Enterprise AI Analysis
The dual footprint of artificial intelligence: environmental and social impacts across the globe
Author: Paola Tubaro, Centre of Research in Economics and Statistics (CREST), CNRS-ENSAE-Polytechnic Institute of Paris, Palaiseau, France.
Abstract: This article introduces the concept of the 'dual footprint' as a heuristic device to capture the commonalities and interdependencies between the different impacts of artificial intelligence (AI) on the natural and social surroundings that supply resources for its production and use. Two in-depth case studies, each illustrating international flows of raw materials and of data work services, portray the Al industry as a value chain that spans national boundaries and perpetuates inherited global inequalities. The countries that drive Al development generate a massive demand for inputs and trigger social costs that, through the value chain, largely fall on more peripheral actors. The arrangements in place distribute the costs and benefits of Al unequally, resulting in unsustainable practices and preventing the upward mobility of more disadvantaged countries. The dual footprint grasps how the environmental and social dimensions of the dual footprint emanate from similar underlying socio-economic processes and geographical trajectories.
Keywords: Artificial intelligence, material footprint, labour footprint, data work, value chains, offshoring.
Executive Impact: Key Takeaways for Your Enterprise
Understanding the global implications of AI's dual footprint is crucial for responsible and sustainable enterprise AI implementation. Here's what this research reveals:
0%
US State Data Center Electricity Consumption (Virginia)
Data centers consume up to 25% of electricity in key regions like Virginia, highlighting AI's massive energy demand.
0x
Projected Mineral Demand by 2040 (Strong Decarb. Path)
Global mineral demand for AI hardware could quadruple by 2040 on a strong decarbonization path, intensifying resource pressure.
0B
Projected Data Work Market Size by 2030
The market for AI-related human data services is projected to reach $17.1 Billion by 2030, indicating growing labor intensity.
Dual Footprint
Capturing AI's Interdependent Environmental & Social Impacts
The 'dual footprint' concept offers a holistic view of AI's impacts, revealing how its material and labor demands are deeply intertwined and often offshored to peripheral economies.
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Data Center Footprint
AI production in data centers consumes massive electricity and water. Training a single large language model can emit as much greenhouse gas as five cars over their lifetime. Even with energy efficiency improvements, the sheer scale of computation means the environmental burden remains significant.
Source: Strubell et al. 2019, Luccioni et al. 2023, Li et al. 2025
Hardware Manufacturing & Mining
Beyond data centers, the production of AI hardware (microprocessors, batteries, user equipment) demands substantial raw materials, particularly critical metals and minerals like lithium, cobalt, nickel, and rare earths. Mining these resources often leads to significant environmental disruption.
Source: Gupta et al. 2022, Berthelot et al. 2024, IEA 2021
AI's Material Value Chain Flow
Raw Material Mining
→
Component Manufacturing
→
Data Center Infrastructure
→
User Equipment Production
→
AI Model Training/Inference
→
E-waste & Recycling
Critical Mineral Demand
Projected 2-4x increase by 2040
The energy transition, while crucial for climate action, paradoxically intensifies the demand for mineral-intensive technologies, including those powering AI. This could lead to a 2 to 4-fold increase in global mineral demand by 2040, disproportionately impacting resource-rich, lower-income countries.
Data Work & Precarious Labor
AI requires extensive human labor for data enrichment (labeling, transcription, verification). These 'data workers' often operate through digital platforms, lacking formal employee status, job security, and social protection. Their work is typically low-paid and precarious, contributing to a 'labour footprint' of 'indecent labour'.
Source: Muldoon et al. 2024, ILO 2021, Tubaro & Casilli 2022
Offshoring & Global Disparities
Data work is increasingly offshored to countries with lower labor costs, leveraging existing global disparities. This reinforces informality and integrates precarious labor pools into the AI value chain, shifting social costs from AI-leading nations to more peripheral ones, where local communities often bear the burden.
Source: Altenried 2020, Chandhiramowuli et al. 2024, Tubaro et al. 2025a
| Aspect | Description/Indicators |
|---|---|
| Employment Status |
|
| Remuneration |
|
| Social Protection |
|
| Job Security |
|
| Working Conditions |
|
| Power Dynamics |
|
| Note: Based on literature defining 'indecent labour' and 'bad labour' footprints (García-Alaminos et al. 2020, Simas et al. 2014, ILO 2021). | |
AI Value Chains: Beyond Code
AI is not just code; it relies on complex global value chains for its production and deployment, encompassing everything from raw material extraction and hardware manufacturing to data labeling and infrastructure operations. These chains transcend national borders, mirroring those of traditional industries.
