimism;
• algorithmic coordination — higher prices due to collusion as AI algorithms set prices and are fed with data on competitors’ pricing strategies;
• discrimination based on race or gender — driven by profit maximisation, the creation by AI algorithms of a disparate impact on minority groups due to the use of historical information about online behaviour, information asymmetry and bounded rationality.

More specifically, biased training data in the foundation models might amplify existing discrimination.

In addition, the use widespread use of AI algorithms to moderate purchase behaviour entails potential psychological harm from enhancing online engagement maximisation (Pałka, 2024). Psychological implications of AI include higher levels of stress, anxiety, and depression, linked to effects of online behaviour on self-esteem, body image, and social comparison, as well as effects on social ties, potentially diminishing empathy and contributing to social isolation (Ettman and Galea, 2023). Ethical dilemmas such as privacy violations, algorithmic decision-making, and lack of transparency further strength the need for adequate regulation (Bengio et al., 2025).

Impacts of rebound effects

The positive impact of AI integration across purchase decision making faces the challenge of rebound effects. While AI improves energy efficiency per unit in areas such smart home heating and supply chain management, the resulting cost reductions and performance gains may increase demand for AI-driven services, creating a “service-level rebound” (Luccioni et al., 2025). Thus, the use of AI applications — such as personalised e-commerce recommendations and chatbots — may lead to economy-wide rebound effects.

Recent studies have estimated rebound effects from AI may range from 30% to 60%, potentially offsetting a considerable share of the expected energy and carbon savings (Oluka et al., 2026). However, studies at the country level point to the fact that stringent environmental policies, diversified energy supply, and context tailored digital infrastructure together with strong human oversight to avoid data manipulation may be capable of mitigating these adverse effects (Alnafrah, 2025).

Hence, AI's impact on sustainable consumption is conditional on a suitable policy mix capable of integrating environmental and digital governance as well as to ensure AI advances sustainability rather than exacerbates its risks.

Governance for sustainable AI

The EU is committed to strengthening its competitiveness in the global AI landscape while advancing sustainable consumption. Building on the AI Continent Action Plan, the AI Act and the forthcoming Cloud and AI Development Act, the EU plans to expand its data centre processing capacity over the next five to seven years and improve the conditions for further technological development.

Recent reporting requirements for data centres are enhancing transparency. Meanwhile, a strategic roadmap for digitalisation and AI in the energy sector, together with the data centre energy efficiency package, signal increasing policy attention related to resource efficiency.

The central challenge is to ensure that the deployment of AI algorithms and expansion of data centres, and the associated risks, are managed according to the principles of sustainable consumption and production. Developments must be aligned with the Directive on Empowering Consumers for the Green Transition, the Consumer Rights Directive and the Unfair Commercial Practices Directive, as well as the green public procurement initiative and the strategic public procurement framework. Coordination between AI policies and EU consumer regulations is key to supporting the changes needed in purchasing patterns to achieve climate and environmental objectives under the European Green Deal.