Source: Crawford 2021, Valdivia 2024, Attard-Frost & Widder 2025
Extractivism & Dependency Theory
The AI industry exhibits characteristics of 'extractivism,' appropriating natural resources and labor from peripheral regions. This perpetuates inherited global inequalities, where leading countries benefit from cheap inputs, while the environmental and social costs are externalized onto disadvantaged nations, hindering their development.
Source: Mezzadra & Neilson 2017, Chagnon et al. 2022, Valente & Grohmann 2024
Global Burden Transfer in AI Production
Core AI Demand (US, France)
→
Input Procurement (Minerals, Data Labor)
→
Peripheral Supplier Countries (Argentina, Madagascar)
→
Externalized Environmental/Social Costs
→
Obscured Responsibility
→
Perpetuated Global Inequalities
Offshore Obfuscation
Hiding the True Cost of AI
The geographical distance between AI development hubs and resource/labor supply regions creates an 'offshore obfuscation' effect. This makes it difficult for leading countries to acknowledge and take responsibility for the full environmental and social footprint generated overseas, perpetuating unsustainable practices.
Argentina - United States: Lithium & Data Work
Argentina, a key lithium exporter (vital for AI hardware) and a source of qualified data workers, supplies the US. While offering income in hard currency, lithium extraction leads to social conflicts and environmental depletion. Data work, often unprotected, burdens local economies as workers rely on other jobs or family support due to low pay and lack of social security.
Countries Involved: Argentina, United States
Resources/Services: Lithium, Data Work Services
Madagascar - Japan/S. Korea/France: Cobalt, Nickel & Data Work
Madagascar, rich in cobalt and nickel (for AI hardware) and a provider of data work services to France, faces similar challenges. Mining contributes little to local job creation and raises environmental concerns. Data workers, predominantly young and educated, face precarious conditions, low pay, and limited career progression, leading to reliance on family support and unfulfilled development aspirations.
Countries Involved: Madagascar, Japan, South Korea, France
Resources/Services: Cobalt, Nickel, Data Work Services
| Aspect | Argentina (vs. US) | Madagascar (vs. JP/KR/FR) |
|---|---|---|
| Resource Supplied |
|
|
| Economic Status |
|
|
| Mineral Impacts |
|
|
| Data Work Conditions |
|
|
| Key Challenge |
|
|
| Note: Both countries serve as peripheral suppliers in the global AI value chain, experiencing externalized costs. | ||
Calculate Your Potential AI Efficiency Gains
Estimate the annual hours and cost savings your enterprise could achieve by implementing AI solutions tailored to your specific operational context.
Estimated Annual Cost Savings
$0
Annual Hours Reclaimed
0
Your Path to Sustainable AI Implementation
A structured approach is key to integrating AI responsibly and effectively within your enterprise, mitigating risks identified in the dual footprint analysis.
Phase 01: Discovery & Strategic Alignment
Assess current operational needs, define clear AI objectives, and align with your enterprise's sustainability and ethical guidelines.
Phase 02: Data Foundation & Governance
Establish robust data collection, cleaning, and labeling processes, prioritizing ethical sourcing and ensuring data quality and privacy.
Phase 03: Model Development & Impact Assessment
Develop and train AI models while performing early assessments of potential environmental (energy, materials) and social (labor displacement, bias) impacts.
Phase 04: Integration & Responsible Deployment
Integrate AI solutions with existing systems, deploying with transparency and mechanisms for human oversight and intervention.
Phase 05: Monitoring, Audit & Adaptation
Continuously monitor AI system performance, conduct regular audits for ethical compliance, and adapt to emerging environmental and social considerations.
Phase 06: Scaling & Long-Term Sustainability
Scale AI applications strategically, ensuring that growth does not compromise environmental stewardship or perpetuate social inequalities in the supply chain.
Ready to Build a Sustainable AI Future for Your Enterprise?
Leverage our expertise to navigate the complexities of AI development and ensure your solutions contribute positively to both your bottom line and global well-being. Book a free consultation today.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat